CHANGELOG.md

Changelog

0.15.0

Breaking Changes

• Agent plugin channel retired: Orbit no longer ships Claude, Codex, or Cursor agent plugins; install and use the standalone Orbit CLI instead. ([ORB-10995])

Highlights

• Built-in code-review sweep: fresh workspaces include a disabled code-review auto-task that teams can enable to schedule recurring reviews. ([ORB-10997])
• Reliable concurrent task updates: task writes now coordinate safely across simultaneous CLI, MCP, and dashboard activity. ([ORB-10988])
• Simpler binary-first onboarding: the Quick Start now leads with installing the standalone CLI and brings your existing repository conventions into Orbit. ([ORB-10992])

0.14.0

Breaking Changes

• Fresh installs no longer seed the legacy qa crew: new orbit init workspaces use the supported system crew; update prompts that name the seeded qa crew. Existing user-authored qa configuration remains compatible. ([ORB-10955])
• proc.spawn activities now require an allowlist: v2 activities that grant proc.spawn must declare proc_allowed_programs; add the allowed programs explicitly. ([ORB-10959])

Highlights

• Cursor Agent Plugin support: Orbit now ships as a Cursor-compatible Agent Plugin alongside the Claude and Codex plugin formats. ([ORB-10943])
• More managed CLI executors: GitHub Copilot CLI and Cursor Agent CLI are first-class workflow executors, with improved installed-agent detection. ([ORB-10946])
• Operator-authorized MCP setup: orbit workspace init --mcp can generate MCP configuration for a workspace. ([ORB-10960])
• Dashboard workspace shell: the dashboard gains a command rail, two-mode dock, and log status bar for faster operator navigation. ([ORB-10972])
• Security-review automation: fresh workspaces include a disabled security-review auto-task that teams can enable when ready. ([ORB-10950])

0.13.0

Breaking Changes

• Store schema 15 → 16: audit actor identity splits role into actor_kind plus a normalized actor. Workspaces migrate on open; a newer store still refuses an older binary. ([ORB-10888])
• Store schema 16 → 17: unauthenticated MCP calls persist a self-reported actor. Same migrate-on-open / no-downgrade rule. ([ORB-10890])
• Task complexity is required: CLI task add, web POST /api/tasks, and orbit.task.add reject a missing complexity — pass low, medium, or hard. ([ORB-10892])

Highlights

• Dashboard delivery controls: inspect, toggle, and mint auto-tasks; operate routines and the sweep clock; the Tasks view is clearer with safer live-log and inline edits. ([ORB-10876])
• Complexity is first-class: every new task must set it, and the dashboard charts it as a dimension. ([ORB-10892])
• Governed MCP tools need an operator: unauthenticated sessions can no longer invoke those tools. ([ORB-10916])
• Claude Keychain OAuth under sandbox: Claude can read the user Keychain for OAuth; an explicit denyRead still wins. ([ORB-10929])
• System crew is seeded: jobs can name a portable system crew; qa stays for compatibility. ([ORB-10877])

0.12.1

Highlights

• Release cut: version files and the GitHub tag aligned at 0.12.1. ([ORB-10870])

0.12.0

Breaking Changes

• Seeded skill directories retired: orbit-task, orbit-search, orbit-workflow, and the plugin orbit-task-pilot skill are gone — use the orbit router skill and its references/. ([ORB-10860])

Highlights

• One Orbit skill: four seeded skills collapsed into a single orbit router with on-demand references. ([ORB-10860])
• Stale skill links cleaned on upgrade: workspace init removes dangling Claude/Codex skill symlinks left after skill retirement. ([ORB-10869])
• Doctor reports leftover skill directories: orphaned skill dirs without SKILL.md are a warning, not a clean bill of health. ([ORB-10862])
• Safer artifact redaction: env-value redaction no longer silently eats ordinary words in task prose. ([ORB-10867])

0.11.0

Breaking Changes

• Store schema 11 → 15: workspaces migrate on open (friction SQLite, workspace-claim scope, learning-projection removal, invocation audit context). A newer store still refuses an older binary. ([ORB-10680])
• Native project learnings removed: orbit learning, orbit.learning.*, and learning search are gone — use tasks, docs, feature ADRs, and frictions. ([ORB-10736])
• Native ADR store and tools removed: orbit adr / orbit.adr.* and .orbit/adrs/ are gone — author decisions in each feature's 4_decisions.md. ([ORB-10726])
• Duel, independent review, and role: retired: planning-duel and review CLI/MCP/jobs are gone; activity YAML role: no longer loads — use crew: or the run-resolved crew, and ship via task_pr_pipeline. ([ORB-10628])
• Jobs are asynchronous by default: orbit run job returns a durable run immediately; --backend http|cli|auto is gone — use agent_cli and pass --wait when you need to block. ([ORB-10801])
• MCP and fs.* builtins trimmed: auto-task MCP advertises only list+mint; fs.read, fs.delete, and remaining fs.* tools are gone — use CLI/orbit tool run for auto-task authoring and provider-native file tools. ([ORB-10798])

Highlights

• Asynchronous job execution: job submission returns a durable run immediately, with --wait and terminal diagnostics when you need to block. ([ORB-10801])
• Epic-owned continuous delivery: an epic run drains descendant work through one owned worktree and lands the epic branch only after the epic's own destination checks pass. ([ORB-10815])
• One workspace selector grammar: CLI and MCP accept the same registered name, logical ws_* id, or absolute checkout path. ([ORB-10758])
• Network MCP transports: orbit mcp listen and related tunnel/loopback paths serve isolated sessions with explicit workspace routing and claim-gated remote execution. ([ORB-10690])
• Fail-closed Linux sandbox: Bubblewrap plus policy-derived grants give Linux CLI agents and task-pilot a read-only, fail-closed execution path. ([ORB-10552])

0.10.0

• Telemetry no longer discards completed runs: the invocations insert-bound columns (cache_create_1h_tokens, provider_cost_usd) reach existing databases through a new invocation_telemetry_columns migration (schema v10) instead of only the v1 baseline, and a failed invocation-trace write is now logged and recorded as a telemetry.persist_failed event on the run rather than failing the job. ([ORB-10367])
• PR review status clearing is documented: orbit.task.update now describes its empty-string clear convention, with tool-host coverage for the persisted null result. ([ORB-10229])
• Task PR status updates persist in v2 storage: local tool and dashboard PATCH updates now retain pr_status alongside status and execution-summary changes. ([ORB-10223])
• Bundled skills are repository-agnostic: the embedded skill tree drops Orbit-source paths, private Constellation names, workspace-local artifact IDs, and fixed design-doc filenames; a portability regression test and byte-aligned plugin mirrors guard against reintroduction. ([ORB-10208])
• CLI agent envelopes reach workflows: provider-wrapped, prose-prefixed successful responses now validate and project their result object before workflow templating; malformed or failed envelopes fail closed. ([ORB-10216])
• Temporary init validation isolates skill discovery: test fixtures now sandbox HOME before seeding skills, preserving existing agent discovery links after temporary-root cleanup. ([ORB-10220])
• Verbose CLI agent runs stay healthy: stdout and stderr capture now retain bounded diagnostic prefixes and protocol-safe tails while continuing to drain providers, preserving final response envelopes and recording observed byte counts instead of killing successful runs. ([ORB-10221])
• Scheduled provider discovery fixed: systemd routine sweeps now use a portable user PATH so provider launchers installed in ~/.local/bin remain available. ([ORB-10214])
• Log rotation and pipeline spawning tolerate replaced paths: missing log archive directories are silent no-ops, while long-lived Orbit processes resolve Linux deleted-inode executable paths back to the installed binary before spawning workers. ([ORB-10213])
• Skill guidance reflects friction triage: shipped skills use positional learning-show IDs and document mutable friction statuses, direct triage commands, and task-driven resolution. ([ORB-10210])
• Store schema v4 compatibility restored: the append-only job_run_archive_stage migration remains supported after an agent-main history rewrite dropped its registry entry, so current CLIs reopen already-upgraded stores without a destructive downgrade. ([ORB-10209])
• Workspace routines are opt-in: init seeds disabled auto-task, triage, and workspace-local ship-sweep definitions, preserves authored routines on re-init, and delegates shipment synchronously to the normal backlog pipeline. ([ORB-10207])
• Workspace re-init honors explicit routing: re-running init safely merges explicit ship-mode/base-branch changes while preserving identity, creation time, and authored routines; workspace list/show report the effective ship mode. ([ORB-10215])
• Legacy task friction status removed: tasks no longer deserialize, list, update, admit, or render status: friction; standalone friction artifacts are the only supported surface. Task attribution and record-parameter construction now share implementations without changing admission or triage behavior. ([ORB-10202])
• Activity and job catalogs share layered loading: typed adapters now share recursive discovery, directory deduplication, first-wins layering, and duplicate detection while preserving their distinct precedence and validation policies. ([ORB-10201])
• Configuration admission is table-driven: fixed settings now declare parsing, defaults, validation metadata, and CLI projection once, with generated registry/snapshot lookup completeness checks. ([ORB-10199])
• Activity/job precedence docs match execution trust rules: the design now distinguishes workspace-preferred job listings from default-authoritative named job and activity execution. ([ORB-10203])
• Systemd routine workers survive sweep exit: the user sweep service now limits shutdown to its main oneshot process, allowing detached pipeline workers to claim and complete clock-dispatched runs. ([ORB-10153])
• Task creation surface is narrower and consistent: task creation now exposes only legal initial statuses, list flags share repeat/comma parsing, and redundant agent/comment/instructions inputs are removed while model and managed identity attribution remain intact. ([ORB-10155])
• CLI command meaning has one compiler-enforced registry: dispatch, runtime bootstrap, audit metadata, JSON error formatting, and hook suppression now derive from one exhaustive command-operation declaration, eliminating wildcard policy defaults without changing CLI behavior. ([ORB-10200])

Breaking Changes

• Task locks returned to task administration: orbit locks is now orbit task locks. This partially reverses ORB-10000: unlike the other removed task verbs, which were consolidated into existing verbs, locks were only relocated to a more prominent top-level surface. Moving them back completes the original surface-reduction intent without changing the MCP tool surface. ([ORB-10428])
• Learning vote and comment surfaces removed: orbit.learning.upvote and all orbit.learning.comment.* tools and CLI subcommands are gone — use priority + search rank for ranking, update/supersede for corrections, and evidence for provenance. ([ORB-10046])
• orbit task trimmed from 19 to 12 subcommands: seven state-transition and read-projection verbs removed. approve/reject/unarchive/delete → orbit task update --status …; locks → orbit locks; prune-context → orbit task lint --fix; templates dropped. MCP tool surface unchanged. ([ORB-10000])

