Agent Platform Bootstrap v2.42.0

Agent Platform Bootstrap v2.42.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.42.0

Added

• Install profiles — --profile=lite|core|full; lite = skills pack without handoff/enterprise layer; profile-filter.mjs + AGENTS-lite.md + QUICK-REF-lite.md
• Cherry-pick install — --mode=add --add=skill:id and --mode=list --list=skills
• 11 lifecycle skills — modular SKILL.md pack including context-engineering (/context) and verification-before-completion (/verify) from Mode 4 R001/R005
• Lifecycle commands — /plan /build /test /code-simplify /webperf /context /verify (Cursor + Claude)
• Claude marketplace plugin — .claude-plugin/plugin.json for agent-platform-skills
• docs/cursor-setup.md — Cursor distribution guide (no marketplace; --profile=lite)
• Unified scan registry — MAINTAINER/scan-results/registry.md + REPORT-SCHEMA.md; Mode 4 governance scan archive (2026-06-09)
• Platform Sync Gate (PSG) — mandatory maintainer auto-sync after Mode 1–4 changes; platform-maintainer-sync.mdc

Changed

• /spec routes to interview-me skill (idea-refine when exploratory)
• QUICK-REF — profiles matrix, lifecycle when/how table, skills catalog, common confusions (/test vs /verify, etc.)
• README Install — profile recommendation table + capability matrix + lite→full upgrade path
• Mode 2 web audit (2026-06) — OWASP 2025 supply chain, fail-closed errors, API auth batching, LLM action screening in security/backend/devops agents

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform

Agent Platform Bootstrap v2.41.0

Agent Platform Bootstrap v2.41.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.41.0

Added

• Agent-skills ingest (P0 + P1) — rationalization gates, doubt review, source-driven dev, TDD pyramid/Beyoncé/DAMP, Hyrum's Law, Chesterton's Fence, CWV measure-first; attribution in MAINTAINER/ingest/agent-skills-p0-SOURCES.md
• 2 new playbooks (20 total): requirements-clarification, deprecation
• Reference library — .agent/references/ (testing, security, performance, accessibility, orchestration-patterns)
• Context template — spec-outline.md · add-feature Step 0 (spec clarity)
• Lifecycle slash commands — Claude Code (.claude/commands/) and Cursor (.cursor/commands/): /session-start, /session-end, /spec, /audit, /review, /release, /ship, /quick-ref, /platform-help, /caveman (+ helpers); Cursor-only /implement for Plan mode handoff
• Cursor Plan mode handoff — plan-mode-handoff.mdc resumes add-feature from Step 3 after plan approval
• User CONTRIBUTING.md — how to submit rules and playbook ideas
• docs/DISTRIBUTION.md — install paths, IDE notes, honest comparison vs skill packs

Changed

• Marketing / discovery — FRAMEWORK-README "When to use what" table; QUICK-REF key-principle column; PLATFORM-HELP "Start here" path; README named-concepts section; presentation decks + STORY-PLAN aligned with v2.41 lifecycle, slash commands, and Plan handoff
• Install banner — dynamic playbook count from manifest (apply.js)
• User-facing docs — playbook counts, agent-skills DNA callouts, slash command list

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform

Agent Platform Bootstrap v2.40.0

Agent Platform Bootstrap v2.40.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.40.0

Added

• 8 new playbooks (18 total): nfr-definition, production-readiness, performance-budget, observability-setup, accessibility-audit, compliance-review, org-maturity-assessment, incident-postmortem
• context/nfr-log.md — ISO 25010 / 14-category NFR register (threshold + measure + verify)
• context/compliance-evidence-log.md — SOC 2 / ISO 27001 control → artifact mapping
• context/incident-log.md — incident register and DORA rollup table
• Expanded security-audit playbook — structured OWASP/CVE/secrets pass with mandatory Critic (replaces stub)
• Full project audit — Phases 4b/5b/8b (performance, WCAG, observability) + Phase 10 governance/compliance/maturity; report is Phase 11
• Critic dimensions — [ACCESSIBILITY], [OPERABILITY], [BC] and expanded review scope
• WCAG 2.2 AA — frontend-agent accessibility baseline
• UX interaction principles — Nielsen + Shneiderman golden rules in frontend-agent
• Maintainer command — "Sync user-facing docs for vX.Y.Z" and 18-playbook inventory in maintainer agent

