sinapse-ai 7.7.3 → 7.7.4

package/docs/guides/parallel-collaboration-source-of-truth.md

@@ -0,0 +1,481 @@
+# Parallel Collaboration Source Of Truth
+
+**Status:** Active  
+**Audience:** Caio, Sawyer/Soier, maintainers, and AI agents operating with SINAPSE  
+**Purpose:** Single source of truth for safe parallel collaboration in `sinapse-ai` and in any other repository that adopts the same setup
+
+---
+
+## Why This Document Exists
+
+This document defines the canonical operating model for two or more people evolving the same repository in parallel without overwriting code, weakening quality, or depending on informal memory.
+
+It exists for two reasons:
+
+1. `sinapse-ai` is a critical framework and must be protected with professional repository discipline.
+2. The same collaboration model must be portable to other projects that Caio and Sawyer/Soier maintain together.
+
+If another document says something different about parallel collaboration, this file wins.
+
+---
+
+## Core Principle
+
+**Nobody works directly on `main`.**
+
+All meaningful work must happen in an isolated branch, preferably inside an isolated worktree, and must reach `main` only through pull request review plus repository protections.
+
+That one rule removes the main source of accidental overwrite.
+
+---
+
+## Non-Negotiables
+
+These rules apply to `sinapse-ai` and to any project using this collaboration model.
+
+1. No direct work on `main`.
+2. No direct push to `main`.
+3. Every collaborator uses an isolated branch.
+4. Prefer one isolated worktree per feature, story, or scoped task.
+5. Every merge to `main` happens through PR review.
+6. Branches must be up to date with `origin/main` before push or merge.
+7. Obvious secrets must be blocked before commit.
+8. Quality gates must run before merge.
+9. Ownership and reviewer routing must use real GitHub users with write access.
+10. Destructive Git operations are blocked by default.
+
+---
+
+## What Problem This Solves
+
+This model is designed to prevent four common failure modes:
+
+1. Two people edit the same branch and silently overwrite each other.
+2. Someone starts work on `main`, mixes unrelated changes, and makes later recovery difficult.
+3. A branch is pushed while behind `origin/main`, making conflicts appear too late.
+4. Sensitive changes reach the repository without the other maintainer seeing them first.
+
+The goal is not "zero conflicts ever."  
+The goal is: **conflicts are detected early, contained safely, and never land invisibly in `main`.**
+
+---
+
+## Canonical Workflow
+
+### 1. Starting new work from a clean default branch
+
+Use:
+
+```bash
+npm run collab:start -- <story-id> <slug>
+```
+
+Example:
+
+```bash
+npm run collab:start -- 7.7.4 codex-collab-hardening
+```
+
+This command is the standard way to begin work safely. It:
+
+- checks that the current branch is the default branch
+- checks that the working tree is clean
+- fetches from `origin`
+- fast-forwards from `origin/main` when needed
+- detects the maintainer prefix
+- creates a dedicated worktree
+- creates a dedicated branch for that work item
+
+### 2. Recovering work that already started on `main`
+
+Use:
+
+```bash
+npm run collab:adopt -- <story-id> <slug>
+```
+
+Example:
+
+```bash
+npm run collab:adopt -- codex-runtime codex-functional-upgrade
+```
+
+This is the migration bridge when someone already has local work in `main`. It keeps the current files as they are and moves the work to a safe branch.
+
+### 3. Auditing the current state
+
+Use:
+
+```bash
+npm run collab:check
+```
+
+This tells the maintainer:
+
+- current branch
+- default branch
+- detected owner prefix
+- whether the tree is clean
+
+---
+
+## Branch Strategy
+
+### Canonical human maintainer branch format
+
+```text
+{owner}/{type}/{story-and-slug}
+```
+
+Examples:
+
+- `caio/feat/7-7-4-codex-collab-hardening`
+- `soier/refactor/framework-codex-runtime-hardening`
+- `caio/docs/parallel-collaboration-sot`
+
+### Owner detection
+
+The bootstrap detects the owner from local Git identity and environment:
+
+- `caio` -> `caio/`
+- `matheus`, `soier`, or `sawyer` -> `soier/`
+- anything else -> `dev/`
+
+### Important note about legacy internal worktree flow
+
+`sinapse-ai` still has an internal ADE worktree model using:
+
+```text
+auto-claude/{storyId}
+```
+
+That flow is still valid for internal engine-driven work.