Highlights

• Graph-tool guidance lives only in the MCP surface: removed graph-tool invocation instructions from activity prompts and deleted the orbit-graph skill; graph tools self-describe when present, and essential selector/usage knowledge now lives in the orbit.graph.* tool descriptions. ([ORB-10151])

• CI green again — MCP tools/list snapshot synced: ORB-10149's four orbit_auto_task_* tools reached the MCP surface without a snapshot refresh, failing mcp_serve_tools_list_matches_production_snapshot on every push; the round-trip snapshot now includes them (additive — no tools removed). ([ORB-10152])

• QA sweep is an auto-task: the checked-in six-hour definition routes hands-on validation to crews.qa, dedupes open runs, and files findings through Orbit; no-diff-expected exempts side-effect-only tasks from workflow diff gates. Legacy QA code, config, CLI, routines, and deployment units are removed. ([ORB-10148])

• Auto-task primitive: recurring chores are data, not code. orbit auto-task add/list/show/update/toggle (+ orbit.auto_task.* tools) define .orbit/auto_tasks/*.yaml templates with a cron/interval schedule and dedupe policy; one seeded scheduler routine mints tasks from the due ones with catch-up collapse and skip_if_open dedupe. Provider-neutral (ADR-0217). ([ORB-10149])

• Ship --mode pr fails loudly on an empty diff: an implement step that writes nothing no longer reports success — an empty commit and a 0 commits ahead branch now hard-fail (no empty branch pushed, no PR, no promote-to-review), the premature pre-commit push is removed, and the agent envelope gets repo_root plus a fail-closed workspace_path. ([ORB-10134])

• Codex plugin documentation parity: Website install/MCP guides now cover Codex plugin installation, upgrades, fresh-task discovery, and read-only MCP smoke checks alongside Claude; the npm proxy README clarifies how both supported plugins consume the binary without manual asset copying. ([ORB-10117])

• Routine scheduler health over HTTP: GET /api/routines on the dashboard exposes each routine's last fire (timestamp, ok/error outcome, duration) and next-due slot, so a stopped scheduler is visible remotely as a stale last_fire without box ssh. ([ORB-10138])

• ADR create over HTTP: POST /api/adrs records a Proposed ADR (mirroring orbit.adr.add), so remote orchestrators can author decisions without an on-box run; malformed payloads get a structured 400. Friction/learning create routes stay absent. ([ORB-10141])

• Learning + ADR update routes on the HTTP API: PATCH /learnings/:id (summary/scope/tags/body/evidence/priority) and PATCH /adrs/:id (status/tags and mutable metadata) delegate to the CLI tools, preserving supersede-don't-delete and invalid-transition rejection; no create routes added. ([ORB-10143])

• Default failed-run triage: a seeded task_triage_pipeline (+ hourly routine, orbit run triage) diagnoses tasks blocked by failed runs, re-backlogs environmental failures with a bounded retry budget, and leaves the rest blocked with a diagnosis attached. See ADR-0215/ADR-0216. ([ORB-10129])

• Orphaned pending job runs now reconcile to interrupted: workers claim their queued run's pid, the open-time orphan scan and orbit doctor cover pending runs, and orbit run cancel <run_id> terminalizes stuck runs. ([ORB-10070])

• Remote dashboard task actions now reach the selected workspace: approve/reject/archive used a raw fetch() that skipped the workspace-routing helper, so they silently no-op'd against a non-default remote workspace; the workspace selector also no longer renders its filesystem path beneath the dropdown. ([ORB-10124])

• CI runner ENOSPC fixes: the coverage and ci jobs reclaim ~25-30 GB of preinstalled runner bloat before building, set CARGO_INCREMENTAL=0 and CARGO_PROFILE_TEST_DEBUG=line-tables-only to shrink target/, and the coverage cache is now deps-only (no instrumented target/). ([ORB-10126])

• Constellation crew catalog uses model-level names: the checked-in Orbit workspace now offers Claude opus/sonnet/fable and Codex sol/terra/luna crews, defaulting to opus. ([ORB-10133])

• Codex init default avoids nested sandboxing: fresh Orbit configurations now seed Codex with danger-full-access, leaving Orbit and the host as the execution boundary. ([ORB-10131])

• Crews are one provider-model assignment: [crews.<name>] now uses flat model, provider, and backend fields shared by every activity role; legacy three-role crews load through their implementer assignment. See ADR-0213. ([ORB-10130])

• Model defaults de-hardcoded and centralized: a new orbit-common::model_defaults module is the single source of truth for production model defaults; default Claude CLI models are now the unversioned opus/sonnet aliases and default Codex is gpt-5.6-terra. Existing workspaces are unchanged until orbit init --refresh-defaults. See ADR-0211. ([ORB-10051])

• orbit web serve defaults to global (multi-workspace) mode: the all-workspaces dashboard dropdown is always present; --global is now a deprecated no-op. ([ORB-10029])

• orbit-cmd extracted from orbit-core: the CLI-facing command layer moved into a new internal orbit-cmd crate (depends on orbit-core, never the reverse). Import paths orbit_core::command::{doctor,migrate,diagnostics,…} → orbit_cmd::…; CLI/MCP/on-disk behavior unchanged. See ADR-0203. ([ORB-10016])

• Ops runbook — docs/OPERATIONS.md: a command-first day-2 operations guide (state inventory, WAL-safe backup/restore, stuck-job debugging, corrupted-DB recovery, log rotation, health checks, orbit migrate), linked from the README. ([ORB-10014])

• Crate-boundary error translators enforced: every typed error crossing into OrbitError now goes through its owning crate's *_error_to_orbit translator, enforced mechanically by a new scripts/check-error-translation.sh guardrail. ([ORB-10013])

• Versioned .orbit/ upgrade path — orbit migrate: the whole workspace layout is now versioned via a state/layout.version marker and an ordered migration registry; older layouts auto-migrate on open, newer ones refuse to open. orbit migrate [--dry-run] reports and inspects. ([ORB-10012])

• One Selector parser instead of two: the duplicated selector grammar is unified on the orbit-common implementation, with orbit-graph-extract's module now a re-export shim; orbit-common's surface now recognizes module:/command:. See ADR-0202. ([ORB-10011])

• CI guardrails: coverage visibility, CHANGELOG freshness, declared MSRV: an informational cargo-llvm-cov coverage job, a new check-changelog-freshness.sh guardrail, and a declared rust-version = "1.88" MSRV enforced by a pinned-toolchain CI job. ([ORB-10010])

• Policy deny rules can no longer be dodged by respelling a path: glob-path normalization rebuilds paths from their normal components so secret/./key.txt, secret//key.txt, and trailing-slash spellings all match rules for secret/key.txt; adds Linux enforcement integration tests and a CI sandbox/policy job. ([ORB-10009])

• MCP round-trip integration tests + a session-workspace wire fix: end-to-end MCP coverage over real transports with snapshotted tools/list payloads; fixes a wire bug where _meta.orbit.workspace never reached tool dispatch over a real transport. ([ORB-10007])

• Dashboard endpoint integration tests: axum-level coverage of the /api surface over on-disk temp workspaces — GET /api/audit, adversarial artifact-serving traversal/symlink cases, POST /workflows/ship in global mode, and workspace-selection error paths. Test-only. ([ORB-10008])

• Task migration between machines: new orbit task export/import pack and restore task bundles as a portable tar.zst with integrity validation and collision renumbering; orbit task reindex rebuilds the index from bundles, and --task-id-start seeds disjoint id ranges. ([ORB-00034])

• orbit web connect <ssh-host>: view a remote workspace's dashboard over an auto-managed SSH tunnel — picks a free port, starts remote serve, forwards, opens the browser, and tears down on Ctrl-C. ([ORB-00029])

• Global, multi-workspace dashboard: orbit web serve outside any workspace (or with --global) serves every registered workspace via a workspace-keyed runtime map, with a header selector and an aggregate "All workspaces" task view. ([ORB-00030])

• Dashboard shows each workspace's filesystem location: the global dashboard surfaces where each workspace lives on disk (home-abbreviated) in the header selector and task Details box, disambiguating same-named workspaces. ([ORB-00037])

• Aggregate dashboard view no longer 400s: an isAggregateView() predicate gates per-workspace fetches so the "All workspaces" view fetches only the aggregate task list and shows a placeholder for workspace-specific panels instead of erroring. ([ORB-00039])

• Aggregate 400-guard extended to the remaining dashboard tabs: the shared isAggregateView() guard now also covers the Audit, Knowledge, and scoreboard-window fetches in the "All workspaces" view. ([ORB-00040])

• Aggregate 400-guard completed: closes the last gaps — the Diagnostics tab and the stale learning/ADR/friction detail panels now show the aggregate-mode placeholder instead of firing per-workspace endpoints. ([ORB-00044])

• Versioned store-schema migrations: the store SQLite DB tracks numbered migrations in a schema_meta ledger, applies each atomically, adopts pre-ledger DBs via an idempotent baseline, and refuses to open a DB written by a newer orbit. ([ORB-10003])

• SQLite reads no longer queue behind writes: the shared store checks out per-call read-only connections from a small pool while writes keep the single WAL writer, so a read completes even while a write transaction is open. ([ORB-10004])

• Job-run checkpoint/resume for hard interruptions: the v2 executor persists a recovery checkpoint after each completed step; dead-owner runs finalize to a new interrupted state, and orbit job resume <run_id> replays only the remaining steps. ([ORB-10002])

• One shared SQLite pragma helper: WAL/busy_timeout/foreign_keys/synchronous setup is now centralized in orbit_common::utility::sqlite::apply_default_pragmas, fixing per-store drift. ([ORB-10004])

• Always-on operability: detailed healthz and orbit doctor: /healthz?detailed=true runs per-check probes, and orbit doctor runs workspace pass/warn/fail health checks. ([ORB-10005])

• Retry hygiene for external calls: step backoff now uses full jitter, the agent-CLI and embedding-companion transports classify retryable-vs-permanent failures, and validate_job rejects invalid retry: blocks before execution. ([ORB-10006])

• Internal refactor — execution-context decomposition + scoreboard consolidation: orbit-engine's context.rs is split into a context/ module by concern, and orbit-store's scoreboard shapes share one implementation. No behavior change. ([ORB-10015])

• Dashboard task-create workspace selection hardened: POST /api/tasks uses ?workspace=<id> for selection (unknown → 404, inactive → 400) and now rejects a stray workspace body key with a 400 — fixing the bridge task_create workspace no-op. ([ORB-00042])

0.9.2

Highlights

• orbit graph is back as a thin CLI wrapper: the v2 graph is now reachable from the single orbit binary via orbit graph {sync, search, show, refs, callees, impact, trace, overview, implementors, deps, version, db-path, clean}, not only the standalone orbit-graph-cli binary or the in-process MCP adapter. orbit-graph-cli is lib-ified (lib + bin) so both front ends share one command layer with no duplication; the agent-facing graph surface is unchanged (still MCP-only). This amends the ADR-0198 consequence that dropped the subcommand. ([ORB-00396], ADR-0199)
• Task store crash durability hardened: the task registry SQLite connection now runs with synchronous=FULL, and task-bundle creation fsyncs the bundle directory's parent before returning — so a power loss or OS crash mid-write can no longer leave the task store or a freshly-created bundle in a torn, unrecoverable state. ([ORB-00395], [ORB-00394])

0.9.1

This release completes the orbit-graph v2 migration — v1 (orbit-knowledge) is fully decommissioned and v2 is now the only graph surface — and adds a stable protocol for plugging in out-of-process executors.

