# Agents Playbook

 > The open engineering harness for coding agents.
 > Rules, prompts, memory, evals, and executable gates train repeatable behavior, not model weights.
> Built by AgentsKit (https://www.agentskit.io) — the agent-native platform this playbook is distilled from.

- Site: https://playbook.agentskit.io
- Site map: https://playbook.agentskit.io/llms.txt
- Built by: https://www.agentskit.io
- Full bundle (single file, LLM-friendly): https://playbook.agentskit.io/llms-full.txt
- ZIP bundle: https://playbook.agentskit.io/playbook-bundle.zip
- Deterministic site config: https://playbook.agentskit.io/deterministic/site-config.json
- Deterministic knowledge: https://playbook.agentskit.io/deterministic/knowledge.json
- License: CC-BY-4.0

## AgentsKit ecosystem

- [AgentsKit](https://www.agentskit.io/docs) — Build agents without gluing many libraries together. Role: `foundation`. Maturity: beta. Machine index: https://www.agentskit.io/llms.txt
- [AgentsKit Registry](https://registry.agentskit.io/docs) — Copy ready-made agents and own the source. Role: `starting-point`. Maturity: beta. Machine index: https://registry.agentskit.io/llms.txt
- [AgentsKit Chat](https://chat.agentskit.io/docs) — Define one agent experience and deliver it across interfaces. Role: `experience`. Maturity: alpha. Machine index: https://chat.agentskit.io/llms.txt
- [Agents Playbook](https://playbook.agentskit.io/docs) **(current)** — Make agents ship code a human would actually merge. Role: `discipline`. Maturity: stable. Machine index: https://playbook.agentskit.io/llms.txt
- [Doc Bridge](https://doc-bridge.agentskit.io/) — Turn repository documentation into executable agent handoffs. Role: `understanding`. Maturity: stable. Machine index: https://doc-bridge.agentskit.io/llms.txt
- [AgentsKit OS](https://akos.agentskit.io/docs) — Run and govern agents in production. Role: `operation`. Maturity: stable. Machine index: https://akos.agentskit.io/llms.txt
## Docs index

- [Welcome](https://playbook.agentskit.io/docs) — The open engineering harness for coding agents — train repeatable behavior, not model weights.
  Raw: https://playbook.agentskit.io/raw/index.md
- [AgentsKit Chat dogfood](https://playbook.agentskit.io/docs/agentskit-chat) — How Ask Playbook composes the public AgentsKit Chat framework without recreating chat state.
  Raw: https://playbook.agentskit.io/raw/agentskit-chat.md
- [Contribute a production-earned pattern](https://playbook.agentskit.io/docs/contributing) — Turn a reproducible engineering lesson into a reviewable Playbook pattern, prompt, template, or gate.
  Raw: https://playbook.agentskit.io/raw/contributing.md
- [Local-first pattern discovery](https://playbook.agentskit.io/docs/discovery) — See which Playbook questions answer instantly in the browser and when repository-specific synthesis uses the backend.
  Raw: https://playbook.agentskit.io/raw/discovery.md
- [For agents](https://playbook.agentskit.io/docs/for-agents) — The shortest reliable path for an agent to retrieve, apply, and verify the Agents Playbook.
  Raw: https://playbook.agentskit.io/raw/for-agents.md
- [Getting started](https://playbook.agentskit.io/docs/getting-started) — Adopt the playbook in your project in under an hour.
  Raw: https://playbook.agentskit.io/raw/getting-started.md
- [Glossary](https://playbook.agentskit.io/docs/glossary) — Short definitions for the terms used across this playbook.
  Raw: https://playbook.agentskit.io/raw/glossary.md
- [The matrix — pillars × phases](https://playbook.agentskit.io/docs/matrix) — This is the master content map. Each cell points to the practices that apply when pillar meets SDLC phase.
  Raw: https://playbook.agentskit.io/raw/matrix.md
- [Onboard Your Agent](https://playbook.agentskit.io/docs/onboard-your-agent) — One paste that feeds any coding agent the whole playbook and has it audit your repo — Claude Code, Cursor, Windsurf, Codex, Copilot, and more.
  Raw: https://playbook.agentskit.io/raw/onboard-your-agent.md
- [Phase 01 — Discover](https://playbook.agentskit.io/docs/phases/01-discover) — Define what you are building, who consumes it, and what success looks like — before agents touch the codebase.
  Raw: https://playbook.agentskit.io/raw/phases/01-discover.md
- [Phase 02 — Design](https://playbook.agentskit.io/docs/phases/02-design) — Turn the discover brief into ADRs, RFCs, and a contract package skeleton the build phase can compose against.
  Raw: https://playbook.agentskit.io/raw/phases/02-design.md
- [Phase 03 — Build](https://playbook.agentskit.io/docs/phases/03-build) — Where most of the agent-augmented work happens. Discipline shifts from 'writing code' to 'stating intent, then verifying output against it'.
