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/fmind/fkf/agents-mdgit clone --depth 1 https://github.com/fmind/fkfWhat 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 | $0.02131 | $0.02131 |
| Opus 5 | $0.01066 | $0.01066 |
| Sonnet 5 | $0.00426 | $0.00426 |
| Haiku 4.5 | $0.00213 | $0.00213 |
Grade A, and why
fkf 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 yesterday.
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 — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md (Project)
This repository builds fkf: one Go module, one binary, and two embedded skills. A base is a git repository of JSON and Markdown that FKF collects, links as URIs, and retrieves for coding agents. README.md explains the product; this file defines the contributor contract.
Status
The design is implemented. Do not add a compatibility path, migration, or second spelling. Configuration and collected evidence each use fkf: 1; the containing file identifies the contract. The evidence envelope is permanent and additive because providers may no longer retain old records. A change that requires forced re-collection is the wrong design.
Workflow
- Use
.agents/skills/fkf-contribute/SKILL.mdas the route map. This file is authoritative if they disagree. - Run focused tests after behavior changes, then
mise run all. Never weaken, skip, or hide a failing check. - Do not commit, push, publish, rename, delete outside the working tree, or take an irreversible action unless explicitly asked.
- Keep tests hermetic: use a temporary
HOME, unsetFKF_BASE, fake every declared command throughRunner, and use synthetic fixtures. The narrow exceptions are tests ofcore/shell, the synthetic relative-base helper regression, and invariant or status-policy tests that run localgo listorgit.
Product boundaries
- One module and binary. The root module is
github.com/fmind/fkf; do not addgo.work, a nested module, plugin, provider SDK, or second shipped command.docs/go.modonly pins the Hugo theme. - POSIX execution. FKF supports Linux and macOS;
run:,test:, andbody:are direct argv and a helper's shebang selects its interpreter. Cancellation uses POSIX process groups. WSL2 is Linux; native Windows is out of scope. - One base boundary.
<base>/fkf.yamlplus ignoredfkf.local.yamlare the only configuration. There is no global config, profile, bundle, visibility, sensitivity, capture, or policy layer. - Plain durable data. Bases contain JSON, Markdown, and the generated graph TSV. Never store rendered commands, planner state, or mutable execution metadata in evidence.
- Published paths only. Resolve paths through
Store.Resolve. It admits enabled layers, graph artifacts,fkf.yaml, andAGENTS.md; it rejects every other root file and any symlink below the base. - Small, typed Go. Keep cancellation explicit, split natural seams rather than line counts, and prefer a pure helper to a framework.
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.
- yesterday First seen · 70 lines · 2,131 tokens per session scan A 5c8904a2ec7f
fkf AGENTS.md is an instructions file published in the GitHub repository fmind/fkf (2 stars, last pushed 2d ago), licensed MIT. It adds 2,131 tokens to every session, about $0.0107 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
siyuan AGENTS.md
AGENTS.md instructions for siyuan-note/siyuan, covering agents.md, 1. non-negotiable constraints, do not hand-edit, verification and prohibited operations and 2. project-specific rules.
stashbase AGENTS.md
Instructions for liliu-z/stashbase, covering agent maintenance contract for this repo, github access for this repository, temporary worktrees, electron launch environment and documentation route.
Starcat AGENTS.md
Instructions for starcat-app/Starcat, covering agents.md, 🚨 硬性铁律（每次写代码前必读，违反即返工）, 🌿 git 分支与 worktree（强制）, 🧭 主进度索引（每次开工前必读） and 状态符号（与功能实现总览.md 同步）.
Starcat CLAUDE.md
Instructions for starcat-app/Starcat, covering claude.md, 🚨 硬性铁律（每次写代码前必读，违反即返工）, 🌿 git 分支与 worktree（强制）, 🧭 主进度索引（每次开工前必读） and 状态符号（与功能实现总览.md 同步）.
llm-wiki-newsroom AGENTS.md
Instructions for alfadur7/llm-wiki-newsroom, covering llm wiki newsroom — schema & workflow instructions, how to use, directory layout, page format and ingest workflow.
llm-wiki-newsroom GEMINI.md
Instructions for alfadur7/llm-wiki-newsroom, covering llm wiki newsroom — schema & workflow instructions, how to use, directory layout, page format and ingest workflow.