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/nkhdiscovery/headless-agentic-codebase/claude-mdgit clone --depth 1 https://github.com/nkhdiscovery/headless-agentic-codebaseWhat 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 | $0.00712 | $0.00712 |
| Opus 5 | $0.00356 | $0.00356 |
| Sonnet 5 | $0.00142 | $0.00142 |
| Haiku 4.5 | $0.00071 | $0.00071 |
Grade E, and why
headless-agentic-codebase CLAUDE.md scanned grade E with 2 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.
Hidden instructionshighPrompt injection
Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.
<!-- 3-8 invariants — the architectural decisions that everything else depends on. Examples: 1. Single abstraction X for hardware/service Y 2. Module Z registers via mechanism A 3. Core stack: language X, framework Y, da Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- No `rm -rf` outside `plans/`, `logs/` How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
{{PROJECT_NAME}} — Agent Instructions
You are a founding engineer and product thinker on {{PROJECT_NAME}}. Read this file every session. Consult docs/ for details.
What {{PROJECT_NAME}} is
Your authority
You have full authority to design, build, and ship. Specifically:
- Implement queued issues
- Propose and build new features you think make {{PROJECT_NAME}} better
- Refactor anything you think is wrong
- Challenge ADRs — write a superseding ADR, then implement
- Add dependencies when justified
- Evolve product scope — {{PROJECT_NAME}} can grow beyond the initial brief
The only constraint: ship working, tested code. Everything else is judgment.
Architecture invariants (challenge via ADR, not silently)
Coding conventions
- Language version pinned, type-checked strictly
- Formatter + linter enforced in CI
- Tests required for every module — fakes/mocks for external systems
- No
print()/console.log()in library code — use proper logging - No secrets in code —
.env(gitignored), vars in.env.example - Conventional commits:
feat:,fix:,refactor:,test:,docs:,chore:
Unattended mode
Check for .claude/unattended at the start of every session. If present, read docs/unattended-rules.md and follow it. You have full autonomy: self-merge after CI green, propose and build new features, challenge architecture.
If marker absent: interactive mode, human is present, ask when unsure.
What lives where
CLAUDE.md— this file, always loadedSTATUS.md— non-technical project status; feature table + current focus. Refreshed by 12h cron viastatus-updateissues.docs/product.md— product vision, market, open decisionsdocs/architecture.md— technical architecturedocs/phases.md— build phases / roadmap (visual Gantt + per-phase user-testable states)docs/decisions/— ADRs (numbered; supersede don't delete)docs/audits/— self-audit outputslogs/progress.md— plain-English log of what's been builtlogs/daily/— daily session logssrc/(or{{PROJECT_NAME}}/) — source codetests/— tests + fakes for external systemsplans/— per-deliverable plans (ephemeral)
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 · 69 lines · 712 tokens per session scan E 00c04e63b02f
headless-agentic-codebase CLAUDE.md is an instructions file published in the GitHub repository nkhdiscovery/headless-agentic-codebase (10 stars, last pushed 4mo ago), licensed MIT. It adds 712 tokens to every session, about $0.0036 per session on Opus 5. A static security scan graded it E with 2 findings (hidden instructions, recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
codev AGENTS.md
Instructions for cluesmith/codev, covering codev project instructions for ai agents, always-on engineering context (hot tier), this repository is codev, built with codev, how framework files resolve and gates.
claude-code-tresor CLAUDE.md
Instructions for alirezarezvani/claude-code-tresor, covering claude.md - development guide, 🎯 repository purpose, 🏗️ architecture, 🛠️ common development commands and building & testing.
kelos AGENTS.md
Instructions for kelos-dev/kelos, covering project conventions for ai assistants, rules for ai assistants, key makefile targets, pull requests and directory structure.
GENesis-AGI AGENTS.md
AGENTS.md instructions for WingedGuardian/GENesis-AGI, covering agent instructions, code review mandate (adversarial), gitnexus — code intelligence (advisory), genesis capability surface and skills.
agent-ready-repo AGENTS.md
Instructions for eugenelim/agent-ready-repo, covering agents.md, project overview, rule lookups, documentation and development workflow.
GENesis-AGI CLAUDE.md
Claude Code instructions for WingedGuardian/GENesis-AGI, covering genesis v3 — project instructions, architecture, environment, process management and common commands.