Changed

• production-readiness — P0 NFR verification, compliance evidence, vuln SLA, SBOM, change-management gates
• nfr-definition — DORA KPI elicitation and compliance evidence cross-links
• nfr-log.md — example rows for compliance (NFR-C01–C02) and DORA (NFR-DP01–DP04)
• data-agent — N+1 detection, indexes, bounded reads, EXPLAIN discipline
• devops-agent — container image vulnerability scan; observability operability section
• add-feature Step 5b — [PERFORMANCE] / [ACCESSIBILITY] when triggered; checks nfr-log.md P0/P1
• Architect / Security agents — NFR, compliance-evidence, and incident-log discipline
• AGENTS.md routing — NFR, PRR, performance budget, observability, a11y, compliance, maturity, postmortem
• CHECKLIST — NFR, a11y, observability, compliance evidence, vuln SLA, change traceability
• User-facing docs — README, FRAMEWORK-README, QUICK-REF, PLATFORM-HELP, agent-platform-beta.html, STORY-PLAN

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform

Agent Platform Bootstrap v2.37.0

Agent Platform Bootstrap v2.37.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.37.0

Added

• document-api playbook — OpenAPI/Swagger tasks now route to .agent/playbooks/document-api.md instead of *(none)*; enforces spec-from-code alignment and mandatory Critic review before handoff
• Session-end Step 2a — Critic catch-up gate when application code changed but playbook Step 5b was skipped
• Session-end Step 2e — mandatory test + coverage verification when application code changed
• CHECKLIST quality gates — Critic, Security gate, and playbook Step 5b output tracked explicitly

Fixed

• Critic gate skipped on bug-fix / add-feature — Step 5b marked MANDATORY with required ▶ Critic review — output and CURRENT.md logging; HARD RULE blocks marking work done without review
• PowerShell commit failure at session-end — Step 2c forbids && chaining; each git command runs as a separate shell invocation
• Docs route without playbook — AGENTS.md docs row now points to document-api.md (prevents spec drift / docs-before-code)

Changed

• add-feature Step 5a — Security gate outputs required ▶ Security gate — signal when triggered

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform

v2.36.0

What's new in v2.36.0

Added

• Backwards compatibility policy — every expert agent and playbook now checks for BC breaks before writing any code. A structured ⚠️ BC BREAK notice (what changes, who is affected, severity, migration path) is required and blocks implementation until explicitly approved. Defined once in BEST-PRACTICES.md; enforced across all 9 experts, the Critic [BC] dimension, and 5 playbooks (add-feature, bug-fix, api-integration, release, refactor).
• Critic [BC] review dimension — 8th dimension added to the adversarial review. Checks API contract changes, field removals, env/config renames, exported interface changes, auth mechanism changes, and whether the BC notice was issued and approved. Non-migratable break = Critical; undocumented migratable break = High.
• Team adoption presentation — presentation/team-adoption.html (12-slide deck) and presentation/STORY-PLAN.md (story beats, live demo script, objection handling) for onboarding developer teams.

Changed

• Release playbook BC scan — Step 3b now scans all commits since last tag for BC breaks before semver determination. Any BC break forces a Major bump; blocks if bump is downgraded without explicit approval.
• Cursor multi-model documentation — .cursor/README.md and SYNC-POINTS.md now document that model switches within Cursor (GPT-4o → Claude → Gemini) are invisible to the platform and should be treated as mini-handoffs: run session-end before switching models mid-task.

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your PROJECT sections are never touched. Platform rules update automatically.

v2.35.2 — Session-start no longer triggers expert routing

Fixed

• ▶ fires on session-start command — the router was treating Read .agent/session-start.md and execute it. as a dev task and incorrectly attaching a ▶ Agent Platform · backend expert · add-feature playbook routing signal. Session commands now explicitly bypass routing in both AGENTS.md and the global ~/.claude/CLAUDE.md stub.