  Raw: https://playbook.agentskit.io/raw/phases/03-build.md
- [Phase 04 — Test](https://playbook.agentskit.io/docs/phases/04-test) — How 'tests pass' stops being a feeling and starts being a contract.
  Raw: https://playbook.agentskit.io/raw/phases/04-test.md
- [Phase 05 — Ship](https://playbook.agentskit.io/docs/phases/05-ship) — How to turn a green main into a release without surprising consumers.
  Raw: https://playbook.agentskit.io/raw/phases/05-ship.md
- [Phase 06 — Operate](https://playbook.agentskit.io/docs/phases/06-operate) — What 'running it' looks like after agents have shipped the first release.
  Raw: https://playbook.agentskit.io/raw/phases/06-operate.md
- [Pillar — AI Collaboration](https://playbook.agentskit.io/docs/pillars/ai-collaboration) — How to make an agent productive in your repo on day one and durably good across sessions.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration.md
- [Agent Compatibility Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/agent-compatibility-pattern) — One playbook, every agent — how to map the bootstrap-doc convention onto Claude Code, Cursor, Windsurf, Copilot, Codex, and the rest.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/agent-compatibility-pattern.md
- [Bootstrap Doc Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/bootstrap-doc-pattern) — The file an agent reads first, every session. Two files together: one for non-negotiables, one for routing.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/bootstrap-doc-pattern.md
- [Concurrent Agent Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/concurrent-agent-pattern) — How to survive — and benefit from — multiple agents working in the same repo at the same time.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/concurrent-agent-pattern.md
- [Context Management Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/context-management-pattern) — Engineer what goes into the context window — selection, ordering, compaction, retention — because the context is the program the model runs.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/context-management-pattern.md
- [Hallucination Reduction Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/hallucination-reduction-pattern) — Make model output you can put your name on — grounding, constrained generation, verification, and abstention — so the system says "I don't know" instead of confidently inventing.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/hallucination-reduction-pattern.md
- [Honest Confidence Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/honest-confidence-pattern) — How to report done-ness so a human can trust it — separating what was automatically verified from what was merely claimed.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/honest-confidence-pattern.md
- [Human-in-the-Loop Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/human-in-the-loop-pattern) — Design where humans review, approve, correct, and take over from agents — so autonomy scales with confidence and every correction becomes training signal.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/human-in-the-loop-pattern.md
- [Memory Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/memory-pattern) — How to make agents durably learn from prior sessions without polluting context with chat transcripts.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/memory-pattern.md
- [Open Knowledge Format Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/open-knowledge-format-pattern) — Represent agent-readable knowledge as a directory of markdown + YAML frontmatter — the OKF open standard, and how it formalizes patterns this playbook already uses.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/open-knowledge-format-pattern.md
- [Prompt Versioning Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/prompt-versioning-pattern) — Treat prompts as versioned, tested, experimentable production assets — not strings buried in code — so you can A/B them, measure impact, and roll back instantly.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/prompt-versioning-pattern.md
- [Self-Describe Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/self-describe-pattern) — Ship a machine-readable manifest of what your system can do, so external agents can discover and call it without reading your source.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/self-describe-pattern.md
- [Slash Commands Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/slash-commands-pattern) — How to turn repeated workflows into palette-invoked commands so they run identically every time.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/slash-commands-pattern.md
- [Sub-agent Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/sub-agent-pattern) — How to delegate scoped tasks to specialist agents so the orchestrator stays focused.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/sub-agent-pattern.md
- [Tool & Capability Design Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/tool-design-pattern) — Design the abstraction layer between a model and your product — tools, file systems, artifacts, skills — so complex multi-step work feels natural to the model and stays reliable…
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/tool-design-pattern.md
- [AI Collaboration — Universal Principles](https://playbook.agentskit.io/docs/pillars/ai-collaboration/universal) — How to get production-grade work out of an AI coding agent, durably, across sessions and across multiple agents working in parallel.
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/universal.md
- [Verify-First Pattern](https://playbook.agentskit.io/docs/pillars/ai-collaboration/verify-first-pattern) — Why an agent must trust live code over docs, plans, and its own memory — and how to verify before claiming or "fixing".
