@sienklogic/plan-build-run 2.1.0 → 2.1.1

package/plugins/copilot-pbr/references/continuation-format.md

@@ -0,0 +1,212 @@
+# Continuation Protocol
+
+How to spawn a fresh executor agent to continue from a checkpoint or partial completion.
+
+---
+
+## Why Fresh Agents
+
+When an executor hits a checkpoint, it STOPS and returns. A new, fresh agent must be spawned to continue because:
+1. The checkpoint may require user input that changes the execution path
+2. Context isolation prevents accumulated state from causing issues
+3. Fresh agents have full context budget for remaining work
+
+---
+
+## Lean Continuation State File
+
+Instead of inlining all prior context into the continuation prompt, the build skill writes a compact state file to disk. The continuation executor reads this file + the original PLAN.md, keeping the prompt lean.
+
+### File Location
+
+```
+.planning/phases/{NN}-{slug}/.continuation-state.json
+```
+
+### Schema
+
+```json
+{
+  "plan_id": "02-01",
+  "completed_tasks": [
+    {
+      "id": "02-01-T1",
+      "commit": "abc1234",
+      "files": ["src/auth/discord.ts", "src/auth/types.ts"]
+    },
+    {
+      "id": "02-01-T2",
+      "commit": "def5678",
+      "files": ["src/middleware/auth.ts"]
+    }
+  ],
+  "checkpoint": {
+    "task": "02-01-T3",
+    "type": "human-verify",
+    "resolution": "passed"
+  },
+  "resume_at": "02-01-T3"
+}
+```
+
+### Fields
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `plan_id` | string | The plan being executed |
+| `completed_tasks` | array | Tasks completed before checkpoint. Each has `id`, `commit` hash, and `files` changed. |
+| `checkpoint` | object | The checkpoint that triggered continuation. `task` = task ID, `type` = checkpoint type, `resolution` = user's response. |
+| `resume_at` | string | Task ID to resume execution from |
+
+### How the Build Skill Uses It
+
+**Writing** (when executor hits a checkpoint):
+1. Executor returns checkpoint info to the build orchestrator
+2. Build orchestrator writes `.continuation-state.json` with completed tasks, checkpoint details, and resume point
+3. Build orchestrator prompts user for checkpoint resolution
+4. Build orchestrator updates `checkpoint.resolution` with user's response
+
+**Reading** (when spawning continuation executor):
+1. Build orchestrator reads `.continuation-state.json` from disk
+2. Build orchestrator reads the original PLAN.md from disk
+3. Build orchestrator constructs a lean continuation prompt referencing both files
+4. Continuation executor reads `.continuation-state.json` itself to verify prior commits
+
+This approach replaces inlining the full completed-tasks table into the prompt, saving significant context budget on plans with many tasks or multiple checkpoints.
+
+---
+
+## Continuation Context Structure
+
+The continuation prompt must include:
+
+### 1. Completed Tasks Table
+
+A table of all tasks completed before the checkpoint, with their commit hashes:
+
+```markdown
+| Task ID | Task Name | Commit | Files Changed | Status |
+|---------|-----------|--------|---------------|--------|
+| 02-01-T1 | Create auth module | abc1234 | src/auth/discord.ts, src/auth/types.ts | complete |
+| 02-01-T2 | Add auth middleware | def5678 | src/middleware/auth.ts | complete |
+| 02-01-T3 | Wire OAuth callback | — | — | checkpoint |
+```
+
+### 2. Resume Task Number
+
+Which task to resume from:
+- If the checkpoint task needs to be RE-EXECUTED with user input: resume at the checkpoint task
+- If the checkpoint task was completed and the next task is blocked: resume at the NEXT task
+
+### 3. User Response
+
+The user's response to the checkpoint:
+- For `decision`: which option they chose
+- For `human-verify`: whether they passed or failed each item
+- For `human-action`: confirmation that the action was taken
+
+---
+
+## Continuation Prompt Template
+
+```
+You are the executor agent. Continue executing a plan from a checkpoint.
+
+<plan>
+[Inline the FULL PLAN.md content — all tasks, not just remaining ones]
+</plan>
+
+<completed_tasks>
+The following tasks have already been completed. DO NOT re-execute them.
+Verify their commits exist before proceeding.
