Skip to content

Latest commit

 

History

History
276 lines (223 loc) · 17.2 KB

CHANGELOG.md

File metadata and controls

276 lines (223 loc) · 17.2 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

6.0.0 - 2026-03-05

Added

  • Specification Standards — synced from ai-jumpstart v6.0.0: complete spec template inline in AGENTS.md defining prototype hierarchy (v1 → Milestones → Workstreams → Sections → Dimensions)
  • Mistral AI provider expansion — additional model variants in .config/opencode/opencode.json (mistral-large-2512, codestral-2508, codestral-2502)
  • Vibe agent support — added to run.sh deployment configuration

Changed

  • AGENTS.md — synced from ai-jumpstart v6.0.0 with canonical spec template
  • CHANGELOG.md — synced from ai-jumpstart v6.0.0
  • .zshrc environment loading — refactored with modular functions (load-buffet-env, load-clawable-env), manifest-based key discovery, AGENT_ENV profile support
  • run.sh formatting — normalized indentation (tabs instead of spaces)
  • VERSION — bumped to 6.0.0

Removed

  • project_spec.md — ai-jumpstart-specific, not a template for general use

5.1.0 - 2026-03-05

Added

  • Standard Make target taxonomy in AGENTS.md — canonical table for dev, up, down, lint, test, build, _clean, qa, qa-smoke; bans make quality, forbids make qa-headed as a shared target
  • Design system defaults in skills/frontend-design/SKILL.md — Geist font pair, CSS token palette (#FF6B35 neon orange, #39FF85 terminal green, #0A0A0A dark base), dot-grid background, hero glow, card hover, agent terminal surface patterns
  • docs/pilot-spec.md — pilot program specification

Changed

  • skills/create-scaffold/SKILL.md — replaced make quality with make lint; added full target taxonomy and web-specific targets table
  • skills/e2e-qa-playwright/SKILL.md — removed make qa-headed; added BASE_URL switching pattern; chromium-only CI rule
  • skills/frontend-design/SKILL.md — stack updated to Vite 7 + Tailwind v4 CSS-first (@import "tailwindcss"); Make targets section added
  • docs/scaffolds/website-template.md — updated stack to Vite 7 + Tailwind v4; Make targets table; verification uses make lint/test/build/qa-smoke
  • skills/oracle/SKILL.md — confirmed canonical v0.9.0 CLI-centric version
  • run.sh — added Vibe config and agent entries

5.0.0 - 2026-03-02

Added

  • Tombstone v5.0 - Complete rewrite in Zig with new architecture
    • Milestone 1: Foundation - CLI parsing, Makefile, tests, 212KB binary
    • Milestone 5: Setup automation - replaces run.sh with tombstone setup
    • Milestone 6: Directory sync - automates ai-jumpstart → dotfiles sync
    • New spec: docs/spec/v5/PROJECT_SPEC.md with 7 milestones to v1.0
    • Commands: tombstone setup, tombstone sync, tombstone status
    • Performance: 3-5ms execution, <500KB binary target

Changed

  • Project focus shift - From agent router to development automation tool
  • Milestones reordered: Setup (5) and Sync (6) prioritized over routing (2-4)

4.2.0 - 2026-03-02

Added

  • Tombstone agent router specification - comprehensive design doc for intelligent agent routing, setup automation, and directory sync (see tombstone/SPEC.md)

4.1.1 - 2026-02-25

Changed

  • Migrated Codex feature flag from deprecated collab = true to [features].multi_agent = true in .codex/config.toml

4.1.0 - 2026-02-25

Added

  • Mistral AI provider support — added MISTRAL_API_KEY to Proton Pass AGENTS_BUFFET vault
  • Pi agent configuration — added Mistral provider with mistral-large-latest and codestral-latest models
  • Opencode configuration — added Mistral provider with OpenAI-compatible API endpoint
  • Dotfiles sync — synced .pi/agent/models.json and .config/opencode/opencode.json to dotfiles repo

4.0.0 - 2026-02-23

Changed

  • Profile maintenance: cleaned up blank lines in Owner sections (AGENTS.md, mac-vm.md)