Highlights

• orbit-graph v2 is now the sole graph surface: the orbit-knowledge (v1) crate and its orbit.graph.* builtin tools are removed, the orbit graph CLI command is dropped (the v2 graph ships as the standalone orbit-graph-cli binary plus an always-on in-process MCP adapter), and orbit init no longer builds a graph up front — v2 syncs on demand. The MCP graph surface is now search / show / refs / callees / impact / trace / overview / implementors / deps; pack is gone and callers is folded into refs. ([ORB-00391], [ORB-00389])
• External Executor Protocol v1: register a homegrown agent or CLI as an Orbit executor without forking core, through a documented stdin/stdout envelope contract and dynamic out-of-process registration. ([ORB-00384])
• Graph reads are faster and more accurate: queries are index-backed and decoupled from a full-worktree rescan, show returns UTF-8 text instead of a raw byte array, trace resolves far more command handlers, and refs/impact fall back to fuzzy-name matches when the precise floor is empty. ([ORB-00377], [ORB-00380], [ORB-00386], [ORB-00387], [ORB-00381])

0.9.0

This is a security-hardening release: the bulk of the changes close secret-leak, sandbox-escape, network-exposure, and workspace-containment issues found in a focused audit (ORB-00355–ORB-00374).

Breaking Changes

• Shell activity/job type removed: the v2 type: shell activity surface is deleted end to end — a workspace activity or job declaring type: shell now fails to deserialize at load instead of executing. The shell program allowlist was self-asserted within the same workspace-supplied YAML (the gate was a tautology) and run_shell spawned with no OS sandbox, cwd confinement, or policy consultation. The shell_reference asset and the pure-shell job fixtures/demos are removed; use the sandboxed agent-CLI execution path instead. ([ORB-00374], [ORB-00363])
• [execution.env] inherit config key removed: an auto-discovered workspace config.toml could set inherit = true and — because workspace config replaces global — flip agent subprocess spawning to full environment inheritance, leaking the orbit process's API keys and tokens into spawned agents. Env inheritance is now fixed to false and is no longer configurable; only the pass allowlist remains tunable. Stale inherit keys are silently ignored. ([ORB-00365])

Highlights

• Secret redaction hardened across audit and logs: redaction now covers non-UTF-8 audit blobs (previously bypassed entirely), error/bytes fields recorded via tracing, and Gemini's HTTP transport no longer places the API key in the URL query string. The pattern redactor also learned major Google / SCM / cloud key formats it previously missed. ([ORB-00358], [ORB-00367], [ORB-00371], [ORB-00359])
• Sandbox and execution confinement tightened: the macOS sandbox profile no longer grants unrestricted file-read* plus network*, an SBPL (?i) inline-flag bug that disabled sandbox-exec for all macOS CLI runs is fixed, the run_command automation action no longer shell-interpolates rendered templates, and unbounded child stdout/stderr capture (a memory-exhaustion DoS) is now bounded. ([ORB-00370], [ORB-00372], [ORB-00364], [ORB-00362])
• Dashboard network exposure closed: orbit web serve now refuses to bind any non-loopback host (only 127.0.0.0/8 and ::1) — the unauthenticated read/write API can no longer be exposed to the network via --host 0.0.0.0; bind loopback behind an authenticated tunnel/reverse proxy for remote access. Stored XSS from untrusted task/learning/ADR/friction content rendered in the dashboard is also fixed. ([ORB-00360], [ORB-00355])
• Secret-file policy matching fixed: denyRead/denyModify globs now match dotenv-variant secret files (.env.local, etc.), policy glob matching is no longer case-sensitive on case-insensitive filesystems, and audit blobs and JSONL logs are no longer written world-readable. ([ORB-00357], [ORB-00366], [ORB-00368])
• Workspace containment enforced: a malicious workspace can no longer override seeded .orbit/resources/{jobs,activities}/ assets, and graph operations stay confined to the worktree even when an MCP session omits the workspace. ([ORB-00356], [ORB-00361])

0.8.2

Fixes

• Intel-macOS semantic companion dropped from the release: orbit-search-companion cannot link on x86_64-apple-darwin — ort-sys (2.0.0-rc.12) ships no ONNX Runtime prebuilt for Intel macOS. The Intel-mac release leg now builds the orbit CLI only, and the release publishes companion binaries for macOS-arm64, Linux x86_64, and Linux arm64 only. The Intel-mac CLI tarball still ships; semantic search is unsupported on Intel Macs (it never successfully shipped). Together with the glibc fix in 0.8.1, this completes recovery of the v0.8.0 release, whose GitHub Release never published while @orbit-tools/cli@0.8.0 was already on npm. ([ORB-00351])

0.8.1

Fixes

• Linux release build of the semantic companion: orbit-search-companion failed to link on the release runners — its prebuilt ONNX Runtime (via ort / fastembed) references glibc 2.38+ symbols (__isoc23_*) that don't exist on ubuntu-22.04 (glibc 2.35), failing both Linux build legs and blocking the GitHub Release. The release workflow now builds the Linux companion on ubuntu-24.04 (glibc 2.39) while keeping the orbit CLI on ubuntu-22.04, so the CLI's glibc-2.35 compatibility floor is preserved. This fixes forward after the v0.8.0 GitHub Release failed to publish while @orbit-tools/cli@0.8.0 had already been published to npm. ([ORB-00350])

0.8.0

Breaking Changes

• Agent MCP surface slimmed; orbit.review-thread.* alias dropped: six tools moved off the agent-facing MCP surface to register_inactive — orbit.task.reject, orbit.friction.list, orbit.friction.show, orbit.friction.resolve, orbit.learning.comment.list, and orbit.learning.upvote. They stay reachable via the CLI (orbit tool run) and runtime.run_tool, but tools/list over MCP no longer advertises them and orbit mcp serve rejects agent calls to them — these are operator/triage actions, while agents file friction via orbit.friction.add and discover learnings via orbit.search. Separately, the duplicate orbit.review-thread.* alias (added in ORB-00273) is removed; the canonical orbit.task.review_thread.* form that the README, the orbit-review-task skill, and the audit/redaction policy reference is unchanged. Seeded activity allowlists and skill docs were updated to drop the stale references. EXPECTED_INACTIVE_TOOL_NAMES length canary moves 21 → 27. ([ORB-00346], [ORB-00348])

Highlights

• Dashboard and scoreboard redesign: the dashboard adopts the new visual mock and the scoreboard gets a phase-1 reskin plus a window-aware backend — a ?window= selector (1h / 24h / 7d / 30d / all) now scopes scoreboard metrics, on an additive snapshot schema bump (v5 → v6). ([ORB-00345], [ORB-00336], [ORB-00337])
• Crew and duel composition adapt to detected agents: [workflow].default_crew and [duel].candidates are now derived from the agent families detected at orbit init rather than hard-coded, so a workspace's defaults match the agents actually installed. ([ORB-00347])
• Semantic-companion install hardened: the release workflow now publishes the standalone orbit-search-companion-* binaries alongside the CLI tarballs, so orbit semantic install can resolve signed companion assets from the GitHub Release. ([ORB-00299])

0.7.1

Fixes

• Rotate release signing key and recover package installs: adds orbit-release-key-3 to the shell and npm installer trust sets, makes it the canonical plugin/npm/release-signing.pub, and documents the new current signing path. This patch fixes forward after the v0.7.0 GitHub Release publish failed before assets were created while @orbit-tools/cli@0.7.0 had already been published. ([ORB-00292])

