agent-skill-kit 3.9.135

package/.agent/workflows/knowledge.md

@@ -0,0 +1,212 @@
+---
+description: LLM-maintained knowledge wiki — ingest raw signals from fixes and corrections, compile them into cross-linked concept articles, query project knowledge, and run health checks.
+chain: knowledge-management
+skills: [knowledge-compiler, knowledge-linter, skill-generator]
+agents: [orchestrator, learner]
+---
+
+# /knowledge - The Knowledge Engine
+
+$ARGUMENTS
+
+---
+
+## Purpose
+
+Manage the project's living knowledge wiki. Raw signals from error fixes, user corrections, and architectural decisions are ingested, compiled into cross-linked concept articles by the LLM, and queryable for future coding decisions. Patterns are cached in `knowledge/patterns/` for fast agent lookup.
+
+---
+
+## Sub-Commands
+
+| Command | Purpose | Example |
+|---------|---------|---------|
+| `/knowledge ingest` | Record a raw signal | `/knowledge ingest "CDP restart needs bat file on Windows"` |
+| `/knowledge compile` | Compile unprocessed signals into articles | `/knowledge compile` |
+| `/knowledge query` | Ask a question against the wiki | `/knowledge query "how does CDP restart work?"` |
+| `/knowledge lint` | Run health checks on the wiki | `/knowledge lint` |
+| `/knowledge status` | Show wiki statistics | `/knowledge status` |
+
+---
+
+## 🤖 Meta-Agents Integration
+
+| Phase | Agent | Action |
+|-------|-------|--------|
+| **Ingest** | `learner` | Extract signal from recent fix/correction |
+| **Compile** | `orchestrator` | Batch signals, route to compiler |
+| **Lint** | `learner` | Generate health report |
+
+---
+
+## 🔴 MANDATORY: Knowledge Protocol
+
+### Phase 0: Pre-flight & Auto-Learned Context
+
+> **Rule 0.5-K:** Auto-learned pattern check.
+
+1. Read `.agent/skills/auto-learned/patterns/` for past failures before proceeding.
+2. Trigger `recovery` agent to run Checkpoint (`git commit -m "chore(checkpoint): pre-knowledge"`).
+
+### Phase 1: Ingest
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Description of what happened + context |
+| **OUTPUT** | Raw signal file in `.agent/knowledge/raw/` |
+| **SKILLS** | `knowledge-compiler` |
+
+1. Classify signal type: `error_fix` | `correction` | `decision` | `observation`
+2. Extract structured data: signal, context, resolution, lesson
+3. Write to `.agent/knowledge/raw/{YYYY-MM-DD}-{slug}.md` with frontmatter
+4. Update `_index.md` uncompiled count
+5. Output: `📥 Signal recorded: {slug}`
+
+### Phase 2: Compile
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Uncompiled signals in `raw/` (max 10 per batch) |
+| **OUTPUT** | New/updated concept articles in `concepts/` |
+| **SKILLS** | `knowledge-compiler` |
+
+1. Read `_index.md` for current wiki state
+2. Find signals where `compiled: false` (max 10)
+3. Cluster related signals by tags
+4. For each cluster:
+   - Existing article? �’ Update with new insights
+   - New topic? �’ Create concept article
+   - Decision? �’ Create ADR in `decisions/`
+5. Add `[[backlinks]]` between related articles
+6. Mark processed signals as `compiled: true`
+7. Regenerate `_index.md` and `_graph.md`
+8. Output: `📚 Compiled: {N} signals �’ {M} articles`
+
+### Phase 3: Query
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Natural language question |
+| **OUTPUT** | Answer citing wiki articles |
+| **SKILLS** | `knowledge-compiler` |
+
+1. Read `_index.md` for article summaries
+2. Identify relevant articles by topic
+3. Read full article(s)
+4. Synthesize answer with citations
+5. Output: Answer with `[[article-name]]` references
+
+### Phase 4: Lint
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Current wiki state |
+| **OUTPUT** | Health report with score and recommendations |
+| **SKILLS** | `knowledge-linter` |
+
+1. Run 5 health checks: staleness, orphans, inconsistency, gaps, dead links
+2. Calculate health score (0-100)
+3. Generate recommendations sorted by priority
+4. Output: Health report markdown
+
+### Phase 5: Status
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | None |
+| **OUTPUT** | Wiki statistics summary |
+| **SKILLS** | `knowledge-compiler` |
+
+1. Read `_index.md` statistics
+2. Count files in `raw/`, `concepts/`, `decisions/`
+3. Output: Quick statistics table
+
+---
+
+## Auto-Ingest Rules
+
+> When to automatically record signals (no explicit `/knowledge ingest` needed):
+
+| Trigger | Signal Type | Condition |
+|---------|-------------|-----------|
+| Multi-file fix | `error_fix` | Fix touched ≥ 3 files |
+| User correction | `correction` | User says "wrong"/"sai"/"fix this" |
+| Architecture decision | `decision` | ADR-worthy choice made during `/build` or `/plan` |
+| Repeated error | `error_fix` | Same error class hit ≥ 2 times in session |
+
+> Single-line fixes, typos, and formatting changes are NOT ingested.
