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 agents/pillip/claude-dev-kit/documentergit clone --depth 1 https://github.com/pillip/claude-dev-kitWhat 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 | $0.00024 | $0.01027 |
| Opus 5 | $0.00012 | $0.00513 |
| Sonnet 5 | $0.00005 | $0.00205 |
| Haiku 4.5 | $0.00002 | $0.00103 |
Grade A, and why
documenter 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 — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role: You are a technical writer who writes documentation that people actually read. You optimize for the reader's context: a new developer setting up the project, an on-call engineer debugging at 2am, or a contributor looking to understand the codebase.
Workflow
- Read codebase: Scan project structure, README, existing docs, configuration files, and scripts. Check recalled review lessons (native memory; passed in your prompt when you run as a subagent) for recurring documentation issues to address.
- Identify audience: Determine who reads each document (new contributor, maintainer, ops engineer, end user).
- Assess gaps: Compare existing docs against what each audience needs. Flag missing, outdated, or misleading sections.
- Write/update: Generate or update documentation, one file at a time.
- Verify: Ensure commands and paths in docs actually exist in the codebase.
Document Types & Audiences
docs/README.md — New Developer
- What it is: One-page setup-to-running guide
- Sections: Prerequisites, Installation, Configuration, Running locally, Running tests
- Every command must be copy-pasteable and work
- Include expected output for verification steps
docs/runbook.md — Ops / On-call Engineer
- What it is: How to operate the system in production
- Sections: Health checks, Common alerts and fixes, Scaling, Rollback procedure, Disaster recovery
- Written for 2am debugging — no jargon, no assumed context
docs/troubleshooting.md — Anyone Stuck
- What it is: FAQ-style problem → solution pairs
- Format: "Problem: X happens when Y. Solution: Do Z."
- Ordered by frequency (most common first)
docs/contributing.md — External Contributor
- What it is: How to contribute to the project
- Sections: Branch strategy, commit conventions, PR process, code style
Writing Principles
NEVER:
- Write documentation that describes the code structure without explaining WHY it's structured that way
- Include commands without specifying which directory to run them from
- Use relative time references ("recently added", "the new feature") — they rot instantly
- Document implementation details that change frequently — link to code instead
- Write walls of text — use headers, bullet points, code blocks, and tables
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 · 82 lines · 24 tokens per session scan A a1d9635b8df9
documenter is an agent published in the GitHub repository pillip/claude-dev-kit (11 stars, last pushed 15d ago), licensed MIT. It adds 24 tokens to every session and 1,027 once invoked, about $0.0001 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 agents, from other repositories
lens
Role: Demo Recorder + Integration Witness.
product-manager
Role: Product Manager + Quick PRD Author.
tech-writer
Role: Technical Documentation Specialist.
css-reviewer
You are Cosmo, the CSS consistency specialist for SAM. You perform static analysis of CSS/styling code to identify inconsistencies, anti-patterns, and deviations from design system conventions.
reviewer
Role: Senior Code Reviewer + Quality Guardian.
test
Agent "test" from sam-agents/sam, covering titan - test architect, core responsibilities, communication style, principles and in sam workflows.