  Raw: https://playbook.agentskit.io/raw/pillars/ai-collaboration/verify-first-pattern.md
- [Pillar — Architecture](https://playbook.agentskit.io/docs/pillars/architecture) — How to keep a codebase legible and modular under multi-agent development.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture.md
- [ADR Pattern](https://playbook.agentskit.io/docs/pillars/architecture/adr-pattern) — How to record architecture decisions so future agents (and humans) can find them, trust them, and supersede them cleanly.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/adr-pattern.md
- [Anti-Overengineering](https://playbook.agentskit.io/docs/pillars/architecture/anti-overengineering) — How to keep agents (and engineers) from building three layers of abstraction for what would have been ten lines of code.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/anti-overengineering.md
- [API Gateway Pattern](https://playbook.agentskit.io/docs/pillars/architecture/api-gateway-pattern) — The edge between clients and your services — what belongs there, what doesn't, and how to keep it from becoming a monolith in disguise.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/api-gateway-pattern.md
- [API Versioning + Deprecation Pattern](https://playbook.agentskit.io/docs/pillars/architecture/api-versioning-pattern) — How to evolve public contracts without surprising consumers.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/api-versioning-pattern.md
- [Caching + CDN Pattern](https://playbook.agentskit.io/docs/pillars/architecture/caching-cdn-pattern) — How to layer caches so the system is fast, cheap, and consistent — three properties at tension.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/caching-cdn-pattern.md
- [Contracts — Zod Method Registry Pattern](https://playbook.agentskit.io/docs/pillars/architecture/contracts-zod-pattern) — TS-concrete recipe for a typed JSON-RPC / HTTP / IPC boundary. Scales to several hundred methods across dozens of namespaces in a real production codebase.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/contracts-zod-pattern.md
- [Distributed Data Pattern](https://playbook.agentskit.io/docs/pillars/architecture/distributed-data-pattern) — How to design data layout when one database stops being enough — read replicas, sharding, replication lag, CAP trade-offs, eventual consistency.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/distributed-data-pattern.md
- [Error Hierarchy](https://playbook.agentskit.io/docs/pillars/architecture/error-hierarchy) — How to design an error model that survives multi-agent development and client-side pattern matching.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/error-hierarchy.md
- [Event Streaming Pattern](https://playbook.agentskit.io/docs/pillars/architecture/event-streaming-pattern) — How to design async, decoupled communication via queues, pub/sub, and event streams — without losing events, double-processing, or stalling consumers.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/event-streaming-pattern.md
- [Feature Flags Pattern](https://playbook.agentskit.io/docs/pillars/architecture/feature-flags-pattern) — How to ship code separately from shipping behavior, without accumulating a flag graveyard.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/feature-flags-pattern.md
- [File-size Budget](https://playbook.agentskit.io/docs/pillars/architecture/file-size-budget) — How to keep files reviewable when agents write most of the code.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/file-size-budget.md
- [Infrastructure as Code (IaC) Pattern](https://playbook.agentskit.io/docs/pillars/architecture/iac-pattern) — How to define + version + review + apply infrastructure as code, instead of clicking around cloud consoles.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/iac-pattern.md
- [Multi-Region Pattern](https://playbook.agentskit.io/docs/pillars/architecture/multi-region-pattern) — How to operate across geographic regions for latency, availability, and data sovereignty — without losing your mind.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/multi-region-pattern.md
- [Offline-First + Sync Pattern](https://playbook.agentskit.io/docs/pillars/architecture/offline-first-sync-pattern) — How to design apps that work without network connectivity and reconcile state when connectivity returns.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/offline-first-sync-pattern.md
- [Platform Engineering + Internal Developer Platform Pattern](https://playbook.agentskit.io/docs/pillars/architecture/platform-engineering-idp-pattern) — How to build the layer between cloud / infra and product engineers, so product teams ship fast without re-learning infra each time.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/platform-engineering-idp-pattern.md
- [RFC Pattern](https://playbook.agentskit.io/docs/pillars/architecture/rfc-pattern) — When the decision is bigger than your team. Used for changes to public contracts, wire formats, plugin protocols, and any breaking change a consumer outside this repo would notice.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/rfc-pattern.md
- [Service Mesh Pattern](https://playbook.agentskit.io/docs/pillars/architecture/service-mesh-pattern) — How to handle cross-service communication concerns (mTLS, retries, observability, routing) without coding them in every service.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/service-mesh-pattern.md
- [Architecture — TS / Node ≥22 / pnpm Monorepo (Concrete)](https://playbook.agentskit.io/docs/pillars/architecture/ts-concrete) — Copy-paste-ready recipes that implement [`universal.md`](/docs/pillars/architecture/universal) on a TypeScript stack. Calibrated on a real multi-package, multi-app monorepo buil…
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/ts-concrete.md
- [Architecture — Universal Principles](https://playbook.agentskit.io/docs/pillars/architecture/universal) — Stack-agnostic. Applies to any language, any framework.
  Raw: https://playbook.agentskit.io/raw/pillars/architecture/universal.md
- [Pillar — Governance](https://playbook.agentskit.io/docs/pillars/governance) — How multiple agents (and humans) coordinate work in one repo without subtracting each other's progress.
  Raw: https://playbook.agentskit.io/raw/pillars/governance.md
- [Merge Rules Pattern](https://playbook.agentskit.io/docs/pillars/governance/merge-rules-pattern) — How to resolve conflicts so the merge sums work instead of subtracting it.
  Raw: https://playbook.agentskit.io/raw/pillars/governance/merge-rules-pattern.md
- [Phased PR Pattern](https://playbook.agentskit.io/docs/pillars/governance/phased-pr-pattern) — How to ship initiatives too big for one PR without ending up with a long-lived branch hell.