0.7.0

Breaking Changes

• Agent MCP surface trimmed: six admin/destructive tools moved off the agent-facing MCP surface — orbit.task.delete, orbit.task.lint, orbit.semantic.uninstall, orbit.adr.list, orbit.learning.prune, orbit.learning.comment.delete. They remain registered (register_inactive) and reachable via the CLI (orbit task delete, orbit adr list, etc.) and runtime.run_tool; tools/list over MCP no longer advertises them and orbit mcp serve rejects calls to them. Agents that need ADR enumeration use orbit.search --kind adr. A new orbit adr list CLI subcommand routes through the tool to preserve filter parity (--include-remote, --status, --owner, etc.). EXPECTED_INACTIVE_TOOL_NAMES length canary in crates/orbit-cli/src/command/mcp/tests/mod.rs bumps 15 → 21. ([ORB-00289])
• Semantic companion lookup no longer searches $PATH: normal runtime lookup uses the managed ~/.orbit/embed/bin/ install only. Dev/CI overrides must set ORBIT_SEARCH_COMPANION=<absolute path> plus ORBIT_SEARCH_COMPANION_ALLOW_UNSAFE=1; managed downloads now require a release-signed checksum manifest before replacement or execution. ([ORB-00265])
• Envelope and embedding rebuild verbs renamed: orbit docs reindex / orbit.docs.reindex is now orbit docs index / orbit.docs.index for doc-corpus embeddings, and orbit learning reindex / orbit.learning.reindex is now orbit learning sync / orbit.learning.sync for YAML-to-SQLite envelope reconciliation. No compatibility aliases are retained. ([ORB-00206], [ORB-00244])
• orbit search mode split and per-kind status syntax: the query surface now has three visible CLI forms: orbit search <query>, orbit search similar <id>, and orbit search path <path>. The old orbit search --semantic <id> and orbit search --path <path> forms are removed. --status now requires kind:value tokens such as task:open,doc:active,adr:proposed; bare tokens like --status open are rejected. CLI --field / --model and MCP field / embedding_model are removed from orbit search; MCP model remains provenance-only. ADR-0179 supersedes ADR-0175. ([ORB-00205])
• Per-domain search subcommands removed; cross-kind filters on orbit search: orbit task search, orbit docs search, orbit learning search (and the matching orbit.task.search, orbit.docs.search, orbit.learning.search MCP tools) are removed. Replacement: orbit search --kind {task,doc,learning,adr,all} <query> for content similarity and kind-specific list commands for structural filters. orbit search --all, --tag, per-kind --status, and orbit search path <path> cover the previous include-superseded/path/tag use cases. orbit learning list --path now uses glob containment rather than exact string match. ([ORB-00202])
• orbit search flag rename: free-text vector ranking is now --hybrid / hybrid: true; task-neighbor lookup is now --semantic <id> / semantic: "<id>". The old --semantic boolean and --related <id> surfaces are hard-removed; JSON mode values are now hybrid and neighbor. Historical phase-1 audit payloads carrying semantic: true are orphaned by the no-shim rename. ([ORB-00204])
• Design-doc tooling retired: orbit design check, orbit design init/list/show, orbit.design.*, the orbit-design skill, the wrapper script, and make check-design-docs are gone. Use orbit-docs, orbit.docs.*, and the same-PR doc update rule; orbit workspace init --design still seeds docs/design/CONVENTIONS.md. ([ORB-00112], [ORB-00177], [ORB-00182])
• Search namespace split: orbit.semantic.search, orbit.semantic.related, and the orbit-semantic skill are removed in favor of orbit.search and orbit-search; orbit semantic reindex is now orbit semantic index. Historical semantic.search / semantic.related audit event names are orphaned by this hard break because there are no external audit-history consumers yet. ([ORB-00196])
• Low-use CLI commands retired: orbit scoreboard, orbit metrics, and orbit adr migrate are removed. Their supported replacements are the dashboard scoreboard and metrics views, dashboard HTTP metrics endpoints, and the already-migrated orbit.adr.* artifact tools. ([ORB-00183], [ORB-00190], [ORB-00197])
• Legacy executor models: compatibility removed: executor YAML must use model_pair_override:; the one-release models: alias and deprecation warning path are gone. ([ORB-00051])

Highlights

• Unified knowledge search: lexical and hybrid search now spans tasks, docs, learnings, and ADRs, with doc/learning/ADR embeddings, task-neighbor lookup, path applicability, and a single orbit search mental model. ([ORB-00196], [ORB-00202], [ORB-00203], [ORB-00205], [ORB-00206], [ORB-00216], [ORB-00217])
• Worktree-aware artifact foundations: Orbit now distinguishes shared and local .orbit/ roots, adds a global artifact ID allocator, and lays the groundwork for worktree-local ADR/Learning artifacts while keeping existing stores on the shared root. ([ORB-00199], [ORB-00200], [ORB-00201])
• Docs and learning context in the agent loop: PreToolUse learning reminders, orbit-docs, indexed design docs, and task.show --with-context make project knowledge easier for agents to retrieve before editing. ([ORB-00160], [ORB-00162], [ORB-00163], [ORB-00166])
• Dashboard and steering improvements: Audit / Diagnostics side panels, scoreboard grouping, a task resolver, metrics panels, review-thread message panels, and hook-driven async steering make the dashboard a fuller operations surface. ([ORB-00142], [ORB-00144], [ORB-00154], [ORB-00191], [ORB-00211], [ORB-00213], [ORB-00273], [ORB-00274])
• Release and runtime hardening: installers authenticate release archives, semantic-companion binaries require signed checksums, key rotation is documented in the installer trust set, and the companion freshness check closes a TOCTOU window. ([ORB-00265], [ORB-00266], [ORB-00270], [ORB-00271])
• Safer agent writes and execution: artifact-write redaction, ambient MCP session context, and restricted v2 proc.spawn execution reduce the odds of leaking secrets or running in the wrong workspace. ([ORB-00138], [ORB-00256], [ORB-00262])

0.6.0

Release scope

• Grok onboarded as the fourth first-class agent family: all_agent_families() 3 → 4, sandbox profile, MCP init, executor YAML, commit identity, and a parity sweep across scoreboards/duels/docs. ([ORB-00043], [ORB-00044], [ORB-00045], [ORB-00046], [ORB-00047], [ORB-00048], [ORB-00049], [ORB-00050], [ORB-00052])
• Agent identity collapsed to family: model strings are configuration-only; family (codex / claude / gemini / grok) is the durable identity across tasks, scoreboards, friction, audit, planning-duel artifacts, and git author lines. ([ORB-00080], [ORB-00081], [ORB-00088], [ORB-00089], [ORB-00090], [ORB-00091], [ORB-00106])
• Per-task crew abstraction: replaces role-keyed [agent.<role>] config with named [crews.*] registries selectable per task, and gives the duel-plan agent pool the same configurable surface. ([ORB-00058], [ORB-00072], [ORB-00076], [ORB-00078])
• First-class Knowledge tab in dashboard: Learnings/Frictions/ADRs subtabs with inline lifecycle controls, plus task-detail enrichment (tags, external_refs, relations, job_run_id, review_threads, locked-files panel, per-task crew selector). ([ORB-00060], [ORB-00061], [ORB-00062], [ORB-00063], [ORB-00067], [ORB-00068], [ORB-00069], [ORB-00073], [ORB-00076], [ORB-00082], [ORB-00083], [ORB-00084], [ORB-00097])

Breaking Changes

• Branching model flipped: main is now the release/production branch; agent-main is the dev integration branch where task PRs land. Each release tags on agent-main then promotes to main via merge commit; hotfixes branch from main and back-merge to agent-main. Install URLs in README.md and the website now point at main. Retired stub crates/orbit-core/assets/activities/examples/promote_agent_main.yaml removed. See RELEASING.md §10b and §Hotfix flow. ([ORB-00054])
• Crew registry replaces role-keyed agent config: [agent.planner], [agent.implementer], [agent.reviewer] blocks are removed in favor of named [crews.<name>] registries selected via [workflow].default_crew or per-task crew. Workspaces with stale schema are rejected at load. ([ORB-00058])
• ship-auto and ship-local aliases removed: orbit run ship is now the unified async-by-default command — empty task IDs trigger auto-backlog mode, explicit IDs queue-and-wait instead of fail-fast. ([ORB-00075])
• Agent identity schema collapsed to family: PlanningRoleAssignment drops model; planning-duel artifact paths and signatures rename to {slot}.md and *authored by: {family} / {slot}*; resolve_agent_model_pair* helpers and alias canonicalization removed; scoreboard by_model → by_family. Includes a read-side migration; downstream consumers indexing on by_model keys break. ([ORB-00080])
• Family-identity migration script required after upgrade: scripts/migrate_family_identity.py rewrites persisted task/friction/run/audit/scoreboard records and four SQLite stores to family-keyed shapes. Dry-run by default; back up before applying. ([ORB-00081])
• Audit task.locks.* and job run-pipeline-worker events populate task_id/job_run_id semantically: previously these overloaded target_id. Downstream analytics joining on target_id for these events need updating. ([ORB-00085])
• FrictionStatus::Default derive swap: enum default impl moved from a hand-rolled impl to #[derive(Default)] to satisfy a clippy-deny lint. Behaviorally a no-op; listed for completeness. ([ORB-00086])
• Task relation enum gains Produces/Resolves; frictions auto-close on Review → Done: relation enum extension is additive but the auto-friction-close on task completion is a semantics change. A resolved_by_task back-pointer is added to friction records. ([ORB-00093])
• Learning storage layout: flat → per-entity directories: .orbit/learnings/<L-id>.yaml moves to .orbit/learnings/<L-id>/learning.yaml. Legacy-layout load returns a typed error directing operators at orbit learning migrate-layout. ([ORB-00096])
• ADR allocation policy: new ADR headings must first allocate globally via orbit.adr.add; local 3-digit ADR headings under 4_decisions.md are grandfathered. docs/design/CONVENTIONS.md §4 updated. ([ORB-00098])
• Knowledge-graph workspace_root attribution fixed: graph refs are now keyed on the worktree's actual branch rather than falling back to the main repo's agent-main; the missing-ref case rebuilds rather than silently reading the default branch. Cached selectors keyed off the old behavior may need refresh. ([ORB-00099], [ORB-00105])
• orbit.task.update persists source_task_id: the property was previously silently dropped on writes; clients that depended on the drop now see persistence. Empty-string clears the field. ([ORB-00101])
• MCP schema emitter no longer degrades object_list params to "string": evidence and similar fields now emit array-shaped schemas. MCP clients that worked around the bug by string-encoding payloads must send arrays. ([ORB-00102])
• Ship batch commit message template: feat: parallel batch [ORB-id] is replaced by a deterministic template — <type>: <truncated title>… [ORB-id] [EXT-id]… with an optional full-title line, execution-summary paragraph, and Planned-By / Implemented-By trailers. Release-note builders and git log --grep 'parallel batch' workflows break. ([ORB-00107])