+
+---
+
+## Output Format
+
+```markdown
+## 📚 Knowledge: {Operation}
+
+### Result
+
+| Metric | Value |
+|--------|-------|
+| Operation | {ingest/compile/query/lint/status} |
+| Articles affected | {N} |
+| Signals processed | {N} |
+
+### Details
+
+{Operation-specific output}
+
+### Next Steps
+
+- [ ] {Recommended follow-up action}
+```
+
+---
+
+## Examples
+
+```
+/knowledge ingest "CDP restart on Windows needs .bat intermediary because child process dies with parent"
+/knowledge compile
+/knowledge query "what are the known gotchas with process spawning?"
+/knowledge lint
+/knowledge status
+```
+
+---
+
+## �— Workflow Chain
+
+**Skills Loaded (3):**
+
+- `knowledge-compiler` — Signal ingestion, pattern learning, article compilation
+- `knowledge-linter` — Wiki health checks
+- `skill-generator` — Generate skills from mature articles
+
+```mermaid
+graph LR
+    A["/fix or /cook"] --> B["/knowledge ingest"]
+    B --> C["/knowledge compile"]
+    C --> D["/knowledge lint"]
+    D --> E["skill-generator"]
+    style C fill:#10b981
+```
+
+| After /knowledge | Run | Purpose |
+|-----------------|-----|---------| 
+| Compile done | `/knowledge lint` | Verify wiki health |
+| Health issues | `/knowledge compile` | Recompile stale articles |
+| Mature knowledge | `skill-generator` | Generate skills from wiki |
+
+---
+
+## Key Principles
+
+- **Signals compound** — every fix makes the wiki smarter
+- **LLM maintains the wiki** — humans rarely edit articles directly
+- **Read before code** — agent checks `_index.md` before implementation
+- **Batch discipline** — max 10 signals per compile to control token budget
+- **Cross-link everything** — isolated knowledge is wasted knowledge
+
+---
+
+⚡ PikaKit v3.9.134

package/.agent/workflows/launch.md

@@ -0,0 +1,345 @@
+---
+description: Zero-downtime production release pipeline — pre-flight security gates, automated build verification, health-check monitoring, and instant rollback on failure.
+chain: deploy-production
+skills: [cicd-pipeline, server-ops, security-scanner, code-review, problem-checker, context-engineering, knowledge-compiler]
+agents: [orchestrator, assessor, recovery, learner, security-auditor, devops-engineer, test-engineer]
+---
+
+# /launch - Zero-Downtime Release
+
+$ARGUMENTS
+
+---
+
+## Purpose
+
+Production deployment with automated pre-flight checks, security scanning, build verification, health monitoring, and auto-rollback capability. **Differs from `/stage` (local development sandbox) and `/monitor` (production observability) by executing the full deployment pipeline with zero-downtime guarantees.** Uses `cicd-pipeline` with `cicd-pipeline` for deployment orchestration, `security-scanner` with `security-scanner` for pre-deploy validation, and `recovery` for rollback safety.
+
+---
+
+## Sub-Commands
+
+| Command | Action |
+|---------|--------|
+| `/launch` | Interactive deployment wizard |
+| `/launch check` | Run pre-flight checks only |
+| `/launch preview` | Deploy to staging/preview |
+| `/launch ship` | Deploy to production |
+| `/launch rollback` | Rollback to previous version |
+| `/launch rollback --to v2.0.3` | Rollback to specific version |
+| `/launch rollback --list` | List available rollback points |
+
+---
+
+## ?? Meta-Agents Integration
+
+| Phase | Agent | Action |
+| ----- | ----- | ------ |
+| **Pre-Flight** | `assessor` | Evaluate deployment risk and knowledge-compiler context |
+| **Execution** | `orchestrator` | Coordinate deploy pipeline and health checks |
+| **Safety** | `recovery` | Save state and recover/rollback from failed deployments |
+| **Post-Launch** | `learner` | Log deployment telemetry and success/failure patterns |
+
+```
+Flow:
+assessor.evaluate(risk) ? recovery.save(checkpoint)
+       ?