+
+| Task ID | Task Name | Commit | Status |
+|---------|-----------|--------|--------|
+| {id} | {name} | {hash} | complete |
+| {id} | {name} | {hash} | complete |
+| {id} | {name} | — | checkpoint |
+</completed_tasks>
+
+<checkpoint_resolution>
+Checkpoint type: {human-verify | decision | human-action | architectural-change}
+User response: {the user's response to the checkpoint}
+
+{For decision: "User chose Option {X}: {description}"}
+{For human-verify: "User confirmed: {pass/fail per item}"}
+{For human-action: "User confirms action completed: {what they did}"}
+{For architectural-change: "User approved approach: {new approach}"}
+</checkpoint_resolution>
+
+<resume_instructions>
+Resume execution at Task {N}.
+{If checkpoint task should be re-executed: "Re-execute Task {N} with the user's input."}
+{If checkpoint task is done: "Skip Task {N} and continue from Task {N+1}."}
+</resume_instructions>
+
+<project_context>
+[Same project context as original spawn — config, CONTEXT.md, STATE.md]
+</project_context>
+
+<prior_work_this_phase>
+[Include all SUMMARY.md files from other completed plans in this phase]
+</prior_work_this_phase>
+
+Instructions:
+1. Verify completed task commits exist: git log --oneline -n {count}
+2. If commits are missing, STOP and return error
+3. Resume execution at Task {N}
+4. Create atomic commits for each remaining task
+5. Write SUMMARY.md when complete (include ALL tasks — completed and newly executed)
+6. Run self-check
+```
+
+---
+
+## Verification Before Resuming
+
+The continuation agent MUST verify prior work before continuing:
+
+```bash
+# Check that expected commits exist
+git log --oneline -n {expected_commit_count}
+
+# Check that expected files exist
+ls -la {each file from completed tasks}
+```
+
+If verification fails:
+- STOP immediately
+- Report which commits/files are missing
+- DO NOT attempt to re-create them (that's the orchestrator's job)
+
+---
+
+## SUMMARY.md for Continued Plans
+
+The final SUMMARY.md must include ALL tasks — both pre-checkpoint and post-checkpoint:
+
+```markdown
+## Task Results
+
+### Task 1: {name}
+- **Commit**: `{hash}` - {message}
+- **Status**: complete (pre-checkpoint)
+
+### Task 2: {name}
+- **Commit**: `{hash}` - {message}
+- **Status**: complete (pre-checkpoint)
+
+### Task 3: {name} (checkpoint: {type})
+- **Checkpoint resolution**: {user response}
+- **Commit**: `{hash}` - {message}
+- **Status**: complete (post-checkpoint)
+```
+
+---
+
+## Multiple Checkpoints
+
+If a plan has multiple checkpoint tasks:
+- Each checkpoint requires a separate continuation agent spawn
+- The completed_tasks table grows with each continuation
+- Each continuation verifies ALL prior commits, not just the most recent

package/plugins/copilot-pbr/references/deviation-rules.md

@@ -0,0 +1,112 @@
+# Deviation Rules
+
+When execution doesn't go exactly as planned, these 5 rules govern what the executor can do on its own versus what requires user approval.
+
+---
+
+## Rule 1: Bug Discovered
+
+**Trigger**: Verify command fails due to a bug in the code just written.
+
+**Action**: Auto-fix the bug, re-run verify. If it passes, commit as normal. Note the bug and fix in SUMMARY.md under "Deviations".
+
+**Limit**: 3 fix attempts maximum. After 3 failures, STOP and return error with failing verify output.
+
+**Approval needed**: No
+
+---
+
+## Rule 2: Missing Dependency
+
+**Trigger**: Action or verify fails because a package/dependency is not installed.
+
+**Action**: Auto-install the dependency using the project's package manager. Continue execution.
+
+**Commit**: Include dependency installation in the task's commit (e.g., package.json, package-lock.json changes).
+
+**Limit**: Only install dependencies directly referenced in the plan's `<action>` or clearly required by the code being written. Do NOT install "nice-to-have" packages.
+
+**Approval needed**: No
+
+---
+
+## Rule 3: Critical Gap
+
+**Trigger**: While implementing, a critical gap is noticed that would cause runtime errors — missing error handling, null checks, input validation for security.
+
+**Action**: Add the minimal necessary code. Note it in SUMMARY.md under "Deviations".
+
+**Limit**: This is for preventing crashes and security holes ONLY. Not for adding features, improving UX, or refactoring.
+
+**Examples of valid critical gaps**:
+- Missing null check that would cause runtime crash
+- Unhandled promise rejection
+- SQL injection vulnerability
+- Missing input sanitization
+- Missing error boundary
+
+**Examples that are NOT critical gaps (use Rule 5 instead)**:
+- Better error messages
+- Input validation for UX (e.g., email format)
+- Performance optimization
+- Code organization improvements
+
+**Approval needed**: No
+
+---
+
+## Rule 4: Architectural Change
+
+**Trigger**: The plan's approach won't work as specified. The architecture needs to change. A fundamental assumption was wrong.