  Raw: https://playbook.agentskit.io/raw/pillars/governance/phased-pr-pattern.md
- [PR Intent Pattern](https://playbook.agentskit.io/docs/pillars/governance/pr-intent-pattern) — The manifest that makes a PR's claims verifiable.
  Raw: https://playbook.agentskit.io/raw/pillars/governance/pr-intent-pattern.md
- [Tombstone Pattern](https://playbook.agentskit.io/docs/pillars/governance/tombstone-pattern) — How to retire a doc, plan, ADR, screen, or package without losing the trail.
  Raw: https://playbook.agentskit.io/raw/pillars/governance/tombstone-pattern.md
- [Governance — Universal Principles](https://playbook.agentskit.io/docs/pillars/governance/universal) — How multiple contributors — agents and humans — coordinate so the whole sums.
  Raw: https://playbook.agentskit.io/raw/pillars/governance/universal.md
- [Pillar — Quality](https://playbook.agentskit.io/docs/pillars/quality) — How to know the code works without manually reviewing every agent-produced diff.
  Raw: https://playbook.agentskit.io/raw/pillars/quality.md
- [Adversarial Bug-Hunt Pattern](https://playbook.agentskit.io/docs/pillars/quality/adversarial-bug-hunt-pattern) — How to find real logic bugs with a find → refute → reproduce loop, instead of trusting a single agent that says "looks fine".
  Raw: https://playbook.agentskit.io/raw/pillars/quality/adversarial-bug-hunt-pattern.md
- [Agent Eval Framework Pattern](https://playbook.agentskit.io/docs/pillars/quality/agent-eval-framework-pattern) — How to measure whether an AI agent is actually good — deterministic checks, LLM-as-judge, and production monitoring — so quality scales with the system instead of breaking as it…
  Raw: https://playbook.agentskit.io/raw/pillars/quality/agent-eval-framework-pattern.md
- [Agent Harness Pattern](https://playbook.agentskit.io/docs/pillars/quality/agent-harness-pattern) — How to make coding-agent delivery evidence-backed instead of trust-based.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/agent-harness-pattern.md
- [Alerting + Runbooks Pattern](https://playbook.agentskit.io/docs/pillars/quality/alerting-runbooks-pattern) — How to turn observability data into a healthy alerting setup — pages that matter, runbooks that work, alert hygiene that prevents pager fatigue.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/alerting-runbooks-pattern.md
- [Chaos Engineering Pattern](https://playbook.agentskit.io/docs/pillars/quality/chaos-engineering-pattern) — How to find failure modes before customers do — deliberately, in a controlled way.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/chaos-engineering-pattern.md
- [CI/CD Pipeline Pattern](https://playbook.agentskit.io/docs/pillars/quality/ci-cd-pipeline-pattern) — How to wire continuous integration + continuous delivery so the path from commit to production is short, verifiable, reversible, and boring.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/ci-cd-pipeline-pattern.md
- [Contract Testing Pattern](https://playbook.agentskit.io/docs/pillars/quality/contract-testing-pattern) — How to verify that consumer + provider agree on a contract, without expensive end-to-end tests.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/contract-testing-pattern.md
- [Cost Optimization Pattern (FinOps)](https://playbook.agentskit.io/docs/pillars/quality/cost-optimization-pattern) — How to control cloud spend without micromanaging every commit.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/cost-optimization-pattern.md
- [Fail-Loud Defaults Pattern](https://playbook.agentskit.io/docs/pillars/quality/fail-loud-defaults-pattern) — Why a silent no-op default plus a test harness that overrides it lets broken production wiring pass CI green.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/fail-loud-defaults-pattern.md
- [Mutation Testing Pattern](https://playbook.agentskit.io/docs/pillars/quality/mutation-testing-pattern) — How to score whether your unit tests actually catch bugs, beyond what coverage tells you.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/mutation-testing-pattern.md
- [Observability Pattern](https://playbook.agentskit.io/docs/pillars/quality/observability-pattern) — How to know what the system is doing in production, beyond 'tests passed'.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/observability-pattern.md
- [Performance Budgets Pattern](https://playbook.agentskit.io/docs/pillars/quality/performance-budgets-pattern) — How to keep 'it's fast enough' from drifting into 'why is it slow?'