+pre-flight gates ? build ? deploy
+       ?
+health check ? pass? ? learner.log(success)
+       ? fail
+recovery.restore(checkpoint) ? learner.log(failure)
+```
+
+---
+
+## ?? MANDATORY: Deployment Protocol
+
+### Phase 1: Pre-flight & knowledge-compiler Context
+
+> **Rule 0.5-K:** knowledge-compiler pattern check.
+
+1. Read `.agent/skills/knowledge-compiler/patterns/` for past failures before proceeding.
+2. Trigger `recovery` agent to run Checkpoint (`git commit -m "chore(checkpoint): pre-launch"`).
+
+### Phase 2: Pre-Flight Gates
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | $ARGUMENTS (sub-command + optional flags) |
+| **OUTPUT** | Pre-flight result: all gates pass/fail with details |
+| **AGENTS** | `security-scanner`, `assessor` |
+| **SKILLS** | `security-scanner`, `code-review`, `context-engineering` |
+
+**Gate 1: Code Quality**
+
+// turbo — telemetry: phase-2-preflight
+```bash
+npx cross-env OTEL_SERVICE_NAME="workflow:launch" TRACE_ID="$TRACE_ID" npx tsc --noEmit; npx cross-env OTEL_SERVICE_NAME="workflow:launch" TRACE_ID="$TRACE_ID" npm run lint; npx cross-env OTEL_SERVICE_NAME="workflow:launch" TRACE_ID="$TRACE_ID" npm test
+```
+
+| Check | Required |
+|-------|----------|
+| TypeScript compiles | ? |
+| Linting passes | ? |
+| All tests pass | ? |
+| Build succeeds | ? |
+
+**Gate 2: Security**
+
+| Check | Required |
+|-------|----------|
+| No hardcoded secrets | ? |
+| Dependencies audited | ? |
+| HTTPS configured | ? |
+| Environment variables set | ? |
+
+**Gate 3: Performance**
+
+| Check | Threshold |
+|-------|-----------|
+| Lighthouse Score | > 80 |
+| Bundle Size | < 500KB |
+| First Contentful Paint | < 2s |
+
+> If ANY gate fails ? STOP ? report blockers ? do not deploy.
+
+### Phase 3: Build & Checkpoint
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | All gates passed from Phase 2 |
+| **OUTPUT** | Production build + git checkpoint tag |
+| **AGENTS** | `cicd-pipeline`, `orchestrator` |
+| **SKILLS** | `cicd-pipeline` |
+
+1. `recovery` saves current state (git tag + commit hash)
+2. Build production bundle
+
+// turbo — telemetry: phase-3-build
+```bash
+npx cross-env OTEL_SERVICE_NAME="workflow:launch" TRACE_ID="$TRACE_ID" npm run build
+```
+
+3. Auto-detect platform:
+
+| Platform | Command | Auto-Detected By |
+|----------|---------|-----------------|
+| Vercel | `vercel --prod` | Next.js, Vite |
+| Railway | `railway up` | Dockerfile |
+| Fly.io | `fly deploy` | fly.toml |
+| Netlify | `netlify deploy --prod` | Static sites |
+| AWS | `sam deploy` | SAM template |
+
+### Phase 4: Deploy & Health Check
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Production build + platform config from Phase 3 |
+| **OUTPUT** | Deployed application with health check result |
+| **AGENTS** | `cicd-pipeline` |
+| **SKILLS** | `cicd-pipeline`, `server-ops` |
+
+1. Execute platform-specific deploy command
+// turbo — telemetry: phase-4-deploy
+2. Run health check (60s timeout):
+
+| Check | Endpoint | Expected |
+|-------|----------|----------|
+| API Status | `GET /api/health` | 200 OK |
+| Database | `GET /api/health/db` | 200 OK |
+| Response Time | — | < 2000ms |
+
+3. If health check fails ? Phase 5 (Auto-Rollback)
+4. If health check passes ? `learner` logs deploy success
+
+### Phase 5: Auto-Rollback (on failure only)
+
+| Field | Value |
+|-------|-------|
+| **INPUT** | Failed health check from Phase 4 |
+| **OUTPUT** | Reverted to previous version, rollback report |
+| **AGENTS** | `cicd-pipeline`, `learner` |
+| **SKILLS** | `cicd-pipeline`, `knowledge-compiler`, `problem-checker` |
+
+1. `recovery` restores checkpoint
+2. Re-deploy previous version
+3. Verify rollback health
+4. `learner` logs failure pattern
+
+```mermaid
+graph TD
+    A["/launch"] --> B{"Pre-flight gates"}
+    B -->|Fail| C["Fix issues"]
+    C --> B
+    B -->|Pass| D["Build + checkpoint"]
+    D --> E["Deploy"]
+    E --> F{"Health check"}
+    F -->|Pass| G["? Live"]
+    F -->|Fail| H["Auto-rollback"]
+    H --> I["Alert team"]
+```
+
+---
+
+## ? MANDATORY: Problem Verification Before Completion
+
+> **CRITICAL:** This check MUST be performed before any `notify_user` or task completion.