+
+**Action**: STOP immediately. Return a checkpoint-style response describing:
+- What the plan says to do
+- Why it won't work
+- What the alternative would be
+- Which tasks are affected
+
+**DO NOT** make architectural changes without approval. DO NOT try to "make it work" with hacks.
+
+**Examples**:
+- Plan says to use library X, but it doesn't support the required feature
+- Plan assumes a REST API, but the service only has GraphQL
+- Plan's database schema won't support the required queries
+- Plan's authentication approach conflicts with the framework
+
+**Approval needed**: YES — full stop, present to user
+
+---
+
+## Rule 5: Scope Creep
+
+**Trigger**: While implementing, something is noticed that would be nice to have, could be improved, or is related but not in the plan.
+
+**Action**: Log it in SUMMARY.md under "Deferred Ideas". DO NOT implement it. Continue with the plan as written.
+
+**Examples**:
+- "This could use caching" -> Deferred
+- "The error messages could be better" -> Deferred
+- "This should have pagination" -> Deferred
+- "The existing code has a bug unrelated to this plan" -> Deferred (unless it blocks the plan)
+- "This API should have rate limiting" -> Deferred
+- "The UI could use loading states" -> Deferred
+
+**Approval needed**: No (just log and continue)
+
+---
+
+## Decision Tree
+
+```
+Execution deviates from plan
+  |
+  ├─ Code bug? ──────────────── Rule 1: Auto-fix (3 attempts)
+  |
+  ├─ Missing package? ──────── Rule 2: Auto-install
+  |
+  ├─ Security/crash risk? ──── Rule 3: Minimal fix + note
+  |
+  ├─ Architecture wrong? ───── Rule 4: STOP + checkpoint
+  |
+  └─ Nice to have? ─────────── Rule 5: Log + continue
+```

package/plugins/copilot-pbr/references/git-integration.md

@@ -0,0 +1,284 @@
+# Git Integration Reference
+
+Plan-Build-Run's commit conventions, commit points, branching strategy, and hook scripts.
+
+---
+
+## Commit Message Format
+
+```
+{type}({scope}): {description}
+```
+
+### Components
+
+| Part | Description | Example |
+|------|-------------|---------|
+| `{type}` | Conventional commit type | `feat`, `fix`, `chore` |
+| `{scope}` | Descriptive word for the area of change | `auth`, `executor`, `changelog` |
+| `{description}` | Imperative, lowercase description | `add discord oauth flow` |
+
+The format is configurable via `config.json` at `git.commit_format`. The default is `{type}({scope}): {description}`.
+
+### Full Example
+```
+feat(auth): implement discord oauth client
+```
+
+---
+
+## Commit Types
+
+| Type | When to Use | Example |
+|------|------------|---------|
+| `feat` | New feature or functionality | `feat(auth): implement discord oauth client` |
+| `fix` | Bug fix (including during execution) | `fix(api): handle null user profile from discord` |
+| `refactor` | Code restructuring, no behavior change | `refactor(auth): extract token validation into helper` |
+| `test` | Adding or modifying tests | `test(auth): add failing tests for oauth flow` |
+| `docs` | Documentation changes | `docs(planning): add api endpoint documentation` |
+| `chore` | Build config, dependencies, tooling | `chore(deps): configure typescript and eslint` |
+| `style` | Formatting, whitespace (no logic change) | `style(auth): fix import ordering` |
+
+---
+
+## Commit Type Discipline
+
+| Type | Use For | Appears in Changelog |
+|------|---------|---------------------|
+| `feat` | User-visible features ONLY (new commands, new config options, changed behavior users notice) | YES |
+| `fix` | Bug fixes users would notice | YES |
+| `refactor` | Internal restructuring, TDD GREEN commits, code cleanup | No |
+| `test` | Test additions including TDD RED commits | No |
+| `docs` | Documentation (hidden — mostly planning artifacts) | No |
+| `chore` | Build config, deps, tooling, internal scaffolding | No |
+| `wip` | Work in progress (use sparingly) | No |
+
+**TDD commits**: RED phase → `test({scope}): RED - ...`, GREEN phase → `refactor({scope}): GREEN - ...`, REFACTOR phase → `refactor({scope}): REFACTOR - ...`
+
+---
+
+## Special Commit Scopes
+
+Beyond descriptive scopes, Plan-Build-Run recognizes these additional patterns:
+
+| Pattern | When Used | Example |
+|---------|-----------|---------|
+| `{type}(quick-{NNN})` | Quick-task commits via `/pbr:quick` | `feat(quick-001): add health endpoint` |
+| `docs(planning)` | Planning document commits | `docs(planning): add phase 3 plans` |
+| `wip: {desc}` or `wip({area}): {desc}` | Work-in-progress (use sparingly) | `wip(auth): partial oauth implementation` |
+| Merge commits | Git merge operations | `Merge branch 'plan-build-run/phase-02-auth'` |
+
+---
+
+## Session Affinity Metadata
+
+When a session ID is available, append a `[session:{short_id}]` suffix to the commit description. The `short_id` is the first 8 characters of the session_id. This enables conflict tracing across concurrent sessions.