Upgrade

Your global stub is already patched in place (no action needed for ~/.claude/CLAUDE.md).

To update AGENTS.md in your consumer repo:
```
npx github:zafrirron/Agent-Platform --mode=upgrade
```

v2.35.1 — Fix ▶ routing signal silenced by global stub

Fixed

• ▶ routing signal silenced by global stub — the global ~/.claude/CLAUDE.md template said "silently load the correct expert/playbook" which instructed the model to suppress all routing output, overriding every project-level ▶ exemption. Removed the word "silently" and added the exact ▶ format with an explicit platform-signal exemption from caveman mode/compression.
• Project CLAUDE.md: ▶ format now inline — the ▶ format is now embedded directly in CLAUDE.md so no prior AGENTS.md read is required before outputting the first response line.

Upgrade

In your consumer repo:
```
npx github:zafrirron/Agent-Platform --mode=upgrade
```

Also run the global stub update (fixes the root cause in your personal Claude config):
```
npx github:zafrirron/Agent-Platform --mode=global
```

Agent Platform Bootstrap v2.35.0

What's new in v2.35.0

Added

• Agent Platform branding on routing status lines — every expert/playbook activation now shows ▶ Agent Platform · [Expert] · [Playbook] so users can see the platform working
• Release playbook — DevOps owns changelog + version bump — Step 3 expanded with four concrete sub-steps: collect commits since last tag, determine semver bump level, write CHANGELOG.md entry, bump version in all relevant files (DevOps agent, not docs-agent)
• Release playbook — full release commit + tag + GitHub release page — Steps 6–8 added: chore(release): vX.Y.Z commit, git tag, push --tags, gh release create using the CHANGELOG entry verbatim, and announce via release page URL
• CHANGELOG.md starter template ships on install — professional Keep a Changelog template created automatically for new projects; includes [Unreleased] section, full example entry, authoring rules, semver guide, and comparison link stubs; existing changelogs are never touched
• Changelog retrofit capability — DevOps agent can convert any existing changelog to Keep a Changelog format; 9-step workflow: audit → confirm plan → map entries → drop internal noise → show diff → write on approval; trigger: "retrofit my changelog"
• Changelog management documented — new section in user-facing README covering install-time creation, release-time authoring, and the retrofit command

Fixed

• Routing status line suppressed — session-start-shared.md contained "Never announce what you are reading" which silenced the ▶ status line entirely; replaced with a precise rule that preserves the status line while still preventing file-read narration
• Session-start: same-framework auto-resume — sessions on the same framework now resume silently without re-prompting the user
• Mode 3 ingest clarified — ingest is maintainer-curated, not user self-serve; audit now verifies docs and presentation are in sync
• README: maintainer-only commands removed — Mode 4 / ingest commands removed from user-facing upgrade section
• release.ps1 BOM bug — Set-Content -Encoding UTF8 wrote a BOM that broke JSON.parse; switched to [System.IO.File]::WriteAllText with explicit no-BOM UTF-8

Documented

• All 4 Jest coverage output formats (HTML, Clover XML, LCOV, JSON summary) documented with glossary entries

Agent Platform Bootstrap v2.34.0

Agent Platform Bootstrap v2.34.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.34.0

Added — Honest limitation notice · Glossary · Responsive presentation · Uninstall restore visibility · Antigravity rules scan

Honest limitation notice (3 places):

• apply.js install output: new "⚠ Rules are guidance — not deterministic enforcement" section with call-to-action for --mode=install-guards
• .agent/PLATFORM-HELP.md: new "⚠ Important" section at top explaining probabilistic vs deterministic enforcement
• presentation: amber callout on "What It Is" slide

Uninstall restore visibility:

• Restore messages now have their own clearly separated section with ────── dividers and ✅ Restored: X ← your original file is back formatting
• Previously buried between ✔ Removed: lines; users were missing that their files were restored