3.0.0 - 2026-02-22

Changed

  • Oracle CLI migrated — all references updated from @steipete/oracle to @indykish/oracle (v0.9.2) across AGENTS.md, CLAUDE.md, and all skills
  • Oracle install — changed from git clone+build to npm install -g @indykish/oracle / npx @indykish/oracle
  • Model defaults bumped — primary claude-4.6-sonnet, escalation claude-4.6-opus (was 4.5/4.1)
  • oracle SKILL.md — restructured: inline CTO/Engineer review as primary path, @indykish/oracle CLI moved to secondary; added Review Severity Levels (BLOCKER/MAJOR/MINOR/NIT)
  • DX Platform Stack — moved from AGENTS.md into docs/STACK.md; AGENTS.md now has a one-line pointer

Added

  • Cognitive Discipline section in AGENTS.md — Non-trivial definition, Confusion Management, Simplicity Enforcement, Dead Code Hygiene (from AGENTS_OLD learnings)
  • Oracle Escalation Levels (Level 1–4) in AGENTS.md — single-agent → inline review → CLI → parallel worktrees
  • "Do Not Drift" clause in AGENTS.md Skills Policy
  • docs/BEHAVIORAL_GUARDRAILS.md — cognitive discipline extracted from AGENTS_OLD: assumption surfacing, confusion management, simplicity, scope discipline, dead code hygiene, push-back triggers, failure modes, post-work template
  • docs/STACK.md — DX platform stack defaults (Website, CLI, Desktop, Mobile, Frontend, Accessibility)
  • Dotfiles sync tracking in AGENTS.md — file sources table, tree compare one-liner (now includes docs/), sync command block; README.md excluded

2.8.0 - 2026-02-21

Added

  • E2E Networks internal setup in preflight-tooling skill — openfortivpn, mysql-client, VPN launch scripts with LaunchAgent plists
  • VPN env var workflow — credentials via ~/.env_mac (no hardcoded passwords); run.sh warns to set VPN_PROD_* / VPN_STAGE_*
  • Tooling inventory — E2E internal tools section with openfortivpn, mysql-client, font-hack-nerd-font

Changed

  • Oracle skill rewritten — defaults to --engine api --model claude-4.5-sonnet (Claude-first); browser path optional
  • AGENTS.md — Oracle tool defaults updated to Claude Sonnet primary, Opus escalation
  • CLAUDE.md — Added Oracle skill defaults pointer

Removed

  • AGENTS_OLD.md — legacy CTO/Engineer dual-role model (superseded by Oracle Operating Model)
  • docs/01-preflight-vpn-localdev.md — spec consumed and implemented
  • docs/00-oracle-claude-browser-feasibility.md — spec consumed and implemented

2.7.0 - 2026-02-20

Changed

  • Dotfiles sync standardized — Tree structure aligned with ai-jumpstart as canonical source of truth
  • Stale files cleaned up — Removed orphaned skills/go.md, skills/python.md, skills/SKILL.md, root-level status-line.sh, setup-owner.sh, and duplicate docs/mac-vm.md from dotfiles
  • mise.toml consolidated — Removed stale root-level mise.toml from dotfiles; authoritative config lives at .config/mise/config.toml

2.6.0 - 2026-02-20

Added

  • Pi agent support — New .pi/agent/ directory with auth.json, models.json, and settings.json configs
  • Environment-based secrets — All API keys in configs use environment variables ($OPENROUTER_API_KEY, $ZAI_API_KEY, $MINIMAX_API_KEY, etc.)
  • scripts/setup-owner.sh — Interactive setup script to personalize AGENTS.md and runbooks with user details
  • Git email auto-detection — Setup script attempts to read email from git config before prompting
  • Status line scripts — .claude/status-line.sh and .claude-e2e/status-line.sh with emoji-rich output showing project info and costs
  • Version display — run.sh now shows version from VERSION file on startup