+
+The rule is:
+
+- use `auto-claude/{storyId}` for the existing internal ADE worktree pipeline
+- use `npm run collab:start` for human maintainer collaboration
+
+Do not mix both branch styles in the same change.
+
+---
+
+## Worktree Policy
+
+The preferred policy is:
+
+- one worktree per active feature or story
+- one branch per worktree
+- one branch owner
+
+Why worktrees matter:
+
+- each collaborator gets a physically separate working directory
+- switching context becomes safer
+- uncommitted work from one feature does not leak into another
+- `main` can stay clean while several efforts run in parallel
+
+For `sinapse-ai`, the standard location is under:
+
+```text
+.sinapse/worktrees/
+```
+
+---
+
+## What Each Person Does
+
+### Caio
+
+- starts work with `collab:start` or `collab:adopt`
+- works inside his own branch or worktree
+- opens PRs for review
+- reviews Soier's PRs
+
+### Sawyer/Soier
+
+- starts work with `collab:start` or `collab:adopt`
+- works inside his own branch or worktree
+- opens PRs for review
+- reviews Caio's PRs
+
+### Shared rule
+
+The coordination point is the PR, not synchronous chat memory.
+
+---
+
+## Local Safety Rails
+
+The local repository should enforce these protections:
+
+### Pre-commit
+
+- block commits on `main`
+- block obvious secret leakage from staged files
+- keep existing framework protection checks
+
+### Pre-push
+
+- block push from `main`
+- block direct push to `main`
+- block push when the branch is behind `origin/main`
+
+These protections make the safe path the easiest path.
+
+---
+
+## GitHub Protection Baseline
+
+Every repo using this model should configure the default branch with the following minimum protections:
+
+### Required
+
+- PR required before merge
+- at least 1 approval required
+- dismiss stale reviews
+- require the most recent pusher to be approved by someone else
+- require conversation resolution
+- block force push
+- block branch deletion
+- require a validation check or CI summary check
+
+### Recommended
+
+- enforce admins
+- auto-delete merged branches
+- auto-merge when checks and approval are complete
+- CODEOWNERS enforcement after valid ownership is present in the default branch
+- secret scanning and push protection
+
+### Important sequencing rule
+
+Only enable GitHub's `require_code_owner_reviews` after the `CODEOWNERS` file already merged into the default branch is valid and references real collaborators with write access.
+
+Otherwise the repository can lock itself into a broken review state.
+
+---
+
+## CODEOWNERS Policy
+
+`CODEOWNERS` is not just documentation. It is operational ownership.
+
+Rules:
+
+1. Use real GitHub usernames only.
+2. Every listed owner must already have write access.
+3. Critical surfaces should require both maintainers as owners.
+4. Broad fallback ownership is acceptable, but sensitive paths must still be explicit.
+
+Recommended critical paths:
+
+- `.github/`
+- `.husky/`
+- `.claude/hooks/`
+- `.claude/rules/`
+- `.sinapse-ai/core/`
+- `.sinapse-ai/infrastructure/`
+- `bin/`
+- `packages/`
+- `package.json`
+- `package-lock.json`
+
+---
+
+## Merge And Review Model
+
+The standard model is:
+
+1. Maintainer opens feature branch or worktree
+2. Maintainer works locally
+3. Maintainer pushes only their feature branch
+4. Maintainer opens PR to `main`
+5. The other maintainer reviews
+6. CI and repository rules pass
+7. PR merges into `main`
+8. Local branches and worktrees are cleaned up
+
+### Review expectations
+
+The reviewer checks:
+
+- scope matches the branch intent
+- no unrelated files leaked in
+- required sync commands were run
+- docs and manifests stayed aligned when needed
+- the branch is safe to merge
+
+---
+
+## Conflict Handling
+
+### If both collaborators edit different files
+
+Usually there is no issue. Separate branches and PRs are enough.
