agent-kit CLAUDE.md

A set of Claude Code instructions for working on the agent-kit project. It defines when to search existing work, how to check project memory, and when to keep a task diary.

In plain words
What is it for?
Use it when adding a feature or changing three or more files in agent-kit. It helps search past work, check project history, and document implementation decisions in .tasks/.
Why use it?
It gives larger changes a consistent starting process and records useful findings. It also clarifies what to do when the project is not using Git or its memory index is unavailable.

Instructions file

Install

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.

agentmods
npx agentmods add instructions/work0r-ai/agent-kit/claude-md
Clone the repo
git clone --depth 1 https://github.com/work0r-ai/agent-kit
Per session 1,262 This file is loaded in full into every session.
When invoked 1,262 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.01262 $0.01262
Opus 5 $0.00631 $0.00631
Sonnet 5 $0.00252 $0.00252
Haiku 4.5 $0.00126 $0.00126

Measured 2d ago against content hash ef72f32739e2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-kit 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 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.

CLAUDE.md · 107 lines

How it starts

The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.

workorai-agent-kit — Agent Instructions


Implementation Protocol

Applies when a task is a feature or touches 3+ files. For smaller changes (1-2 files, typo, question, research-only), skip this protocol and work normally.

Preconditions (first step)

  • git: if the repo is under git, the commit rule is active. If not, skip commits and note Commit: N/A (no git) in the diary.
  • clawmem: check the index (index_stats / status). If not indexed, index it first (reindex; add the project to ~/.config/clawmem/index.yml if needed), then continue. If indexing is impossible, mirror the durable summary into the diary with clawmem: N/A.

1. Prior-art search (before planning)

Search whether this was built or researched before — including by other agents in other projects (clawmem covers all indexed collections simultaneously):

  • clawmem: search / intent_search for direct matches; find_similar for semantically close patterns; kg_query / find_causal_links for graph-based cross-project discovery
  • git: git log --grep and git log -S
  • tasks: grep past diaries in .tasks/

Record findings in the diary. If a similar feature exists in another project, reuse or adapt — note what was borrowed and why.

If a brainstorm precedes the work, run this same search during the brainstorm too — the findings sharpen the design doc. This does not replace the pre-planning search; do both.

2. Diary (.tasks/, gitignored)

Create .tasks/YYYY-MM-DD-<slug>.md first. One file per task. The agent only creates and appends — it never deletes (the human cleans manually).

Structure:

  • Plan: checklist of tickets (T1, T2, …)
  • Prior art: clawmem / git / tasks findings + conclusion
  • Per ticket on close: what done, how, deviations from plan + why, research, commit hash (if git), clawmem id

3. Stages & commits

A stage = an atomic, revertable unit that can be described as one change. Group tickets into a stage by this criterion, not by ticket count.

Read the full file on GitHub · 107 lines

Changes

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.

  1. 2d ago First seen · 107 lines · 1,262 tokens per session scan A ef72f32739e2

Subscribe to this mod's changes

agent-kit CLAUDE.md is an instructions file published in the GitHub repository work0r-ai/agent-kit (0 stars, last pushed 2mo ago), licensed MIT. It adds 1,262 tokens to every session, about $0.0063 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.