Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/gigsmart/haiku-method/claude-mdgit clone --depth 1 https://github.com/gigsmart/haiku-methodWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/gigsmart/haiku-method/claude-md)<a href="https://agentmods.dev/instructions/gigsmart/haiku-method/claude-md"><img src="https://agentmods.dev/badge/instructions/gigsmart/haiku-method/claude-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.12922 | $0.12922 |
| Opus 5 | $0.06461 | $0.06461 |
| Sonnet 5 | $0.02584 | $0.02584 |
| Haiku 4.5 | $0.01292 | $0.01292 |
Grade A, and why
haiku-method CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
H·AI·K·U Project
H·AI·K·U = Human + AI Knowledge Unification — a universal lifecycle framework for structured AI-assisted work.
Three-component project: plugin (Claude Code plugin), paper (methodology spec), website (Next.js 15 static site).
- Paper is the source of truth for methodology concepts
- Plugin is the source of truth for implementation (orchestrator, MCP tools, hooks, runtime behavior)
plugin/studios/ARCHITECTURE.mdis the structural reference for studio/stage/unit/hat/feedback boundaries within the plugin — the rules and contracts that apply across studios, distinct from any single implementation file. Read it before structural changes; conflict resolution between this doc and the plugin code is described in the doc's intro.- Website presents all of the above to users
Sync Discipline (CRITICAL)
When modifying any component, check if other components need corresponding updates:
| Change Type | Paper | Plugin | Website |
|---|---|---|---|
| New prompt | Mention in relevant section | Add handler in prompts/*.ts |
Update docs if user-facing |
| New studio | Document in Profiles section | Primary | Update docs |
| New stage | Document in relevant profile | Primary | Update docs |
| New hat (in stage) | Document in relevant profile | Add hats/{hat}.md file in stage directory |
Update docs if user-facing |
| New review agent (in stage) | Document in Quality Enforcement | Add review-agents/{agent}.md file in stage directory (resolved via 3-tier cascade — see "Hat / review-agent cascade" below) |
Update docs if user-facing |
| New fix hat (in stage) | Mention in Fix Loop section | Add hats/{hat}.md file in stage directory + reference in fix_hats: on STAGE.md |
Update docs if user-facing |
| New studio-level intent-completion review agent | Mention in Intent-Completion Review section | Add intent-review-agents/{agent}.md in the studio directory (per-studio), OR plugin/intent-review-agents/{agent}.md for a GLOBAL agent every studio inherits (studio files override global by name). NOT per-stage. Directory was renamed from review-agents/ 2026-05-18 so the studio-level slot in the stage cascade is free |
Update docs if user-facing |
| New global hat / review agent (any studio) | Document if added | Add to plugin/hats/{hat}.md or plugin/review-agents/{agent}.md — resolved by every stage in every studio unless a more-specific tier overrides |
Update docs if user-facing |
| New studio-level fix hat | Mention in Intent-Completion Review section | Add fix-hats/{hat}.md — plugin/fix-hats/{hat}.md for a GLOBAL one every studio inherits (e.g. validator, reconciler), OR plugin/studios/{name}/fix-hats/{hat}.md for a studio override (overrides global by name). NOT per-stage. Resolved via the global → studio cascade (project .haiku/ overrides plugin at each tier) |
Update docs if user-facing |
| New stage elaboration guidance | Mention in Stages section if needed | Add phases/ELABORATION.md (additive elaborate-prompt content, not an engine override) |
Update docs if user-facing |
| New reflection dimension | Document in Reflection section | Add reflections/{dim}.md file in studio directory — injected as a lens into the automatic intent-close record_reflection phase |
Update docs if user-facing |
| New lifecycle phase | Document as new section | Implement | Update docs |
| Terminology change | Update all references | Update all references | Update all references |
| New principle | Document in Principles section | Implement if applicable | Update if referenced |
| Concept refinement | Update definition | Update implementation | Update docs |
| Persistence change | N/A (environment-detected) | Update state-tools.ts isGitRepo | Update docs if user-facing |
| New harness support | N/A | Add entry to HARNESS_REGISTRY in harness.ts, update rewriting rules in harness-instructions.ts |
Update docs if user-facing |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 152 lines · 12,922 tokens per session scan A a10ad8618c2f
haiku-method CLAUDE.md is an instructions file published in the GitHub repository gigsmart/haiku-method (24 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 12,922 tokens to every session, about $0.0646 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
pdd GEMINI.md
Instructions for promptdriven/pdd, covering gemini / antigravity notes and pdd cli development.
Chorus CLAUDE.md
Claude Code instructions for Chorus-AIDLC/Chorus, covering claude.md — chorus project guide, what is chorus, tech stack, project structure and key commands.
add AGENTS.md
Instructions for MountainUnicorn/add, covering add, engagement protocol, spec-first invariants, maturity & autonomy ceiling and currently active spec.
ai-interview-kit CLAUDE.md
Instructions for CyannSHI/ai-interview-kit: This project includes 3 AI skills for end-to-end AI phone interview support. Each skill asks the user to choose their language (English/中文) at the start, then adapts accordingly.
SDLC-Enterprise-Framework CLAUDE.md
Claude Code instructions for Minh-Tam-Solution/SDLC-Enterprise-Framework, covering claude.md - ai assistant guidelines for sdlc 6.5.0, repository purpose, key framework concepts, code folder organization (not stage-mapped) and code file naming standards (new in 4.9.1).
smyslokod-starter CLAUDE.md
Claude Code instructions for Avalon-27reg/smyslokod-starter, covering claude.md — карта проекта для claude code, 1. что это за проект, 2. стек, 3. структура репозитория and 4. работа с business/.