SCE - Scene Capability Engine

SCE is a scene capability orchestration engine for AI-native software delivery.
It provides a deterministic path from goal -> scene -> spec -> patch -> verify -> release.

English | 简体中文

---

Why SCE

SCE is designed for teams that want AI agents to deliver software end-to-end without losing control.

• Keep delivery aligned to requirements through Spec-first workflows.
• Scale from single-task execution to multi-agent program orchestration.
• Prevent silent drift with mandatory gates, ontology checks, and release evidence.
• Preserve local work history with timeline snapshots, not just Git pushes.

---

Core Capabilities

Capability	What SCE Provides	Outcome
Scene + Spec model	Scene-governed sessions and Spec lifecycle (requirements/design/tasks)	Stable context across long AI runs
Auto intake + Spec governance	Goal intent detection, auto spec bind/create, scene portfolio governance	Automatic scene-to-spec tracking with bounded spec growth
Studio workflow	studio plan -> generate -> apply -> verify -> release	Structured chat-to-release execution
Autonomous delivery	auto close-loop, close-loop-program, close-loop-controller	Unattended bounded convergence
Multi-agent orchestration	DAG scheduling, retries, 429 adaptive parallel control	Reliable parallel execution at scale
Domain/ontology governance	problem-domain chain + scene template + gate validation	Fewer semantic regressions
Problem closure loop	problem-domain map + chain + problem-contract + closure gate	Root-cause-first fixes with bounded convergence
Problem evaluation routing	Stage-level risk/evidence/readiness scoring with mandatory policy	Adaptive execution strategy with guarded apply/release
Local timeline safety	timeline save/auto/list/show/restore/push + key-event auto checkpoints	Recoverable local history
Errorbook-driven repair	Local + registry-backed error patterns and release gates	Faster diagnosis and safer fixes
Release governance	Git-managed gate, errorbook gate, handoff preflight, tag pipeline	Auditable, reproducible releases

---

3-Minute Quick Start

# 1) Install
npm install -g scene-capability-engine

# 2) Adopt in your project
sce adopt

# 3) Open a primary scene session
sce studio plan --scene scene.demo --from-chat session-demo --goal "bootstrap first feature" --json

# 4) Bootstrap and run one Spec
sce spec bootstrap --name 01-00-first-feature --scene scene.demo --non-interactive
sce spec pipeline run --spec 01-00-first-feature --scene scene.demo

For autonomous execution:

sce auto close-loop "deliver customer + order + inventory baseline"

---

Recommended Workflows

1) Feature Delivery (default)

sce studio plan --scene scene.customer-order --from-chat session-20260302 --goal "optimize checkout"
sce spec bootstrap --name 02-00-checkout-optimization --scene scene.customer-order --non-interactive
sce spec domain coverage --spec 02-00-checkout-optimization --json
sce spec gate run --spec 02-00-checkout-optimization --scene scene.customer-order --json

2) Program-Scale Autonomous Delivery

sce auto close-loop-program "stabilize order lifecycle and release governance" --program-govern-until-stable --json

3) Local History Safety (timeline)

sce timeline save --summary "before risky refactor"
sce timeline list --limit 20
sce timeline restore <snapshot-id>
sce timeline push origin main

4) Release Baseline

sce auto handoff preflight-check --require-pass --json
git tag -a vX.Y.Z -m "vX.Y.Z"
git push origin vX.Y.Z

---

Default Problem-Solving Loop

SCE now enforces a domain-closed diagnosis and repair route by default:

1. Scope the problem first with scene artifacts (problem-domain-map.md, scene-spec.md, problem-domain-chain.json, problem-contract.json).
2. Keep trial-and-error history in incident staging (.sce/errorbook/staging/incidents/) to avoid repeating failed attempts.
3. Use problem evaluation to prioritize likely impact areas before applying/releasing changes.

Hard rule defaults:

• After two failed rounds on the same problem fingerprint, debug evidence is required in subsequent attempts.
• studio verify/release run problem-closure-gate by default when a spec is bound.
• studio plan auto-runs goal intake (bind existing spec or create spec) and writes scene portfolio governance snapshots by default.
• studio plan --manual-spec and --no-spec-governance are blocked by default policy; use policy overrides only when absolutely necessary.
• Historical specs can be scene-governed incrementally via sce studio backfill-spec-scenes --apply (writes .sce/spec-governance/spec-scene-overrides.json).

---

AI Agent Compatibility

SCE is tool-agnostic and works with Codex, Claude Code, Cursor, Windsurf, VS Code Copilot, and other CLI-capable agents.

• Runtime context is managed by .sce/ (not IDE-specific hidden folders).
• Session governance is scene-first: 1 scene = 1 primary session.
• Spec work is attached as child sessions and auto-archived.
• Startup now auto-detects adopted projects and aligns takeover baseline defaults automatically.
• Problem evaluation policy is enabled by default (.sce/config/problem-eval-policy.json) and evaluates every Studio stage.
• Problem closure policy is enabled by default (.sce/config/problem-closure-policy.json) and blocks verify/release bypass when required domain/problem evidence is missing.
• Error handling now follows a full incident loop by default: every record attempt is staged first and auto-closed on verified/promoted outcomes.
• You can inspect or force-align baseline explicitly:
  • sce workspace takeover-audit --json
  • sce workspace takeover-apply --json

---

Important Version Changes

• 3.5.1: Enforced stricter Studio intake defaults (--manual-spec and --no-spec-governance blocked unless policy override), added historical spec scene backfill command (sce studio backfill-spec-scenes) and persisted override mapping (.sce/spec-governance/spec-scene-overrides.json) for portfolio/related-spec alignment.
• 3.5.0: Added Studio automatic goal intake + scene spec portfolio governance (sce studio intake, sce studio portfolio), including default intake policy baseline and governance artifacts for bounded scene spec growth.
• 3.4.6: Added default problem-closure-gate + problem-contract baseline and strengthened mandatory problem evaluation dimensions (problem_contract/ontology_alignment/convergence) for verify/release convergence control.
• 3.4.5: git-managed-gate now treats worktree checks as advisory in default relaxed CI mode (CI/GITHUB_ACTIONS, non-strict), preventing false release blocking.
• 3.4.4: Added SCE_GIT_MANAGEMENT_ALLOW_UNTRACKED=1 / --allow-untracked; release workflow uses it for npm publish after generating release evidence artifacts.
• 3.4.3: Introduced mandatory problem evaluation across Studio stages (plan/generate/apply/verify/release) with policy file .sce/config/problem-eval-policy.json and stage report artifacts.
• 3.4.2: Errorbook incident flow moved to full staging closed-loop (attempt history, incident inspection, resolved archive).
• 3.4.1: Added workspace takeover baseline automation (takeover-audit / takeover-apply) and startup alignment defaults.

---

Documentation Map

Start here:

• Quick Start
• Command Reference
• Autonomous Control Guide
• Scene Runtime Guide
• Value Observability Guide
• Multi-Agent Coordination Guide
• Errorbook Registry Guide
• Documentation Hub

Moqui-focused:

• Moqui Template Core Library Playbook
• Moqui Standard Rebuild Guide

---

Community

• GitHub Discussions
• GitHub Issues

Scan the QR code and note sce to join the WeChat group.

---

License

MIT. See LICENSE.

---

Version: 3.5.1
Last Updated: 2026-03-03