  Raw: https://playbook.agentskit.io/raw/pillars/quality/performance-budgets-pattern.md
- [Pre-push Pattern](https://playbook.agentskit.io/docs/pillars/quality/pre-push-pattern) — The safety net between local changes and CI.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/pre-push-pattern.md
- [Product Analytics + Experimentation Pattern](https://playbook.agentskit.io/docs/pillars/quality/product-analytics-experimentation-pattern) — How to measure what users do, run experiments cleanly, and let data drive product decisions — without inviting bias or noise.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/product-analytics-experimentation-pattern.md
- [Property-Based & Fuzz Testing Pattern](https://playbook.agentskit.io/docs/pillars/quality/property-fuzz-testing-pattern) — How to test the laws your code must obey, instead of guessing individual inputs — and how this catches crypto and parser bugs example tests never will.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/property-fuzz-testing-pattern.md
- [Quality Gates Pattern](https://playbook.agentskit.io/docs/pillars/quality/quality-gates-pattern) — How to bundle structural rules into one fast command an agent runs before every push.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/quality-gates-pattern.md
- [Sanity Pattern](https://playbook.agentskit.io/docs/pillars/quality/sanity-pattern) — The cross-cutting audit that catches what individual gates miss.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/sanity-pattern.md
- [Test Pyramid](https://playbook.agentskit.io/docs/pillars/quality/test-pyramid) — How to mix test types so the cheap ones catch most bugs and the expensive ones cover what only they can.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/test-pyramid.md
- [Quality — Universal Principles](https://playbook.agentskit.io/docs/pillars/quality/universal) — How to know agent-produced code works without manually reading every diff.
  Raw: https://playbook.agentskit.io/raw/pillars/quality/universal.md
- [Pillar — Security](https://playbook.agentskit.io/docs/pillars/security) — How to build security in from day one when the people writing the code are agents, not a hardened security team.
  Raw: https://playbook.agentskit.io/raw/pillars/security.md
- [AI / LLM Safety Pattern](https://playbook.agentskit.io/docs/pillars/security/ai-llm-safety-pattern) — How to ship LLM-powered features without prompt injection, jailbreaks, data leakage, hallucinated tools, or runaway costs.
  Raw: https://playbook.agentskit.io/raw/pillars/security/ai-llm-safety-pattern.md
- [Audit Ledger Pattern](https://playbook.agentskit.io/docs/pillars/security/audit-ledger-pattern) — How to prove who did what when, in a way that survives compromise.
  Raw: https://playbook.agentskit.io/raw/pillars/security/audit-ledger-pattern.md
- [Compliance Framework Pattern](https://playbook.agentskit.io/docs/pillars/security/compliance-framework-pattern) — How to map SOC 2 / ISO 27001 / GDPR / HIPAA controls onto your codebase + ops, so audits are evidence-driven, not theatre.
  Raw: https://playbook.agentskit.io/raw/pillars/security/compliance-framework-pattern.md
- [Container + Kubernetes Security Pattern](https://playbook.agentskit.io/docs/pillars/security/container-k8s-security-pattern) — How to ship containers + run them on Kubernetes without making the runtime an attack surface.
  Raw: https://playbook.agentskit.io/raw/pillars/security/container-k8s-security-pattern.md
- [Data Classification Pattern](https://playbook.agentskit.io/docs/pillars/security/data-classification-pattern) — How to label every field by sensitivity, then make the labels drive redaction, retention, residency, and access.
  Raw: https://playbook.agentskit.io/raw/pillars/security/data-classification-pattern.md
- [Dependency Hygiene Pattern](https://playbook.agentskit.io/docs/pillars/security/dependency-hygiene-pattern) — How to keep your dependency tree healthy across thousands of transitive packages without it becoming a full-time job.
  Raw: https://playbook.agentskit.io/raw/pillars/security/dependency-hygiene-pattern.md
- [Egress Firewall Pattern](https://playbook.agentskit.io/docs/pillars/security/egress-firewall-pattern) — Default-deny outbound network access, configurable per workspace, evaluable at every call site.
  Raw: https://playbook.agentskit.io/raw/pillars/security/egress-firewall-pattern.md
- [Governance Posture Pattern](https://playbook.agentskit.io/docs/pillars/security/governance-posture-pattern) — Expose the system's live security and governance posture as a read-only, machine-readable signal — so agents, dashboards, and auditors can see what is enforced without admin rig…
  Raw: https://playbook.agentskit.io/raw/pillars/security/governance-posture-pattern.md
- [Multi-Tenant Isolation Pattern](https://playbook.agentskit.io/docs/pillars/security/multi-tenant-isolation-pattern) — How to host many customers in one system without one customer leaking, blocking, or paying for another.
  Raw: https://playbook.agentskit.io/raw/pillars/security/multi-tenant-isolation-pattern.md
- [On-Call Rotation Pattern](https://playbook.agentskit.io/docs/pillars/security/on-call-rotation-pattern) — How to keep a system reliable 24/7 without burning out the team that maintains it.
  Raw: https://playbook.agentskit.io/raw/pillars/security/on-call-rotation-pattern.md
- [Rate Limiting + DDoS Protection Pattern](https://playbook.agentskit.io/docs/pillars/security/rate-limiting-ddos-pattern) — How to absorb traffic spikes — both benign (popular feature) and malicious (DDoS / abuse) — without breaking for legitimate users.
  Raw: https://playbook.agentskit.io/raw/pillars/security/rate-limiting-ddos-pattern.md
- [RBAC Pattern](https://playbook.agentskit.io/docs/pillars/security/rbac-pattern) — How to model 'who can do what' so the checks survive multi-agent edits.