Features

• Grok onboarded as fourth agent family: sandbox state dir + SBPL allowances; CLI runner + executor YAML; orbit mcp init --client grok writes .grok/config.toml; duel / scoreboard / friction-stats render zero-grok rows; design folder, commit identity, and docs updated. ([ORB-00043], [ORB-00044], [ORB-00045], [ORB-00046], [ORB-00047], [ORB-00048], [ORB-00049], [ORB-00050], [ORB-00052])
• Knowledge tab in dashboard: Learnings subtab (list, supersede, stats), Frictions subtab (triage, resolve, stats, tag-picker sourced from YAML), ADRs subtab (accept, supersede, related-task deep-links), task-detail enrichment, Locked Files panel, per-task crew selector, markdown rendering. ([ORB-00060], [ORB-00061], [ORB-00062], [ORB-00063], [ORB-00067], [ORB-00068], [ORB-00069], [ORB-00073], [ORB-00076], [ORB-00083])
• Duel-plan agent pool configurable via [duel] candidates + [duel.models] in config.toml; runtime-host trait methods for candidate/model resolution; preserves fallback for non-duel callers. ([ORB-00072])
• Cross-artifact task relations: Produces / Resolves variants enable typed task ↔ friction / learning links; auto-resolves frictions on Review → Done with a resolved_by_task back-pointer. ([ORB-00093])
• Learning enrichments: per-learning comments.jsonl with push-injection rendering; decay-weighted upvotes with task-anchored idempotency; learning-creation wired into agent activity loops with a checkpoint. ([ORB-00077], [ORB-00094], [ORB-00095])
• Direct-agent runtime model injection: ExecutorDef.model_flag enables data-driven -m / --model flag dispatch per step. ([ORB-00053])
• Ship command unified to async-by-default: orbit run ship empty-task-IDs → auto mode, explicit IDs → gated; waiting-reason fields surfaced through orbit run history / show. ([ORB-00074], [ORB-00075])
• Backlog dependency gating: list_backlog_tasks filters by task_dependencies_ready, fixing out-of-order auto-pipeline execution. ([ORB-00057])
• Skill quality nudges: orbit-create-task now teaches optional complexity, dependencies, parent_id, and cross-artifact relations; orbit.friction.add description enumerates the tag taxonomy from YAML. ([ORB-00064], [ORB-00070], [ORB-00104])

Fixes

• Identity attribution end-to-end: runtime ToolContext wire-up, automation-driven Review / Done transitions, git author resolver, and the ship-batch Done loop closed the recurring implemented_by: "system" bug across PR-open, ship, and review paths. ([ORB-00067], [ORB-00088], [ORB-00089], [ORB-00090], [ORB-00091], [ORB-00106])
• Concurrent worktree setup: SHA-resolution + bounded retry eliminates .git/config lock races; post-failure cleanup is idempotent. ([ORB-00059])
• Policy dashboard denial identity: real JobRun IDs are separated from synthetic audit execution_id; task-lock denials now expose actor / requested-files / conflicts. ([ORB-00066])
• CI failure-recovery task type: corrected from invalid "issue" to "bug" (valid types: feature / bug / refactor / chore). ([ORB-00056])
• Clippy expect() violations in orbit-store::legacy_models_warns that blocked PRs after the branching flip. ([ORB-00055])
• Dashboard task-row ID column no longer leaves 50px of dead space before the title — column sizes to content. ([ORB-00097])
• YAML stack-overflow advisory resolved (pre-ORB-scheme task). ([T20260430-16])

Chores

• Family-identity migration script (scripts/migrate_family_identity.py) ships with backups, dry-run-by-default, SQLite normalization, and scoreboard regeneration. ([ORB-00081])
• Learning storage migration helper (orbit learning migrate-layout) ships alongside the flat → per-entity layout flip. ([ORB-00096])
• ADR corpus reconciled: backfilled agent-families, project-learnings, and design-docs orphan ADRs into the global allocation. ([ORB-00103])
• ORB-00080 coverage gap closed: end-to-end planning-duel regression test, crew-driven CLI invocation test, and projection field labeling. ([ORB-00087])
• Dashboard cleanup: API route inventory footer removed; "rejected" status chip removed. ([ORB-00082], [ORB-00084])
• Local CI cleanup: warning-deny clippy across MCP / tools / core / engine / CLI (large-enum boxing, items-after-test ordering, expect formatting, bind-vs-map); task-review scoring coverage extended to the Grok reviewer; design-doc Last updated: refresh across eight docs. ([ORB-00108])
• Unattributed commits: README refreshes ([commit b35724da], [commit 836b307c], [commit 64f9685d], [commit f9397605], [commit 008d172a]); artifact backfill (ADR-0164, L20260517-7, L20260517-8) ([commit 44889e63]); make cleanup-branches target ([commit 51d0777a]); .gitignore lock files ([commit 8f00912a]); track .orbit/learnings/ + .orbit/adrs/ in git ([commit 0106ff3a]); duel configuration in orbit settings ([commit bcefa2a4]); learning-search absolute-path handling fix ([commit 052376ff]); agent model-pair updates for Codex and Claude ([commit 0657b991]); rename executor model-pair override (PR #240) ([commit 2925a4b7]).

0.5.4

Features

• Project-learnings push-injection (L1/L2/L3): relevant learning summaries are now injected into agent context at three layers — engine pre-prompt before runtime spawn, MCP sidecar on path-bearing tools (orbit.graph.show, orbit.graph.refs, orbit.task.show), and a Claude Code PreToolUse hook on Edit | Write | Read. Summary-only payloads with per-session dedup, per-call caps, and an ORBIT_SESSION_ID envelope for cross-process dedup. ([ORB-00009])
• First-class design-docs surface (orbit.design.* + orbit design check CLI): four MCP tools (init, list, show, check) plus a Rust port of the design-doc decay checker, with orbit workspace init --design seeding docs/design/CONVENTIONS.md when absent. make check-design-docs and scripts/check_design_doc_decay.py now wrap the Rust path. ([ORB-00019])
• Default-seeded skills aligned across asset, registry, plugin, and router catalogs: orbit-learning and orbit-design onboarded; orbit-review-task and the semantic-search skill brought into the default seed and the plugin's skills/ symlinks; three drift-detection unit tests guard the four catalogs against recurrence. Default seed bumped 7 → 11. ([ORB-00020], [ORB-00022])
• orbit workspace init --inject-agent-rules: opt-in flag writes an idempotent Orbit-rules block into CLAUDE.md and AGENTS.md at the workspace root, delimited by <!-- orbit-managed:start/end --> markers. Block content sourced from an editable asset; malformed marker pairs refuse to write. ([ORB-00023])
• Inline task status transitions in the dashboard: per-task actions row gains a status selector wired through the existing PATCH /tasks/:id backend, ordered by STATUS_ORDER with done last and excluding rejected/archived/friction. Surfaces a "no longer shown in dashboard list" notice when transitioning to done. ([ORB-00025])
• orbit.learning.* exposed over MCP: the full eight-tool learning surface (add, list, search, show, update, supersede, prune, reindex) is now reachable from every MCP client, not just orbit tool run. Restores parity with the orbit-learning skill instructions. ([ORB-00039])

Fixes

• PID identity & stale-run probe stability: versioned ps-lstart-utc-v1: token replaces raw lstart output (recorded under TZ=UTC/LC_ALL=C) so live workers are no longer falsely marked failed across timezone changes; a new OwnerIdentity::ProbeUnavailable outcome distinguishes transient ps failures from genuinely dead PIDs, preventing single-probe terminalization of live workers. Adds arbiter-side orbit.duel.plan.winner regression coverage. Backward-compat read path (LegacyLiveUnverified) handles existing persisted tokens. ([ORB-00036], [ORB-00037])
• Job-step error messages reach the dashboard: V2AuditEventKind::{StepFinished, RunFinished} gain an optional error_message field; the failure reason is preserved at emit time and surfaced by the audit reader so the Steps and Events tabs no longer show a bare red dot. Backward-compatible — older audit files load unchanged. ([ORB-00026])
• Gemini direct-agent sandbox + planner/arbiter artifact persistence: narrow macOS sandbox allowances for Orbit child-runtime writes (global logs, DB sidecars, task workspace bundles, workspace semantic DB), tightened HOME-derived defaults, and improved planning-duel missing-artifact diagnostics so orbit.duel.plan.add/winner work under sandbox without home-directory re-allow. ([ORB-00027], [commit f3919a99], [commit e706d596])
• Gemini CLI token accounting: the shared response usage parser now reads stats.models.<model>.tokens.{input,prompt,cached,candidates} without double-counting role aggregates, so Gemini direct-agent invocations no longer persist as zero-token traces. ([ORB-00028])
• Graph diagnostics quality: orbit.graph.show now returns did_you_mean suggestions for unresolvable method selectors; exact-name orbit.graph.search ranks definition kinds (trait/struct/enum/type/function/module) above impl-method selectors; orbit.graph.pack carries typed unresolved reasons (not_found / outside_indexed_roots / stale_snapshot); orbit.graph.overview carries a typed downgrade_reason with threshold/actual. ([ORB-00029], [ORB-00030], [ORB-00031], [ORB-00032])
• Task bundle creation no longer leaks a lock sentinel or double-dots its name: TaskBundleStoreV2::create_bundle now locks on the bundle directory target and unlinks the .ORB-XXXXX.lock sentinel inside the locked closure; the old pre-dotted create_lock_path helper was removed. Concurrent-create serialization preserved. ([ORB-00033])
• PR signature attribution corrected: batch_pr_signature no longer falls back to created_by (which often names a planner or a human filer); when no task carries implemented_by, the signature falls back to the PR-opening agent's model identity, which is by construction the author of the commits. ([ORB-00034])
• Dashboard AGENT LOGS rendering: <pre> switched to white-space: pre with horizontal scroll so long single-line stdout (e.g. Codex JSON envelopes) and multi-word stderr stay legible instead of fragmenting one token per row. ([ORB-00035])
• Dashboard friction-chip cleanup: removed dead friction task-status references from STATUS_ORDER and the approve/reject eligibility sets. The diagnostics-tab friction surface (/api/diagnostics/friction) is untouched. ([ORB-00024])

Chores

• make ci-fast introduced for pre-handoff checks: fmt-check + guardrail scripts, no compile. make ci stays the canonical merge gate via PR CI. Agent guidance updated to clarify when make ci failures classify as unrelated CI blockers vs task regressions. ([commit 4c22fa19], [commit 89ebc578])
• Release docs cross-linked: RELEASING.md and docs/RELEASE.md now reference each other so a first-time releaser landing on either file finds both. ([ORB-00040])
• Deprecated activity/executor YAML assets removed: cleanup of seeded resources that are no longer referenced by the runtime. ([commit c9cf36cc])
• devalue bumped 5.7.1 → 5.8.1 in website/. ([commit f08446e1])

0.5.3

Features

• Claude Code plugin SessionStart hook: the Orbit plugin now ships a SessionStart hook (plugin/hooks/check-workspace.sh) that detects an uninitialized workspace via a pure filesystem walk and surfaces a systemMessage instructing the user to run orbit init / orbit workspace init. Closes the silent-no-op gap where orbit mcp serve would attach with zero tools and no in-session signal. ([ORB-00018])

Fixes

• No-diff PR handoffs: pr_open now treats branches with zero commits ahead of the configured base as successful repository-noop handoffs, moves completed tasks to review, returns pr_created: false, and avoids creating an empty GitHub PR. ([ORB-00016])

Chores

• Workspace lint guardrails: codified Rust practice lints for panic surfaces, stdout/stderr usage, and async lock guards in [workspace.lints], with scoped allowlists and CLAUDE.md updated to separate enforced rules from conventions. ([ORB-00013])
• README workspace and MCP surface docs: documented the .orbit/ workspace layout, committed project-memory directories, and the agent-facing MCP tool namespaces. ([commit b99d3796])

0.5.1

Fixes

• Deprecation warning routes through tracing: the knowledge.task_id_pattern deprecation notice was emitted via writeln! directly to stderr from orbit-core, contradicting the CLAUDE.md "tracing for diagnostics; eprintln! only in orbit-cli" rule and leaking into cargo test output. Now goes through tracing::warn! with a structured config field.