+
+### If both collaborators edit the same file, different sections
+
+Git usually merges this automatically or the PR detects it safely before merge.
+
+### If both collaborators edit the same lines
+
+This becomes a visible branch or PR conflict and must be resolved before merge.
+
+That is acceptable. The point of this model is that the conflict appears in a contained place instead of silently overwriting someone else's work.
+
+---
+
+## Quality Gates
+
+For `sinapse-ai`, the baseline quality gates remain:
+
+```bash
+npm run lint
+npm run typecheck
+npm test
+```
+
+Additional sync or validation commands should run when relevant:
+
+```bash
+npm run sync:ide
+npm run validate:codex-sync
+npm run sync:skills:codex
+npm run validate:codex-skills
+npm run generate:manifest
+npm run validate:manifest
+```
+
+Projects that reuse this model should define their own equivalent minimum gates, but the principle remains the same:
+
+- no merge without local validation
+- no merge without remote validation
+
+---
+
+## `sinapse-ai` Specific Additions
+
+Inside `sinapse-ai`, collaboration also interacts with framework boundaries.
+
+### Autonomous lane
+
+Changes can usually proceed without waiting for the other maintainer when they stay within supported extension surfaces like:
+
+- `.sinapse-ai/development/agents/`
+- `.sinapse-ai/development/tasks/`
+- `.sinapse-ai/development/workflows/`
+- `.sinapse-ai/product/`
+- `docs/guides/`
+- `docs/framework/`
+
+### Coordinated lane
+
+Changes should be aligned first when they affect:
+
+- `.sinapse-ai/core/**`
+- `.sinapse-ai/infrastructure/**`
+- `bin/**`
+- `.husky/**`
+- `.claude/hooks/**`
+- release/versioning behavior
+- global framework policy
+
+This distinction is specific to `sinapse-ai`, but the same idea can be reused in other repos: define "safe solo surfaces" and "coordinate-first surfaces."
+
+---
+
+## How To Reuse This In Other Projects
+
+When applying this model to another repo, keep the following portable baseline:
+
+### Portable baseline
+
+- branch per maintainer
+- worktree per feature or story
+- no work on `main`
+- no direct push to `main`
+- PR review required
+- CI required
+- CODEOWNERS with real users
+- secret scan before commit
+- behind-main check before push
+
+### Repo-specific pieces to customize
+
+- branch owner prefixes
+- quality gates
+- critical paths in `CODEOWNERS`
+- CI check names
+- file sync commands
+- release rules
+
+### Adoption checklist for another repo
+
+1. Add maintainer bootstrap command
+2. Add pre-commit and pre-push safety hooks
+3. Add `CODEOWNERS`
+4. Configure GitHub branch protection
+5. Define repo-specific quality gates
+6. Document autonomous vs coordinated paths if the repo has a framework layer
+
+---
+
+## Anti-Patterns
+
+These are forbidden:
+
+- starting serious work directly on `main`
+- pushing directly to `main`
+- sharing one long-lived branch between two people
+- opening PRs from outdated branches
+- relying on memory instead of explicit PR handoff
+- committing `.env` files or obvious secrets
+- using force push as a normal workflow
+- changing critical repository policy without coordinated review
+
+---
+
+## Recovery Rules
+
+### If someone already started in `main`
+
+Use `collab:adopt`.
+
+### If a branch is behind `origin/main`
+
+Update it before push or merge.
+
+### If a conflict appears
+
+Resolve it in the feature branch or PR, never by bypassing protections on `main`.
+
+### If `CODEOWNERS` or GitHub protections are misconfigured
+
+Fix ownership first, then activate stricter review enforcement.
+
+---
+
+## Decision Summary
+
+The canonical answer to "how do two maintainers work in sync without overwriting code?" is:
+
+- isolated branch
+- isolated worktree
+- no work on `main`
+- PR review before merge
+- GitHub protection on `main`
+- local safety hooks
+- quality gates before merge
+
+Everything else is optimization on top of that base.