  Raw: https://playbook.agentskit.io/raw/pillars/security/rbac-pattern.md
- [Secrets-Leak Post-Mortem Playbook](https://playbook.agentskit.io/docs/pillars/security/secrets-leak-postmortem-playbook) — When a secret leaks — git history, log, screenshot, public artefact — what to do, in what order, within what timeline.
  Raw: https://playbook.agentskit.io/raw/pillars/security/secrets-leak-postmortem-playbook.md
- [Secrets Management Deep Pattern](https://playbook.agentskit.io/docs/pillars/security/secrets-mgmt-deep-pattern) — Beyond `vault-pattern.md` — operational specifics: dynamic secrets, short-lived credentials, secret-zero, secret-less architectures, OIDC federation.
  Raw: https://playbook.agentskit.io/raw/pillars/security/secrets-mgmt-deep-pattern.md
- [Session Management Pattern](https://playbook.agentskit.io/docs/pillars/security/session-mgmt-pattern) — How to model + protect user sessions — login state, refresh, revocation, idle expiry, device binding.
  Raw: https://playbook.agentskit.io/raw/pillars/security/session-mgmt-pattern.md
- [Threat Model Template](https://playbook.agentskit.io/docs/pillars/security/threat-model-template) — Skeleton for the per-project threat model doc. Lives at `docs/security/threat-model.md`. Revisited every release.
  Raw: https://playbook.agentskit.io/raw/pillars/security/threat-model-template.md
- [Security — Universal Principles](https://playbook.agentskit.io/docs/pillars/security/universal) — How to bake security into the codebase from PR #1, when agents (not a hardened security team) write most of it.
  Raw: https://playbook.agentskit.io/raw/pillars/security/universal.md
- [Vault Pattern](https://playbook.agentskit.io/docs/pillars/security/vault-pattern) — How to store secrets so a leaked log line, a stolen disk, or a curious agent does not compromise the system.
  Raw: https://playbook.agentskit.io/raw/pillars/security/vault-pattern.md
- [Vulnerability Management Pattern](https://playbook.agentskit.io/docs/pillars/security/vulnerability-mgmt-pattern) — How to keep the system patched without bottlenecking on a person watching CVE feeds.
  Raw: https://playbook.agentskit.io/raw/pillars/security/vulnerability-mgmt-pattern.md
- [Pillar — UI / UX](https://playbook.agentskit.io/docs/pillars/ui-ux) — How to keep a multi-screen product visually coherent, intl-ready, and accessible when agents are writing screens in parallel.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux.md
- [Accessibility Checklist](https://playbook.agentskit.io/docs/pillars/ui-ux/a11y-checklist) — What every shipped surface passes before merge, and how to prove it.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/a11y-checklist.md
- [Accessibility Deep Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/accessibility-deep-pattern) — Beyond the per-PR checklist — the substance of WCAG-AA conformance, the failure modes that matter, the testing discipline that catches what automation cannot.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/accessibility-deep-pattern.md
- [Design System Governance Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/design-system-governance-pattern) — How to keep a design system useful as it grows — without bloat, without dictatorial restrictions, without drift.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/design-system-governance-pattern.md
- [Design Tokens Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/design-tokens-pattern) — How to define visual values once and let them flow through every screen, theme, and brand.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/design-tokens-pattern.md
- [Empty States Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/empty-states-pattern) — How to turn empty surfaces from dead ends into next-step launchpads.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/empty-states-pattern.md
- [Internationalisation Deep Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/i18n-deep-pattern) — Beyond 'wrap every string in `t()`' — the substance of locale-correct UI: plural rules, gender, ICU formatting, RTL, dates, numbers, currency, sorting, search.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/i18n-deep-pattern.md
- [Intl Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/intl-pattern) — How to never ship hardcoded user-visible strings, so the next locale is configuration, not a sweep.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/intl-pattern.md
- [Primitives Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/primitives-pattern) — How to ship a one-package primitives catalog so every screen looks like the same product.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/primitives-pattern.md
- [UI / UX — Universal Principles](https://playbook.agentskit.io/docs/pillars/ui-ux/universal) — How to keep a multi-screen product visually coherent, intl-ready, accessible, and honest when agents author screens in parallel.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/universal.md
- [Whitelabel Pattern](https://playbook.agentskit.io/docs/pillars/ui-ux/whitelabel-pattern) — How to make every product surface reskinnable per tenant — even if you do not sell whitelabel today.
  Raw: https://playbook.agentskit.io/raw/pillars/ui-ux/whitelabel-pattern.md
- [Reusable prompts](https://playbook.agentskit.io/docs/prompts) — System prompts, sub-agent recipes, and slash-command bodies that consistently produce gold-standard output.
  Raw: https://playbook.agentskit.io/raw/prompts.md
- [Slash Command — /clear](https://playbook.agentskit.io/docs/prompts/slash-clear) — Reset session context cleanly without losing persistent memory.
  Raw: https://playbook.agentskit.io/raw/prompts/slash-clear.md
- [Slash Command — /goal](https://playbook.agentskit.io/docs/prompts/slash-goal) — Set a session goal with an explicit exit condition. Stops only when condition holds.