Chores

• Release metadata recovery: backfilled the CHANGELOG entry for v0.5.0 (tagged without one) and aligned Cargo.toml workspace version + Cargo.lock to v0.5.1. v0.5.0 binaries shipped reporting orbit --version 0.4.0 because the workspace Cargo version was not bumped at tag time; v0.5.1 binaries report 0.5.1. Tag-message and commit-message formats now follow RELEASING.md.

0.5.0

Release scope

• Task artifact v2 cutover: the v1 single-file task store is replaced end-to-end by a bundle-based v2 store with status-neutral directories, append-heavy sidecar layout, a SQLite registry for fast lookups, a runtime backend, generated indexes, atomic delete, reservation scoping by workspace binding, search refinements (binary-artifact gating, UTF-8 validation), and a forward-only YAML migration framework keyed on schema_version. ADRs 001/002/003/004/007 for the v2 design flipped to Accepted; the v1 store, the legacy migration helpers, and the [task] artifact_store = "legacy" config gate are all gone.

Breaking Changes

• Legacy task artifact store removed: the v1 single-file task store and its migration helpers were deleted in favor of the v2 bundle store. [task] artifact_store accepts "v2" as a no-op for forward compatibility and rejects "legacy" with an explicit migration error. Workspaces still on legacy storage must migrate under v0.4.0 before upgrading. ([commit e9582eba], [commit 222f6020], [commit 123f89f7])
• Task.workspace_path / Task.repo_root dropped from the update path: TaskAutomationUpdate, TaskRecordUpdateParams, and TaskDocumentUpdateParams no longer accept these fields, and the v2 document layer rejects them at load. Worktree setup and parallel-batch dispatch no longer thread them through. The fields remain on the public DTO via projection from workspace metadata, but write-path inputs that include them now fail. ([commit 6beb14a2])

Features

• ADR-artifact subsystem (orbit.adr.*): ADRs lift out of per-feature 4_decisions.md markdown into first-class artifacts at .orbit/adrs/<status>/<id>/{adr.yaml,body.md} with globally-unique ADR-NNNN IDs and a three-state lifecycle (proposed/accepted/superseded). Ships domain types, a SQLite envelope index, a file store with per-ADR fs2 locks, five tools (add, show, list, update, supersede) with lifecycle audit rows, an orbit adr migrate one-shot, and a parser/sweeper hardened against rollup-bullet form and code-fenced examples. Migration imported 142 ADRs (97 accepted / 39 proposed / 6 superseded) across the existing corpus. The envelope's legacy_id: String field was renamed to legacy_ids: Vec<String> mid-development per the ADR-002 amendment; pre-GA churn, no external consumers yet. ([T20260510-27], [T20260510-28], [T20260511-1], [T20260511-2], [T20260511-3], [T20260511-11])
• Project learnings subsystem (orbit.learning.*): workspace-scoped LearningFileStore under .orbit/learnings/<id>/learning.yaml with a SQLite learnings_index for fast scope-glob lookups on the injection hot path. Eight MCP tools (add, list, search, show, update, supersede, reindex, prune) plus matching orbit learning <verb> CLI; orbit learning migrate-layout upgrades legacy flat workspaces. Learnings travel with the repo via a carved-out .gitignore entry. End-to-end dispatch latency p95 ≈ 0.04 ms at 500 records. ([T20260511-5], [T20260511-6], [ORB-00096])
• Plugin install contract locked down: make release-check enforces version lockstep across plugin/npm/package.json, plugin/.claude-plugin/plugin.json, npm view @orbit-tools/cli, and gh release list -L 1; scripts/smoke-plugin-install.sh exercises the published @orbit-tools/cli@latest postinstall + MCP tools/list handshake; .github/workflows/smoke-plugin-install.yml runs the smoke weekly and on every v* tag on macOS and Linux; docs/RELEASE.md codifies the chain. v0.5.0 switched the npm publish step to manual after the workflow ran into account-level 2FA. ([ORB-00012], [ORB-00014])
• Typed error discriminators: KnowledgeError.kind is now the typed KnowledgeErrorKind enum, and ten per-kind OrbitError::*NotFound(String) variants consolidate into NotFound { kind: NotFoundKind, id: String } with exhaustive matches in every error-code translator. JSON wire shape is preserved via #[serde]; internal Rust matchers move to the typed kind. ([ORB-00001])
• Property + snapshot tests on protocol boundaries: proptest-backed Selector display/parse roundtrip (256 cases per variant), multi-threaded GraphLockGuard/LockStore concurrency with a 5-second deadline, and committed JSON-snapshot coverage for AuditGuard success/failure/denied events. ([ORB-00002])
• Knowledge graph workflow relocation: build/show/search/history workflows moved from orbit-tools into a new orbit_knowledge::workflows module, so orbit-knowledge owns both the tool surface (commands/) and the host application surface (workflows/). ([commit 5fc2b72c])
• Forward-only YAML migration framework: orbit_common::migration::Plan chains Value → Value steps keyed on schema_version; OrbitError::Migration separates chain failures from store/parse errors; the v2 task-bundle envelope read path now flows through task_migrations::envelope_plan() (empty chain today — next schema bump adds one add_step call). ([commit 01928e76])
• Semantic companion install hardening: companion installation is version-aware, --force reinstalls supported, and background-companion stderr is suppressed during task-mutation indexing. ([T20260510-26])

Fixes

• pr_open template tolerates legacy batch_id overrides: worktree_setup emits batch_id as a deprecated alias equal to job_run_id, and required_job_run_id accepts job_run_id → run_id → legacy batch_id in that precedence. Recovers task_pr_pipeline runs whose stale resource overrides still reference {{ steps.worktree.output.batch_id }}. ([ORB-00010])
• Retired v1 / stub paths fail loudly instead of silent success: OrbitRuntime::get_job and RuntimeHost v1 job lookup return an explicit retired-v1 error instead of Ok(None); the promote_agent_main and revert_on_red deterministic actions fail with a "retired stub" error instead of returning skipped JSON. Tightens validation on inputs that were already invalid by spec. ([ORB-00007])
• Architecture and design-doc drift fixed: ARCHITECTURE.md was regenerated from cargo metadata to reflect the live workspace graph; make check-design-docs was cleared by refreshing stale Last updated: metadata and replacing moved/deleted file references across ten design folders. ([ORB-00006])
• Smoke plugin-install assertion fixed: assertion was checking "orbit\. but MCP wire names use underscores (sanitize_tool_name in orbit-mcp/src/adapter.rs replaces . with _). Without this fix the on-tag smoke would have false-failed. ([ORB-00014])

Chores

• Per-crate stability tier markers: [package.metadata.orbit] stability = ... added to all fourteen workspace crates (stable: orbit-common, orbit-store; experimental: orbit-embed-companion, orbit-registry; internal: the remaining ten). scripts/check-stability.sh enumerates members via cargo metadata, validates the marker, and fails closed with named offenders; wired into make stability and make ci. ([ORB-00005])
• Missing-docs guardrail: workspace-wide missing_docs = "warn" in [workspace.lints.rust]; RUSTDOCFLAGS=-D warnings cargo doc --no-deps --workspace wired into scripts/ci-guardrails.sh; 278 accidentally-pub items narrowed to pub(crate) (6.8% of the 4,116-item baseline). Legacy allow-fences in place for the remainder. ([ORB-00004])
• Community health files: CODE_OF_CONDUCT.md (Contributor Covenant v2.1, contact via GitHub Security Advisories), .github/PULL_REQUEST_TEMPLATE.md with linked Orbit task ID + make ci / make check-design-docs checkboxes, and three .github/ISSUE_TEMPLATE/*.yml issue forms with a 14-crate dropdown sourced from live crates/ listing. ([ORB-00011])
• Release runbook (RELEASING.md): pre-1.0 versioning policy with explicit breaking-vs-non-breaking criteria, 11-step release checklist, CHANGELOG conventions, and tag-push CI workflow description. ([T20260510-24])
• Semantic search surfaced in agent instructions: new semantic-search skill modeled on orbit-graph, plus pointers from orbit-create-task, orbit-execute-task, orbit-review-task, agent_implement.yaml, agent_review.yaml, epic_orchestrator.yaml, and dispatch_agent.yaml. All references use "if available" / "optional" language so missing companion never hard-fails a workflow. ([T20260511-4])
• Design-pattern reference docs added: docs/design-patterns/{command,strategy,raii_guard,newtype,error_translation}.md so feature work can copy from documented references instead of inventing new shapes. ([commit 3adcd838], [commit c713cc30], [commit 66389575], [commit f2f82bf1], [commit aa407aa0])
• Design-doc decay check: new scripts/check_design_doc_decay.py and make check-design-docs flag docs/design/* docs whose Last updated: precedes the last commit on any referenced crates/.../*.rs file. ([commit 18c48744])
• Workspace lint table introduced: [workspace.lints] in root Cargo.toml with each crate inheriting via lints.workspace = true; mechanical lint rules moved out of CLAUDE.md prose. ([commit 0cbb037d])
• CLAUDE.md refactored to point at ARCHITECTURE.md: crate layering moved to a dedicated architecture doc; CLAUDE.md shrinks to project rules and judgment calls. ([commit b7c590aa], [commit 9362730b], [commit 7582af9d])
• MCP server configuration for additional environments: .codex/config.toml, .gemini/settings.json, and .vscode/mcp.json added so Codex/Gemini/VS Code agents pick up Orbit MCP out of the box. ([commit 200ee6fd])
• README simplification: Quick Start trimmed; positioning sentences moved to design docs. ([commit 75699824])
• Agent skills symlink: .agents/skills symlink added so external agent harnesses pick up the seeded skill set. ([commit 9f7bf89b])
• Lessons log update: docs/LESSONS.md extended with a workflow lesson. ([commit 024013af])

0.4.0

Release scope

• Pivot to "auditable agentic task management": README and landing-page positioning realigned around intent attribution and audit trails, with throughput and parallel-execution sections refreshed to match.
• Knowledge graph reads on SQLite: per-build graph_index.sqlite sidecar with read-only fast paths for graph.overview, graph.search, and graph.show, plus an output-equivalence harness against the JSON fallback.
• Semantic search foundation (preview): hybrid (BM25 + cosine + RRF) retrieval over tasks, delivered as a separately-installed orbit-embed-companion binary. Preview status — surface may change before v1.