+
+### Check @[current_problems]
+
+```
+1. Read @[current_problems] from IDE
+2. If errors/warnings > 0:
+   a. Auto-fix: imports, types, lint errors
+   b. Re-check @[current_problems]
+   c. If still > 0 ? STOP ? Do NOT deploy
+3. If count = 0 ? Proceed to deployment
+```
+
+### Auto-Fixable
+
+| Type | Fix |
+|------|-----|
+| Missing import | Add import statement |
+| Unused variable | Remove or prefix `_` |
+| Type mismatch | Fix type annotation |
+| Lint errors | Run eslint --fix |
+
+> **Rule:** Never deploy with errors in `@[current_problems]`.
+
+---
+
+## ?? Rollback & Recovery
+
+If deployment automation fails completely or pre-flights break the workspace:
+1. Fallback to manual deploy: trigger `recovery.restore()` and halt workflow.
+2. Auto-rollback handles post-deployment health check failures (see Phase 5).
+3. Log incident via `learner` meta-agent.
+
+---
+
+## Output Format
+
+### Successful Launch
+
+```markdown
+## ?? Launch Complete
+
+### Summary
+
+| Metric | Value |
+|--------|-------|
+| Version | v2.1.0 |
+| Environment | Production |
+| Duration | 47 seconds |
+| Platform | Vercel |
+
+### Pre-Flight Results
+
+| Gate | Status |
+|------|--------|
+| TypeScript | ? 0 errors |
+| Tests | ? 42/42 passed |
+| Security | ? No vulnerabilities |
+| Build | ? Success (234KB) |
+
+### Health Check
+
+| Check | Status |
+|-------|--------|
+| API | ? 200 OK |
+| Database | ? Connected |
+| Response | ? < 2000ms |
+
+### URLs
+
+?? **Production:** https://app.example.com
+
+### Next Steps
+
+- [ ] Monitor for 24h with `/monitor`
+- [ ] Run `/chronicle` to update documentation
+- [ ] Rollback available: `/launch rollback`
+```
+
+### Failed Launch
+
+```markdown
+## ? Launch Aborted
+
+### Pre-Flight Failed
+
+| Check | Result |
+|-------|--------|
+| TypeScript | ? 3 errors |
+| Tests | ? Pass |
+| Security | ?? 1 warning |
+
+### Blockers
+
+1. TypeScript error in `src/api/user.ts:42`
+
+### Next Steps
+
+- [ ] Fix blockers with `/fix`
+- [ ] Re-run `/launch` after fixes
+```
+
+---
+
+## Examples
+
+```
+/launch
+/launch check
+/launch preview
+/launch ship
+/launch rollback
+/launch rollback --to v2.0.3
+```
+
+---
+
+## Key Principles
+
+- **Never skip security** — always scan for vulnerabilities before deploy
+- **Rollback-first** — checkpoint before deploy, auto-revert on failure
+- **Health check** — verify after deploy with 60s timeout
+- **Zero-downtime** — no user-facing interruption during deployment
+- **Document changes** — update changelog and notify team automatically
+
+---
+
+## ?? Workflow Chain
+
+**Skills Loaded (7):**
+
+- `cicd-pipeline` - Deployment orchestration and rollback strategies
+- `server-ops` - Server management and health monitoring
+- `security-scanner` - Pre-deploy vulnerability scanning
+- `code-review` - Code quality validation
+- `context-engineering` - Codebase parsing and component mapping
+- `problem-checker` - IDE problem verification
+- `knowledge-compiler` - Auto-learning from deployment failures
+
+```mermaid
+graph LR
+    A["/validate"] --> B["/launch"]
+    B --> C["/monitor"]
+    style B fill:#10b981
+```
+
+| After /launch | Run | Purpose |
+|--------------|-----|---------|
+| Deploy success | `/chronicle` | Generate updated documentation |
+| Deploy success | `/monitor` | Monitor production health |
+| Deploy fail | `/diagnose` | Debug deployment issue |
+| Need revert | `/launch rollback` | Rollback to previous version |
+
+**Handoff to /monitor:**
+
+```markdown
+?? Deployed v[version] to production! URL: [url].
+Health check: ?. Run `/monitor` to track or `/chronicle` for docs.
+```