Changed

  • run.sh refactored — Fully modular with functions: collect_skills(), deploy_configs(), deploy_agent(), etc.
  • run.sh moved — Relocated from scripts/ to root directory for easier execution
  • AGENTS.md templated — Owner profile now uses {{OWNER_NAME}}, {{OWNER_HANDLE}}, {{DISCORD_HANDLE}}, {{OWNER_EMAIL}}, {{HARDWARE}} placeholders
  • mac-vm.md templated — Account email uses {{OWNER_EMAIL}} placeholder
  • Setup notices added — AGENTS.md and runbooks display "Setup Required" notices until setup-owner.sh is run
  • Shellcheck compliance — All scripts pass shellcheck with proper variable scoping and error handling

Removed

  • scripts/run.sh — Deleted (moved to root)

2.5.0 - 2026-02-20

Changed

  • Hardcoded /Users/kishore/ paths removed — All absolute user paths in AGENTS.md, CLAUDE.md, docs/scaffolds/, and docs/codex-limits.md replaced with $HOME/
  • CLAUDE.md file:// link fixed — Pointer to AGENTS.md now uses a relative path instead of an absolute file:// URL
  • Zoho skill install steps removed — zoho-desk and zoho-sprint SKILL.md files no longer include manual cp commands; run.sh handles script deployment
  • Scaffold clone hints added — skills/create-scaffold/stacks/go.md and stacks/python.md now include fallback clone URLs for missing reference repos
  • run.sh reminder block polished — Replaced raw empty export lines with a clean aligned reminder referencing the Proton Pass vault
  • README.md — Removed Z_AI_API_KEY; kept ZAI_API_KEY

2.4.0 - 2026-02-20

Added

  • KiloCode agent support — run.sh deploys KiloCode config to ~/.config/kilo/opencode.json
  • OpenCode permissions — Added rm deny, trash allow, directory blocks to OpenCode config (was missing entirely)
  • Config file deployment — run.sh now deploys agent configs (OpenCode, KiloCode) via CONFIG_MAP
  • Dotfile backup rule — AGENTS.md and CLAUDE.md require timestamped backup before editing dotfiles
  • Agent deployment targets — run.sh now deploys to ~/.config/agents and ~/.pi/agent in addition to existing agents
  • Claude E2E agent support — run.sh now deploys to ~/.claude-e2e if directory exists
  • Optional agent deployment — Agents are only deployed if their home directory exists (prevents creating unused directories)
  • Handoff skill — Standardized checklist for packaging work state when switching agents
  • Pickup skill — Standardized checklist for rehydrating context when resuming work
  • Subagent coordination guide — docs/subagent.md documents multi-agent tmux workflows
  • Zoho Desk pull enhancements — Rate limiting with credit buffer, --until date filter, configurable concurrency/delay options
  • Zed editor guidance — Notes on using open -a Zed when CLI not on PATH

Changed

  • rm removed from deny list across Claude, Claude E2E, KiloCode, and OpenCode configs. rm now prompts for approval instead of being hard-blocked. trash remains the preferred delete command.
  • Hardcoded paths removed — Scaffold stacks and Zoho skill docs now use $HOME/ instead of /Users/kishore/
  • OpenCode config format — Replaced <ENV:...> placeholders with {env:...} format
  • Agent profile config dir — Renamed from ~/.config/e2e/agent-profiles to ~/.config/clawable
  • run.sh formatting — Normalized indentation from spaces to tabs
  • Zoho Desk thread content — Fixed to use plainText fallback, removed summary field
  • Zoho Desk output format — Pull now writes per-ticket JSONL files in <YYYY>/<MM>/<DD>_<HHMMSS>_<ticket#>.txt structure (was batched YAML)
  • Proton Pass vault naming — Updated from WORK_E2E_AGENTS to AGENTS_BUFFET with GITLAB_PERSONAL_ACCESS_TOKEN item