Breaking Changes

• Friction reports relocated: friction is no longer a task type or status. Records live as append-only markdown under .orbit/frictions/{yyyy}-{mm}/F{nnn}.md and are managed through orbit.friction.add/list/show/stats. orbit.task.add rejects type: friction / status: friction; web API and scoreboard JSON drop friction_bounty. ([T20260510-13])
• Task type taxonomy reduced: the task | feature | epic | issue | bug | chore | refactor | friction enum collapses to feature | bug | refactor | chore. orbit.task.add and orbit.task.update reject the removed values; existing tasks were migrated. ([T20260510-14])
• Attribution narrowed to model: the agent field is removed from Actor; orbit.task.add rejects an agent parameter and Orbit infers the agent family from model via agent_from_model. MCP orbit_task_list, orbit_task_search, and orbit_task_review_thread_list responses are now object-shaped (previously top-level arrays) so Cursor and VS Code accept them. ([T20260510-15])
• Knowledge-graph leaf IDs unified across extractors: Python, Rust, Java, and TypeScript leaf selectors now use a single canonical form so SQL and JSON paths return set-equivalent results. GRAPH_SQLITE_INDEX_SCHEMA_VERSION bumps; consumers caching selectors must rebuild. ([T20260510-7])
• Semantic search requires a companion binary: orbit-embed-companion is installed separately via orbit semantic install; orbit semantic * and the matching MCP tools fail until it is present. ([T20260510-9], [T20260510-10])
• JobV2Step rejects multi-body shapes: YAML steps that previously parsed silently with both target and parallel (or any other body combination) now fail at load. ([T20260509-31])
• orbit-locks skill removed: the seeded orbit-locks/SKILL.md and the ad-hoc orbit.task.locks* instructions in the seeded orbit skill are gone — the gate pipeline still owns reservations. External agent prompts referencing the skill must be updated. ([T20260510-17])

Features

• Knowledge graph SQLite read facade: per-build graph_index.sqlite with versioned schema, read-only facade with graceful JSON fallback, and SQL fast paths for graph.overview (aggregation), graph.search (exact-name, path-prefix, and substring), and graph.show (selector lookup with children repopulated via a forward-pointer edge table). ([T20260509-70], [T20260509-71], [T20260509-72], [T20260509-73], [T20260509-74])
• Knowledge graph latency wins: lazy source hydration via GraphReadOptions, a bounded default-ranking work cap on search, and a BinaryHeap top-K in overview.top_files. ([T20260509-65], [T20260509-67], [T20260509-68])
• Knowledge command surface: graph business logic — ranking, classification, fast-path orchestration — relocated into orbit_knowledge::commands::* so non-tool consumers share canonical behavior. ([T20260510-5])
• Semantic search subsystem: orbit-embed client, orbit-embed-companion binary, embeddings and tasks_fts SQLite schema, paragraph chunker, BLAKE3 dedup, task-mutation index hooks, and orbit semantic install/uninstall/reindex/stats/search/related CLI plus MCP surface. ([T20260510-3], [T20260510-9], [T20260510-10], [T20260510-20])
• Task tags: first-class tags: Vec<String> field with normalized SQLite index and --tag filtering on orbit task list/search. ([T20260510-12])
• Activity/job runtime polish: wildcard-aware tool allowlists honored at dispatch and HTTP-loop schema advertisement, asset-load-time allowlist validation, agent-loop on_denial: continue, literal-boolean condition atoms, and exclusive locking on duel scoreboard appends. ([T20260509-15], [T20260509-22], [T20260509-23], [T20260509-25], [T20260509-32])
• Recovery role configurability: seeded step-failure recovery activity uses role: reviewer and resolves agent/model from [agent.reviewer] config instead of hardcoded Codex. ([T20260509-14])
• Done-task sync cap: website task sync caps generated pages to the 100 most recent done tasks. ([T20260509-20])
• Debug-job-failure skill: seeded orbit-debug-job-failure SKILL.md teaching agents how to investigate failed/stuck/cancelled job runs across run state, audit events, blobs, and live processes. ([T20260509-79])
• Graph-latency benchmark: split benchmarks/CONVENTIONS.md into agent vs perf RESULTS schemas, scaffolded benchmarks/graph-latency/ with three-tier Python/Java/Rust corpora, and ran v1/v2 sweeps against the post-SQLite read paths. ([T20260509-63], [T20260509-87], [T20260510-4])

Fixes

• Output-equivalence between SQL and fallback paths: graph.search SQL widened to substring match aligned with the navigator and graph.show repopulates children via a forward-pointer edge table. ([T20260510-1], [T20260510-2])
• Workflow stop on implementer envelope failure: peek_response_status extracts embedded Orbit envelopes from CLI stdout that contains explanatory prose before the JSON, so failed implementations no longer advance to push/PR. ([T20260509-15])
• ship-auto empty backlog: condition evaluator skip guards no longer fail when bundle_count is zero. ([T20260509-11])
• Parallel dispatcher hang on worker timeout: scoped-thread workers now exit through a cancellable boundary so the pipeline returns within its own timeout. ([T20260509-38])
• Subprocess timeout cleanup: bare spawn_with_timeout starts children in a process group/session so grandchildren are killed and pipes don't leak. ([T20260509-40])
• Stdout no longer duplicated into DispatchOutcome: blob refs are the source of truth, with a bounded preview retained. ([T20260509-43])
• Path-traversal hardening: task store ID validation, policy candidate-path component checks, resource-name validation in policy/executor stores, and an absolute-path probe for sandbox-exec. ([T20260509-26], [T20260509-27], [T20260509-28], [T20260509-30])
• Tool deletion guard: orbit.task.delete MCP tool respects the same protected-status guard as orbit task delete. ([T20260509-44])
• Backend resolution: invalid [runtime] backend values reject before dispatch instead of falling through to preview HTTP. ([T20260509-45])
• Architecture guardrail: scripts/check-dependency-direction.sh derives the workspace-crate list from cargo metadata so new crates can't drift past the check. ([T20260509-46])
• JSON output purity: orbit task approve --all-proposed --json and reject equivalents emit pure JSON on stdout. ([T20260509-47])
• Dashboard dependencies: dependency-status index includes done and archived tasks so visible rows don't misreport completed deps as missing. ([T20260509-48])
• Reject help/help-truth alignment: top-level task help describes the actual reject transition matrix. ([T20260509-50])
• Symlink scanning: knowledge scanner skips and canonicalizes symlinked dirs to prevent index escape and cycles. ([T20260509-33])
• Graph freshness: manifest persists exact Git identity rather than relying on committer timestamp. ([T20260509-34])
• GitHub PR result validation: github.pr.review and github.pr.comment.reply validate JSON shape before reporting success with id 0. ([T20260509-36])
• MCP name collisions: dot-to-underscore name mapping detects ambiguity on startup. ([T20260509-37])
• Git author identity: workflow commits set per-implementer author dynamically without writing repo-local git config user.*. ([T20260508-22], [T20260509-12])
• CI clippy guardrails: cleared manual_contains, needless_borrow, useless_conversion, match_like_matches_macro, empty_line_after_doc_comments, too_many_arguments, doc_lazy_continuation, and question_mark violations under -D warnings. ([T20260509-18], [T20260509-61], [T20260510-15], [T20260510-22])
• fast-uri Dependabot alerts: addressed and documented dev-only fast-uri advisories on website/package-lock.json. ([T20260509-57])

Chores

• Module decomposition: split command/web/api.rs (3,376 LOC), activity_job/job_executor.rs (2,841 LOC), activity_job/cli_runner.rs (2,161 LOC), runtime/orbit_tool_host/mod.rs (2,033 LOC), command/mcp/setup.rs (1,964 LOC), and the activity_job/groundhog.rs runner. ([T20260509-1], [T20260509-2], [T20260509-3], [T20260509-4], [T20260509-5], [T20260509-19])
• Embed crate ownership: relocated vector::* and the semantic command surface from orbit-store and orbit-core into orbit-embed, reversing the dep arrow so orbit-store no longer knows the embedding feature exists. ([T20260510-20])
• Panic audit: classified ~1,864 unwrap / expect / panic! sites and removed the accidental ones in execution-critical paths. ([T20260509-6])
• Test coverage on highest-risk seams: focused tests for the activity/job DAG executor and the macOS sandbox/policy boundary. ([T20260509-7])
• Plan-duel context_files extraction: duel resolver auto-populates task.context_files from the winning plan's Context Files section. ([T20260509-9])
• Knowledge-graph, policy-sandbox, and Groundhog doc hygiene: refreshed owned design docs to current surface.
• Task lineage design (first draft): seeded docs/design/task-lineage/ with edge schema, three derivers, bipartite bridge, feature closure, and symbol-biography renderer. ([T20260510-21])
• Project learnings design (seed): seeded docs/design/project-learnings/ with hook-injection layer rationale; deferred until semantic search is Accepted. ([T20260510-11])
• Semantic search v2 design pivot: switched to companion-binary architecture per ADR-005. ([T20260510-3])
• Orbit-create-task skill: tightened context_files rule (existing modified or deleted files only, prefer file-level selectors). ([T20260509-83])
• make ci alignment: make build failures resolved alongside the dep-direction script. ([T20260510-22])
• Release metadata: bumped Cargo workspace, plugin manifests, and npm proxy metadata to v0.4.0.

