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/ambystechcom/ambykit/agents-mdgit clone --depth 1 https://github.com/ambystechcom/AmbyKitWhat 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 | $0.01216 | $0.01216 |
| Opus 5 | $0.00608 | $0.00608 |
| Sonnet 5 | $0.00243 | $0.00243 |
| Haiku 4.5 | $0.00122 | $0.00122 |
Grade A, and why
AmbyKit AGENTS.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 2d 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 — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — AmbyKit
Guidance for AI coding assistants working on the AmbyKit codebase itself. (For using AmbyKit
in your own project, see README.md and docs/.)
What AmbyKit is
AmbyKit is a Spec-Driven Development (SDD) framework distributed as a Node/TypeScript CLI
(@ambystech/ambykit, bin ambykit). It helps AI coding assistants build better software by
producing structured artifacts — a project constitution, user stories, testable
requirements, UI design, an implementation plan, and an ordered task list — before
code is written.
The core design is author-once, emit-per-tool: one neutral source of truth (templates + phase prompts) is translated by per-tool emitters into each assistant's native command/rules format (Claude Code, OpenCode, GitHub Copilot + CLI, Cursor + CLI, Antigravity IDE + CLI).
Golden rules
- Author once, emit per tool. Never hand-maintain per-tool command files as source. The neutral
source lives in
src/templates/andsrc/prompts/; emitters insrc/emitters/generate tool files (e.g..claude/commands/amby.*.md,.github/prompts/). If you find yourself editing those by hand, stop — edit the neutral source and re-runambykit syncinstead. BaseEmitter/BaseCommandhold shared logic. New tool support = a thin subclass ofBaseEmitterthat sets itscommandSurface/paths and overrides only what differs. New CLI verb = a subclass ofBaseCommandimplementingexecute(). Don't duplicate orchestration.- Token frugality is a feature. Phase prompts and templates must minimize tokens the assistant
reads/writes: progressive disclosure (load only the artifact a phase needs via
@path), reference by stable ID instead of restating, and patch-in-place rather than regenerate. Seedocs/architecture.md→ Token efficiency. - We dogfood AmbyKit on AmbyKit. Features are specced under
specs/using the AmbyKit workflow. The repo's own.claude/etc. are generated byambykit syncand must stay in sync (CI checks).
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.
- 2d ago First seen · 81 lines · 1,216 tokens per session scan A cc03022e9cc7
AmbyKit AGENTS.md is an instructions file published in the GitHub repository ambystechcom/AmbyKit (2 stars, last pushed 6d ago), licensed MIT. It adds 1,216 tokens to every session, about $0.0061 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 2026-08-31.
Other instructions, from other repositories
thinkrail AGENTS.md
Instructions for JetBrains/thinkrail, covering thinkrail, module structure & boundaries (top-priority requirement), engine: pi only, in-process, architecture (three rings) and repo layout.
truthmark AGENTS.md
Instructions for merlinhu1/truthmark, covering authority, documentation scope, product boundary, rules and instruction surface boundary.
thinkrail CLAUDE.md
Instructions for JetBrains/thinkrail, a project described as: Vibe code with pi in a lightweight, real IDE - The Vibe You Need.
truthmark CLAUDE.md
Instructions for merlinhu1/truthmark, covering authority, documentation scope, product boundary, rules and instruction surface boundary.
coograph brutal-honesty.instructions.md
Global communication style — applies to all workspaces and conversations.
claude-code-second-brain AGENTS.md
Instructions for shofer-dev/claude-code-second-brain, covering agents.md, read first, invariants and working here.