+
+Example:
+```
+feat(auth): implement discord oauth client [session:a1b2c3d4]
+```
+
+When no session ID is available (e.g., manual commits), omit the suffix.
+
+---
+
+## Commit Body (Optional)
+
+For commits that need explanation, add a body after a blank line:
+
+```
+feat(02-01): implement discord oauth client [session:a1b2c3d4]
+
+- Uses discord-oauth2 library for token exchange
+- Stores tokens in httpOnly cookies for security
+- Supports identify and email OAuth scopes
+
+Deviation: Added null check for user.email (Rule 3 -- critical gap)
+```
+
+---
+
+## Commit Points
+
+Plan-Build-Run defines strict rules about WHEN commits happen:
+
+### One Task = One Commit
+
+Each successfully completed plan task gets exactly one atomic commit. No more, no less.
+
+### TDD Tasks = Three Commits
+
+TDD tasks (`tdd="true"`) produce exactly 3 commits following Red-Green-Refactor:
+
+```
+test(auth): RED - add failing tests for auth middleware
+refactor(auth): GREEN - implement auth middleware to pass tests
+refactor(auth): REFACTOR - extract token verification helper
+```
+
+### Commit Preconditions
+
+A commit is created only when:
+1. All `<action>` steps in the task are complete
+2. All `<verify>` commands pass with exit code 0
+3. Only files listed in the task's `<files>` element are staged (plus deviation-required files)
+
+### Deviation Commits
+
+When an executor applies a deviation rule during a task, the deviation is included in the same task commit (not a separate commit):
+
+```
+# Rule 1 auto-fix: rolled into the task's commit
+# Rule 2 auto-install: package.json/lock added to same commit
+# Rule 3 critical gap: null check added in same commit
+```
+
+---
+
+## Atomic Commit Rules
+
+1. **One task = one commit** (TDD tasks get 3)
+2. **Commit only after verify passes** -- never commit broken code
+3. **Stage only files listed in `<files>`** plus any deviation-required files
+4. **Never use `git add .` or `git add -A`** -- stage specific files
+5. **Include all files from the task** -- do not leave modified files unstaged
+6. **Commit message must describe what was done** -- not just the task name
+7. **Use the configured commit format** -- from `config.json` `git.commit_format`
+
+---
+
+## Staging Examples
+
+```bash
+# Standard task
+git add src/auth/discord.ts src/auth/types.ts
+git commit -m "feat(auth): implement Discord OAuth client with token exchange"
+
+# Task that also installed a dependency (Rule 2)
+git add src/auth/discord.ts src/auth/types.ts package.json package-lock.json
+git commit -m "feat(auth): implement Discord OAuth client with token exchange"
+
+# TDD RED
+git add tests/auth/discord.test.ts
+git commit -m "test(auth): add failing tests for Discord OAuth flow"
+```
+
+---
+
+## Git Retry Logic
+
+If `git commit` fails with a lock error (`fatal: Unable to create ... .git/index.lock`):
+1. Wait 2 seconds
+2. Retry the commit
+3. Maximum 3 attempts
+4. If still failing after 3 attempts, report the error and stop
+
+This commonly occurs when parallel executors compete for the git index lock.