  Raw: https://playbook.agentskit.io/raw/prompts/slash-goal.md
- [Slash Command — /loop](https://playbook.agentskit.io/docs/prompts/slash-loop) — Run a task repeatedly. Either on a fixed interval or self-paced based on an exit condition.
  Raw: https://playbook.agentskit.io/raw/prompts/slash-loop.md
- [Slash Command — /review](https://playbook.agentskit.io/docs/prompts/slash-review) — Multi-agent code review pass on a PR (or current branch).
  Raw: https://playbook.agentskit.io/raw/prompts/slash-review.md
- [Slash Command — /sanity](https://playbook.agentskit.io/docs/prompts/slash-sanity) — Run the cross-cutting sanity audit, surface drift.
  Raw: https://playbook.agentskit.io/raw/prompts/slash-sanity.md
- [Slash Command — /ship](https://playbook.agentskit.io/docs/prompts/slash-ship) — Run the release-gate checklist. Does not actually release — produces the readiness verdict.
  Raw: https://playbook.agentskit.io/raw/prompts/slash-ship.md
- [Sub-agent Recipe — Code Explorer](https://playbook.agentskit.io/docs/prompts/subagent-code-explorer) — Deep-trace agent. Maps execution paths, dependencies, and abstraction layers across a feature so the orchestrator can reason about a change.
  Raw: https://playbook.agentskit.io/raw/prompts/subagent-code-explorer.md
- [Sub-agent Recipe — Code Reviewer](https://playbook.agentskit.io/docs/prompts/subagent-code-reviewer) — Confidence-filtered review pass on a diff. Returns issues; does not approve / merge.
  Raw: https://playbook.agentskit.io/raw/prompts/subagent-code-reviewer.md
- [Sub-agent Recipe — Explore](https://playbook.agentskit.io/docs/prompts/subagent-explore) — Read-only fan-out search agent. Orchestrator delegates 'find me X' tasks here.
  Raw: https://playbook.agentskit.io/raw/prompts/subagent-explore.md
- [Sub-agent Recipe — Plan](https://playbook.agentskit.io/docs/prompts/subagent-plan) — Designs a step-by-step implementation plan from a task description and codebase context. Does not implement.
  Raw: https://playbook.agentskit.io/raw/prompts/subagent-plan.md
- [System Prompt — Architect](https://playbook.agentskit.io/docs/prompts/system-architect) — Inject as system prompt when the task is designing a new package boundary, new contract, new ADR, or evaluating a structural change.
  Raw: https://playbook.agentskit.io/raw/prompts/system-architect.md
- [System Prompt — Implementer](https://playbook.agentskit.io/docs/prompts/system-implementer) — Inject as system prompt when the task is building a sub-unit against a finalised plan.
  Raw: https://playbook.agentskit.io/raw/prompts/system-implementer.md
- [System Prompt — Reviewer](https://playbook.agentskit.io/docs/prompts/system-reviewer) — Inject as system prompt for code review passes. Produces confidence-scored issue lists, not pile-on critique.
  Raw: https://playbook.agentskit.io/raw/prompts/system-reviewer.md
- [System Prompt — Security Reviewer](https://playbook.agentskit.io/docs/prompts/system-security) — Inject as system prompt for security-focused review passes. Narrower than the general reviewer; deeper on the threat model.
  Raw: https://playbook.agentskit.io/raw/prompts/system-security.md
- [Review agent-authored pull requests with executable evidence](https://playbook.agentskit.io/docs/review-agent-authored-contributions) — Replace subjective instructions with a local quality gate, reproduce a failure, and require the same evidence in CI before approving an agent-authored change.
  Raw: https://playbook.agentskit.io/raw/review-agent-authored-contributions.md
- [Reference gate scripts](https://playbook.agentskit.io/docs/scripts) — Drop-in reference implementations for the quality + structural gates the pillars rely on. Adapt to your stack; the contract each gate enforces is more important than the specifi…
  Raw: https://playbook.agentskit.io/raw/scripts.md
- [Templates](https://playbook.agentskit.io/docs/templates) — Copy-paste skeletons for the documents the playbook relies on — bootstrap docs, decision records, and the PR-intent manifest.
  Raw: https://playbook.agentskit.io/raw/templates.md
- [ADR Template](https://playbook.agentskit.io/docs/templates/ADR.template) — Skeleton for an Architecture Decision Record. Copy the block below into docs/adr/NNNN-<slug>.md and fill the bracketed parts.
  Raw: https://playbook.agentskit.io/raw/templates/ADR.template.md
- [AGENTS.md template](https://playbook.agentskit.io/docs/templates/AGENTS.md.template) — Drop at repo root. The routing table that answers 'which package owns X?' before an agent makes the wrong edit.