PLATFORM-HELP.md discoverability:

• Status block Reference line now shows say "platform help" for full guide — users see it every session instead of only after opening QUICK-REF.md

Glossary with links (README.md):

• New "Glossary — standards and abbreviations" section: 14 terms (OWASP, CWE, CVE, SOLID, DRY, ADR, OpenAPI/Swagger, OIDC, SBOM, CSRF, SSRF, N+1, CQRS) all with clickable links to authoritative sources

Responsive presentation:

• 3-tier CSS breakpoints (960px tablet, 640px mobile, 400px small mobile)
• Grids, fonts, padding, arrows all adapt correctly to different screen sizes

New: Standards Glossary slide (presentation):

• Slide 17: 10 terms explained and linked — OWASP, LLM Top 10, CWE, SOLID, ADR, OpenAPI, OIDC, SBOM, CVE/CSRF/SSRF, DRY/N+1/CQRS

Presentation trimmed (19 → 17 slides):

• Removed: Full Platform Lifecycle Diagram (overlapped How It Works), Caveman Mode (mentioned in Quality Gates), Platform Evolution (rewritten as "Living Platform" user value story)
• SOLID: now shown as badge in Quality Gates slide — named but not taught

Antigravity .agents/rules/ now scanned for pre-existing user rules (was missing; only .agents/prompts/ was scanned)

Upgrade path

npx github:zafrirron/Agent-Platform --mode=upgrade

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform

Agent Platform Bootstrap v2.33.0

Agent Platform Bootstrap v2.33.0

Built by agents. For agents. To build better agents.

---

Install

npx github:zafrirron/Agent-Platform

Start your first session — paste into your AI agent chat:

Read .agent/session-start.md and execute it.

---

What's new in v2.33.0

Fixed — All 9 agent manifests fully synced with actual capabilities (16 gaps closed)

Agent .manifest.json files had drifted behind the actual .md files across multiple releases. Full audit and resync:

Agent	Gaps closed
architect	Added: SOLID-principles, design-gate, code-standards-review, DRY-enforcement, layer-boundaries; routing: threat model, design before code, SOLID
backend	Added: rate-limiting, idempotency-keys, SSRF-prevention, mass-assignment-protection, OWASP-API-compliance; routing: rate limit, SSRF, REST design
frontend	Added: XSS-prevention, CSP-configuration, CSRF-frontend, secure-storage; routing: accessibility, XSS, CSP, localStorage
devops	Added: SBOM, supply-chain-security, artifact-signing, OIDC-credentials, api-version-inventory, linting-enforcement, branching-strategy
test	Added: contract-testing, consumer-driven-contracts, mutation-testing, regression-tests, coverage-report-generation; routing: contract test, mutation testing
docs	Added: docs-registry-audit, new-doc-registration, staleness-detection; routing: docs registry, registry audit
security	Added: prompt-injection-detection, CSRF-prevention, SSRF-prevention, security-audit-logging, LLM-security, property-level-auth; routing: prompt injection, LLM security, agentic
data	Added: idempotency, zero-downtime-migration, N+1-detection, index-optimisation
critic	Added: DRY-detection, SOLID-violations, code-structure-review, magic-numbers, design-gate-verification, cross-framework-review; routing: cross-framework, DRY, magic number

Prevention: MAINTAINER/platform-maintainer-agent.md now has a "Manifest sync" hard rule — manifests must be updated on every agent capability change.

Upgrade path

npx github:zafrirron/Agent-Platform --mode=upgrade

---

Upgrade

npx github:zafrirron/Agent-Platform --mode=upgrade

Your agents get improved rules. Your project customisations are untouched.

Uninstall

npx github:zafrirron/Agent-Platform --mode=uninstall          # dry run — shows what will be removed
npx github:zafrirron/Agent-Platform --mode=uninstall --confirm # removes everything

Removes all platform AI coordination files and restores any AI config files that existed before install.

---

Built by agents. For agents. To build better agents.
https://github.com/zafrirron/Agent-Platform