+
+---
+
+## Branching Strategy
+
+Configured via `config.json` at `git.branching`:
+
+| Strategy | Behavior |
+|----------|----------|
+| `none` | All commits go directly to the current branch (default) |
+| `phase` | Each phase gets its own branch: `plan-build-run/phase-{NN}-{slug}`. Squash-merged to main on completion. |
+| `milestone` | Each milestone gets a branch: `plan-build-run/{milestone}-{slug}`. Merged on milestone completion. |
+| `disabled` | No git operations at all. No commits, no branching. Useful for prototyping or non-git projects. |
+
+### Phase Branching Flow
+
+When `git.branching` is `phase`:
+1. Build orchestrator creates branch: `git checkout -b plan-build-run/phase-{NN}-{slug}`
+2. All executor commits go to the phase branch
+3. After all plans complete and verification passes, orchestrator asks user to confirm squash merge
+4. If confirmed: `git checkout main && git merge --squash plan-build-run/phase-{NN}-{slug}`
+5. Phase branch is deleted after merge
+
+Branch name templates are configured in `config.json`:
+- `git.phase_branch_template`: Default `plan-build-run/phase-{phase}-{slug}`
+- `git.milestone_branch_template`: Default `plan-build-run/{milestone}-{slug}`
+
+### PR Creation
+
+When `git.auto_pr: true` and `git.branching` is `phase` or `milestone`, the build skill creates a GitHub PR after verification passes:
+
+1. Push the phase branch to the remote
+2. Create a PR via `gh pr create` with structured title and body
+3. PR title follows the commit convention: `feat(phase-{N}): {phase slug}`
+4. PR body includes phase goal, key files changed, and must-have verification results
+
+When `git.auto_pr: false` (default), the build skill offers the user a choice after verification:
+- Create PR now
+- Skip PR creation
+- Push branch only (create PR later)
+
+PR creation requires `gh` CLI authenticated with repo write access.
+
+### CI Integration
+
+When `ci.gate_enabled: true`, the build skill checks GitHub Actions status after each wave completes:
+
+1. Run: `gh run list --branch $(git branch --show-current) --limit 1 --json status,conclusion,url`
+2. If `status == completed` and `conclusion == success`: proceed to next wave
+3. If `status == in_progress`: wait up to `ci.wait_timeout_seconds`, re-check
+4. If `conclusion != success` or timeout: surface warning with run URL and offer:
+   - **Wait**: continue polling
+   - **Continue anyway**: proceed despite CI failure (logged as deviation)
+   - **Abort**: stop the build
+
+CI gate requires `gh` CLI and GitHub Actions configured on the repository. The gate only activates when there are commits pushed to a remote branch (branching must be enabled).
+
+### Git Mode
+
+The `git.mode` field controls whether git integration is active:
+
+| Mode | Behavior |
+|------|----------|
+| `enabled` | Normal operation -- commits, branching, hooks all active (default) |
+| `disabled` | No git commands are run. No commits, no branching. Useful for non-git projects. |
+
+---
+
+## Hook Scripts
+
+Plan-Build-Run uses Claude Code hooks (defined in `hooks/hooks.json`) to enforce conventions and track progress during execution. All hooks log via `hook-logger.js` to `.planning/.hook-log` (JSONL format, max 200 entries).
+
+### Hook Summary
+
+| Hook Event | Script | Purpose |
+|------------|--------|---------|
+| `SessionStart` | `progress-tracker.js` | Displays current project progress on session start |
+| `PostToolUse` (Write/Edit) | `check-plan-format.js` | Validates plan file YAML frontmatter after writes |
+| `PostToolUse` (Write/Edit) | `check-roadmap-sync.js` | Ensures ROADMAP.md stays in sync with phase changes |
+| `PreToolUse` (Bash) | `validate-commit.js` | Validates commit message format before `git commit` runs |
+| `PreCompact` | `context-budget-check.js` | Warns about context budget before compaction |
+| `Stop` | `auto-continue.js` | Handles auto-continuation signal when session ends |
+| `SubagentStart` | `log-subagent.js start` | Logs subagent spawn events |
+| `SubagentStop` | `log-subagent.js stop` | Logs subagent completion events |
+| `SessionEnd` | `session-cleanup.js` | Cleans up temporary state on session end |
+
+### validate-commit.js Details
+
+The commit validation hook (`PreToolUse` on Bash commands) enforces the commit format:
+- Checks that `git commit -m "..."` messages match the pattern: `{type}({scope}): {description}`
+- Valid types: `feat`, `fix`, `refactor`, `test`, `docs`, `chore`, `wip`
+- Allows merge commits (starting with "Merge")
+- Allows quick-task, planning, and WIP scope patterns
+- Blocks commits with sensitive file patterns (`.env`, `.key`, `.pem`, credentials) unless they match safe patterns (`.example`, `.template`, `.sample`, test directories)
+- Exit code 2 blocks the commit; exit code 0 allows it
+
+---
+
+## Planning Doc Commits
+
+When `config.json` has `planning.commit_docs: true`, the build orchestrator commits planning artifacts (SUMMARY.md, VERIFICATION.md) after all plans in a phase complete:
+
+```
+docs({phase}): add build summaries and verification
+```
+
+This keeps the planning trail in version control alongside the code it describes.