2.0.0 - 2026-02-16

Added

  • Oracle Operating Model — CLAUDE.md and AGENTS.md auto-loaded by every supported agent
  • scripts/run.sh — Idempotent deploy script that syncs profiles and skills to Claude Code, Codex, OpenCode, AmpCode, and KiloCode
  • 10 production-ready skills — preflight-tooling, review-pr, create-scaffold, create-cli, e2e-qa-playwright, oracle, update-docs-and-commit, frontend-design, zoho-sprint, zoho-desk
  • Agent profile templates in agents/ for all five supported agents
  • Codex config in config/codex.config.toml
  • Operational docs — docs/tooling-inventory.md, docs/worktree-tmux.md, docs/codex-limits.md, docs/RELEASE.md, docs/install-and-test.md
  • Scaffold references in docs/scaffolds/ for Python, Go, Rust, TypeScript, JavaScript CLI, and Tauri stacks
  • Runbooks directory for operational playbooks
  • Zoho integration scripts — scripts/zoho-sprint.mjs, scripts/zoho-desk.mjs
  • Deterministic forge detection — awakeninggit.e2enetworks.net / gitlab.com → glab; github.com → gh

Removed

  • agents/ directory and --local flag from run.sh — redundant copy of profiles already in repo root

Changed

  • README.md rewritten with Getting Started (run.sh), deployment matrix, skills catalog, and working-efficiently guide
  • CHANGELOG.md 1.0.0 entry corrected to reflect repository-local workflow instead of external skills repo
  • update-docs-and-commit skill now includes optional version bumping step
  • Git Forge Policy updated with explicit host-based routing rules

1.0.0 - 2026-02-09

Changed

  • Transformed repo purpose from agent installer to learning resource hub
  • Rewrote README.md as curated collection of AI learning materials:
    • Links to AI Bootcamp presentation series (Week 1-4)
    • Links to MyAccount 2/SRE engineering presentations
    • Links to YouTube videos (PM with AI Tools)
    • Links to team resources (Zoho WorkDrive, internal repos)
    • Design for Agents guidance with Mermaid diagrams
  • Moved skills and scripts into a separate skills distribution flow (now deprecated)
  • Kept CLAUDE.md as reference material for understanding agent configuration patterns

Removed

  • install.sh — Replaced by repository-local workflow docs
  • skills/ directory — Moved to dedicated skills repository
  • scripts/ directory — Moved to dedicated skills repository
  • Agent installation logic — Replaced with standardized npx skills workflow

Migration Notes

Users should use the repository-local assets:

./scripts/run.sh ~/.config/e2e/agent-profiles

Then start from skills/preflight-tooling.md.

0.9.0 - 2026-02-04

Added

  • Agent-agnostic configuration system with CLAUDE.md and AGENTS.md for vendor-neutral workflows
  • Idempotent installer (install.sh) that detects and configures 8 AI coding agents:
    • Claude Code, Cursor, Zed, AmpCode, OpenCode, KiloCode, Antigravity, Pi
  • Centralized configuration via ~/.config/e2e/ai-jumpstart/ with symlinked skills
  • 11 shared skills for common engineering workflows:
    • Development workflow: zoho-setup, create-issue, create-plan, execute, explore, update-docs-and-commit
    • Code review: peer-review, frontend-review, backend-review
    • Design & learning: frontend-design, learning-opportunity
  • Zoho Sprint integration via scripts/zoho-sprint.mjs with OAuth authentication
  • Mermaid diagram standards for architecture documentation with human verification checkpoints
  • Colored terminal output in installer (green ✓, red ✗, yellow ⚠)

Changed

  • Migrated from agent-specific configs to centralized symlink-based architecture
  • Standardized skill paths across all 8 supported agents

Documentation

  • Added PROJECT_SPEC.md with complete Phase 2 specification
  • Added PROJECT_STATUS.md for progress tracking
  • Added README.md with Mermaid architecture diagrams and quick start guide
  • Added CHANGELOG.md (this file)
  • Restructured README.md for clarity and completeness:
    • Added "What You Get" section highlighting 3 main value propositions
    • Added Troubleshooting section with common issues and solutions
    • Added Configuration Files section explaining CLAUDE.md and skills directory
    • Grouped Available Skills into Workflow Automation and Code Quality
    • Enhanced Typical Workflow diagram with code review and CAL usage example
    • Added skill design principles from Principal Engineer review