+
+---
+
+## Related Documents
+
+These documents support this source of truth, but do not override it:
+
+- [Framework Contributor Mode](./framework-contributor-mode.md)
+- [Parallel Workflow](./parallel-workflow.md)
+- [Safe Collaboration Rule](../../.claude/rules/safe-collaboration.md)
+- [Safe Collaboration Template README](../../.sinapse-ai/infrastructure/templates/safe-collab/README.md)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "sinapse-ai",
-  "version": "7.7.3",
+  "version": "7.7.4",
   "description": "SINAPSE AI: Framework de orquestracao de IA — 18 squads, 175 agentes especializados",
   "bin": {
     "sinapse": "bin/sinapse.js",
@@ -20,9 +20,13 @@
     ".claude/CLAUDE.md",
     ".claude/rules/",
     ".claude/hooks/",
+    ".codex/catalog.json",
+    ".codex/command-registry.json",
     ".codex/instructions.md",
     ".codex/agents/",
+    ".codex/scripts/",
     ".codex/skills/",
+    ".codex/tasks/",
     "squads/squad-*/**",
     "squads/claude-code-mastery/**",
     "squads/sinapse/**",
@@ -72,16 +76,20 @@
     "sync:ide:validate": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js validate",
     "sync:ide:check": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js validate --strict",
     "sync:ide:claude": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js sync --ide claude-code",
-    "sync:ide:codex": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js sync --ide codex",
+    "sync:ide:codex": "node .sinapse-ai/infrastructure/scripts/sync-codex-local-first.js",
     "validate:claude-sync": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js validate --ide claude-code --strict",
     "validate:claude-integration": "node .sinapse-ai/infrastructure/scripts/validate-claude-integration.js",
-    "validate:codex-sync": "node .sinapse-ai/infrastructure/scripts/ide-sync/index.js validate --ide codex --strict",
+    "validate:codex-sync": "node .sinapse-ai/infrastructure/scripts/validate-codex-sync.js",
     "validate:codex-integration": "node .sinapse-ai/infrastructure/scripts/validate-codex-integration.js",
+    "validate:codex-commands": "node .sinapse-ai/infrastructure/scripts/validate-codex-command-registry.js",
     "sync:skills:codex": "node .sinapse-ai/infrastructure/scripts/codex-skills-sync/index.js",
     "sync:skills:codex:global": "node .sinapse-ai/infrastructure/scripts/codex-skills-sync/index.js --global --global-only",
     "validate:codex-skills": "node .sinapse-ai/infrastructure/scripts/codex-skills-sync/validate.js --strict",
     "validate:paths": "node .sinapse-ai/infrastructure/scripts/validate-paths.js",
     "validate:parity": "node .sinapse-ai/infrastructure/scripts/validate-parity.js",
+    "collab:adopt": "node bin/utils/collab-start.js --adopt-current",
+    "collab:check": "node bin/utils/collab-start.js --check",
+    "collab:start": "node bin/utils/collab-start.js",
     "validate:semantic-lint": "node scripts/semantic-lint.js",
     "manifest:ensure": "node scripts/ensure-manifest.js",
     "validate:publish": "node bin/utils/validate-publish.js",

package/packages/installer/tests/unit/entity-registry-bootstrap.test.js

@@ -97,9 +97,9 @@ describe('Entity Registry Bootstrap (Story INS-4.6)', () => {
   });
 
   describe('AC3: No duplication with pre-push hook', () => {
-    test('pre-push hook calls ids-pre-push.js (incremental), not populate script', () => {
+    test('pre-push hook calls pre-push-safety.js, not populate script', () => {
       const hookContent = fs.readFileSync(PRE_PUSH_HOOK, 'utf8');
-      expect(hookContent).toContain('ids-pre-push.js');
+      expect(hookContent).toContain('pre-push-safety.js');
       expect(hookContent).not.toContain('populate-entity-registry.js');
     });