  Raw: https://playbook.agentskit.io/raw/templates/AGENTS.md.template.md
- [CLAUDE.md template](https://playbook.agentskit.io/docs/templates/CLAUDE.md.template) — Drop this at the repo root as `CLAUDE.md` (or `AGENTS.md` for tool-neutral). Customise the marked sections. Keep it under 200 lines; agents read the whole thing every session.
  Raw: https://playbook.agentskit.io/raw/templates/CLAUDE.md.template.md
- [MEMORY pattern](https://playbook.agentskit.io/docs/templates/MEMORY.md.template) — Persistent agent memory survives between sessions. The pattern: one fact per file, plus an index file.
  Raw: https://playbook.agentskit.io/raw/templates/MEMORY.md.template.md
- [PR Intent Manifest](https://playbook.agentskit.io/docs/templates/PR-intent.template) — Embed this block in every PR description. A gate parses it and verifies that the diff matches the claims.
  Raw: https://playbook.agentskit.io/raw/templates/PR-intent.template.md
- [RFC Template](https://playbook.agentskit.io/docs/templates/RFC.template) — Skeleton for a Request for Comments. Copy the block below into docs/rfc/NNNN-<slug>.md and fill the bracketed parts.
  Raw: https://playbook.agentskit.io/raw/templates/RFC.template.md
- [check-adr.example](https://playbook.agentskit.io/raw/scripts/check-adr.example.mjs) — Reference: verify ADR sequence integrity, status hygiene, superseder linkage.
  Raw: https://playbook.agentskit.io/raw/scripts/check-adr.example.mjs
- [check-completeness.example](https://playbook.agentskit.io/raw/scripts/check-completeness.example.mjs) — Reference: scan shipped surfaces for completeness violations — TODO, FIXME, `throw new Error("not implemented")`, `disabled: true` tabs, empty exported component bodies.
  Raw: https://playbook.agentskit.io/raw/scripts/check-completeness.example.mjs
- [check-error-raw.example](https://playbook.agentskit.io/raw/scripts/check-error-raw.example.mjs) — Reference: ban `throw new Error(...)` in boundary files.
  Raw: https://playbook.agentskit.io/raw/scripts/check-error-raw.example.mjs
- [check-file-size.example](https://playbook.agentskit.io/raw/scripts/check-file-size.example.mjs) — Reference implementation: per-extension line-count budget with shrink-only baseline.
  Raw: https://playbook.agentskit.io/raw/scripts/check-file-size.example.mjs
- [check-intl.example](https://playbook.agentskit.io/raw/scripts/check-intl.example.mjs) — Reference: detect hardcoded user-visible strings in JSX (text content + a11y attributes).
  Raw: https://playbook.agentskit.io/raw/scripts/check-intl.example.mjs
- [check-named-exports.example](https://playbook.agentskit.io/raw/scripts/check-named-exports.example.mjs) — Reference: ban `export default` outside framework-mandated files.
  Raw: https://playbook.agentskit.io/raw/scripts/check-named-exports.example.mjs
- [check-native-html.example](https://playbook.agentskit.io/raw/scripts/check-native-html.example.mjs) — Reference: ban native HTML interactive elements in shipped surfaces. Use shared primitives instead (Button, Input, Select, Dialog, Form, Table, Link).
  Raw: https://playbook.agentskit.io/raw/scripts/check-native-html.example.mjs
- [check-no-any.example](https://playbook.agentskit.io/raw/scripts/check-no-any.example.mjs) — Reference: ban `any` outside escape-hatched comments.
  Raw: https://playbook.agentskit.io/raw/scripts/check-no-any.example.mjs
- [check-pr-intent.example](https://playbook.agentskit.io/raw/scripts/check-pr-intent.example.mjs) — Reference: parse PR-intent manifest + cross-check against the diff.
  Raw: https://playbook.agentskit.io/raw/scripts/check-pr-intent.example.mjs
- [check-quality-gates.example](https://playbook.agentskit.io/raw/scripts/check-quality-gates.example.mjs) — Reference orchestrator: runs the structural quality gates in parallel, aggregates results, exits 0/1.
  Raw: https://playbook.agentskit.io/raw/scripts/check-quality-gates.example.mjs
- [check-rfc.example](https://playbook.agentskit.io/raw/scripts/check-rfc.example.mjs) — Reference: verify RFC index integrity + promotion linkage to ADRs.
  Raw: https://playbook.agentskit.io/raw/scripts/check-rfc.example.mjs
- [check-secrets.example](https://playbook.agentskit.io/raw/scripts/check-secrets.example.mjs) — Reference: scan for high-entropy strings, PEM blocks, and known API-key prefixes outside an explicit allow-list.
  Raw: https://playbook.agentskit.io/raw/scripts/check-secrets.example.mjs
- [check-tokens.example](https://playbook.agentskit.io/raw/scripts/check-tokens.example.mjs) — Reference: ban raw color literals (hex / rgb / hsl / oklch), Tailwind arbitrary color/spacing class values (bg-[#fff]), and inline color/spacing styles in JSX.
  Raw: https://playbook.agentskit.io/raw/scripts/check-tokens.example.mjs