0.3.1

Features

• Pipeline dispatch reliability: hardened parallel, gate, and epic pipelines with failed-child completion handling, longer task lock coverage, epic timeout/convergence fixes, resolved workspace subprocess cwd, and per-step agent log/error surfacing. ([T20260427-34], [T20260427-36], [T20260427-38], [T20260427-40], [T20260508-8], [T20260508-14])
• Metrics and public docs: split the public metrics surface into Operations and Scoreboard views, added done-task sync pages for orbit-cli.com, refreshed positioning/reference docs, and refined the website UI. ([T20260508-4], [T20260508-16], [T20260508-19], [T20260508-20], [T20260507-21])
• Registry and benchmark tooling: added the orbit-registry crate and identity-key benchmark harness for exercising knowledge graph selector stability. ([T20260507-12], [T20260508-2])

Fixes

• macOS sandbox and CLI execution: allowed Claude's $HOME/.claude.json lock/tmp siblings, re-allowed the active job-run worktree after global deny rules, and demoted successful CLI exits when the inner Orbit envelope reported failure. ([T20260508-13], [T20260508-17])
• Workflow defaults and links: made workflow base branches resolve from [workflow] base_branch when CLI flags are omitted, and fixed task-ID links in generated PR bodies with an opt-in URL template. ([T20260508-11], [T20260508-12])
• CI clippy guardrails: grouped macOS sandbox spawn inputs into a request struct so strict workspace clippy passes under -D warnings. ([T20260508-21])

Chores

• Release metadata: bumped Cargo workspace crates, plugin manifests, install examples, and npm proxy metadata to v0.3.1. ([T20260508-21])
• Release packaging: kept GitHub Release tarballs, checksums, Homebrew tap updates, and installer smoke tests as the supported release path, while removing the npm publish step from the tag workflow.

0.3.0

Release scope

• Stable surface: CLI agent backends. v1 supports backend: cli as the stable agent invocation path, running Codex, Claude Code, Gemini CLI, and other official CLIs as supervised subprocesses. backend: http (LoopTransport) and the Groundhog checkpoint runner remain preview-only for v1; they are exercised in tests but can change before v2.

Breaking Changes

• Activity/job schema v1 removed: loaders now reject schemaVersion: 1 activity/job assets, the v1 reconcile/runtime/store paths are gone, and schemaVersion: 2 is the canonical activity/job surface. ([T20260419-2156], [T20260420-0036])
• Workflow commands reorganized: stable entrypoints are orbit run ship <TASK_ID>..., orbit run ship --mode local <TASK_ID>..., orbit run ship-auto, orbit run duel-plan <TASK_ID>, and orbit run job <JOB_ID>. The direct orbit run <JOB_ID> shorthand and workflow-specific run ship list/show and run duel list/show commands were removed; use orbit run history and orbit run show for job-run inspection. ([T20260417-0248], [T20260419-0355], [T20260425-2010], [T20260426-0742])
• Task attribution history moved to orbit graph history: selector history is graph-owned, so the query now lives next to orbit graph search/show, and rebuilds use orbit graph build. Both orbit graph build and orbit graph history accept --task-id-pattern <regex>; workspace config knowledge.task_id_pattern is the steady-state setting, with CLI flag > config > Orbit default precedence. The selected pattern is recorded in manifest.json, and mismatches emit a stderr warning. orbit.graph.history exposes the same surface to MCP clients. ([T20260426-0507])

Features

• Activity/job v2 runtime: added schema v2 activities and jobs with typed DAG blocks (parallel, fan_out, loop, retry, when), activity name resolution, backend: auto normalization, backend: cli dispatch, HTTP agent loops, session-bound loop steps, and a v2 audit envelope with workspace provenance. ([T20260418-2018], [T20260418-2019], [T20260418-2143], [T20260418-2210], [T20260419-0002], [T20260419-0104])
• Seeded task pipelines: added load-bearing seeded workflows for PR, local, gate, auto-dispatch, and epic shipment, including task reservations, backlog bundling, admission-controlled dispatch, and session-backed epic orchestration. ([T20260419-0622-3], [T20260419-0623], [T20260419-0623-2], [T20260419-2347])
• Knowledge graph: added the Rust orbit-knowledge graph, orbit graph build/update/search/show, graph MCP tools, compact overviews, callers/implementors/dependency navigation, edit buffering, shared locks, auto-refresh, branch-scoped refs, task-ID attribution metadata, and markdown/config/table extraction. ([T20260411-0424], [T20260412-0645-2], [T20260412-0645-3], [T20260421-0358], [T20260421-0528], [T20260422-1540])
• MCP integrations: added the orbit-mcp crate, orbit mcp serve, safe default graph/task tool exposure, external MCP/plugin tooling, and orbit mcp init/remove setup for Claude, Codex, and Gemini clients. ([T20260418-0336], [T20260419-0236], [T20260422-1713], [T20260426-0354])
• Dashboard and observability: added orbit web serve; task, job, audit, scoreboard, and dashboard APIs; diagnostics and recent-runs views; task actions; copyable task IDs; connection health; skeleton/loading states; markdown rendering; and live-data animations. ([T20260417-0346], [T20260417-0412], [T20260417-0427], [T20260417-0437], [T20260417-0528], [T20260418-2004], [T20260426-0354])
• Task planning and search: added structured task plans, dependency support, epic task type support, selector-first task context, agent task search, and richer task field projection for agent/tool callers. ([T20260419-2300], [T20260420-0509-2], [T20260420-0521], [T20260421-0445], [T20260422-1756])
• Groundhog execution model (preview): added Groundhog chronicle serialization, workspace snapshots, verb tools, checkpoint verification, and a dedicated Groundhog v1 activity runner. ([T20260420-0509], [T20260420-0509-3], [T20260420-0509-4], [T20260420-0510], [T20260420-0510-2])
• Provider and evaluation support: added Gemini support, configurable agent/model selection, provider invocation traces, HTTP LoopTransport implementations for Anthropic/OpenAI-compatible/Gemini providers, planning duels, scoreboard attribution improvements, and versioned knowledge-graph benchmark harnesses. ([T20260411-1937-2], [T20260412-0457-2], [T20260412-1939], [T20260412-2129], [T20260418-0645], [T20260418-0759], [T20260422-1609])

Fixes

• Job run observability: orbit run ship --json, orbit run history, orbit run show, and direct orbit job run now retain actionable failure details and durable run-state/job-history records, including synthetic job-level steps for early v2 pipeline failures. ([T20260423-0445], [T20260423-2004-4], [T20260425-2010], [T20260426-0742])
• Branch-scoped knowledge graph refs: graph builds now write .orbit/knowledge/graph/refs/heads/<branch>.json files that point at immutable per-build indexes, reads default to the current git branch with default-branch fallback, and legacy .orbit/knowledge/graph/refs/current.json stores auto-migrate on first open/write. ([T20260421-0358])
• Knowledge graph hardening: graph reads and refreshes recover from corrupted stores, avoid stale worktree data, gate refresh/search hot paths, prune missing context files from locks, and hydrate task IDs idempotently during attribution. ([T20260416-0719], [T20260417-0307], [T20260420-0540], [T20260421-0652])
• Dispatch and locking correctness: task locks now detect directory/file overlaps, backlog selection filters locked groups, failed task-scoped runs move tasks to blocked with job/run/error context, and drained local batches no longer fail spuriously. ([T20260412-0443], [T20260417-0301], [T20260419-2109], [T20260420-0014])
• Workflow compatibility: merged object-valued job defaults with caller input, aligned the Quick Start approval flow with the current task lifecycle, and routed retired workflow inspection docs/errors to orbit run history/show. ([T20260423-0445], [T20260423-0447], [T20260423-2004-2], [T20260425-2010], [T20260426-0742])
• Release and developer tooling: restored release CI targets, repaired advertised developer targets, kept custom roots isolated, and fixed crashes/empty listings after seeded activity/job initialization. ([T20260419-2347], [T20260423-2004], [T20260423-2004-3], [T20260423-2004-5])
• Security and concurrency hardening: added localhost origin checks for web write endpoints, serialized diagnostics JSONL appends, hardened task-store concurrency, tightened filesystem/tool-runtime path boundaries, and strengthened agent protocol handling. ([T20260417-0557], [T20260417-0558], [T20260418-1928])

Chores

• Crate architecture: extracted orbit-common, orbit-knowledge, and orbit-mcp, merged the older orbit-types surface into orbit-common, decomposed execution/runtime modules, and kept crate dependency direction aligned with the documented architecture. ([T20260411-0008], [T20260419-2014])
• Documentation and positioning: added Orbit positioning docs, design-doc conventions, activity-job/knowledge-graph/Groundhog design docs, benchmark reports, and README updates for the current workflow and MCP surfaces.

0.2.0

Features

• Parallel batch execution: dispatch and execute multiple tasks in parallel with file-level conflict detection and shared worktrees
• Auto-cleanup on merge: ship workflow now deletes the remote branch after a successful PR merge

Fixes

• --parallelism flag: serialized as JSON integer instead of string, fixing schema validation failure on orbit run ship --parallelism N
• Stale default artifacts: orbit workspace init now always refreshes default skills, activities, and jobs to their latest embedded versions (custom artifacts are preserved)
• Clippy warning: resolved unused-mut warning and removed clippy from CI

Chores

• Default branch renamed from agent-main to main
• Removed orbit label from PR creation
• Agent configuration updates

0.1.0

Initial release of Orbit.

Core

• Task lifecycle: propose, approve, implement, review, and archive tasks with full history tracking
• Activity system: reusable operations with defined input/output schemas and three spec types (agent_invoke, cli_command, automation)
• Job engine: composable multi-step pipelines with conditional execution, retry logic, nested jobs, and parallel dispatch
• Workflow aliases: orbit run ship, orbit run ship-local, orbit run review as ergonomic entry points over raw job invocation
• Multi-agent orchestration: parallel task workers with file-level locking in shared worktrees
• Multi-model strategy: configurable agent/model per job step (e.g., Opus for planning, Codex for implementation)

CLI

• Grouped command surface: run workflows, manage work, configure and inspect
• JSON and table output modes across all commands
• Audit event logging for every CLI invocation

Infrastructure

• Layered Rust crate architecture (types, policy, exec, tools, store, agent, engine, core, cli)
• Two-root workspace model: global (~/.orbit/) and workspace-local (.orbit/)
• File-based (YAML) and SQLite persistence
• RBAC policy evaluation engine
• Process sandboxing and timeout handling
• Skill system for agent prompt composition