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/igorwarzocha/howaboua-pi-stuff/agents-mdgit clone --depth 1 https://github.com/IgorWarzocha/howaboua-pi-stuffWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/igorwarzocha/howaboua-pi-stuff/agents-md)<a href="https://agentmods.dev/instructions/igorwarzocha/howaboua-pi-stuff/agents-md"><img src="https://agentmods.dev/badge/instructions/igorwarzocha/howaboua-pi-stuff/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00704 | $0.00704 |
| Opus 5 | $0.00352 | $0.00352 |
| Sonnet 5 | $0.00141 | $0.00141 |
| Haiku 4.5 | $0.00070 | $0.00070 |
Grade A, and why
howaboua-pi-stuff 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 today.
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.
What it actually says
This repo publishes through Changesets; every merge to main feeds the version and npm publish workflow.
- Resolve package names by matching their words against immediate subdirectories of packages; search the unique match first and follow direct references only.
- Keep the vendored
src/codex-runtimetrees in pi-codex-web-run and pi-codex-imagegen identical; do not publish a shared runtime package. - Agent-facing text is behavior: keep tool contracts, skill files, prompt metadata, and subagent prompts compact.
- Measure package-emitted tool schemas and system-prompt deltas for cache impact; repository
AGENTS.mdis internal, not product prompt cost. - Do not repeat self-evident contracts across names, descriptions, schemas,
promptSnippet, orpromptGuidelines; use the latter two only for concrete failures. Never rewrite prior tool calls/results to integrate a tool. - Agent-facing prose need not perform grammatical polish; optimize semantic signal per token and omit cosmetic punctuation when it saves tokens. Preserve syntax, structural delimiters, meaning, evidence, caveats, and recovery instructions.
- Contract spine, not feature museum: feature-existence and regression-tour tests die; retain only independent protocol, routing, migration, or model-visible contracts.
- When review questions test scope, cull first: delete whole cases or narrow to the minimum independent contract. Never increase permanent test count unless the user explicitly requests more coverage.
- Never encode agent tool-call mistakes or prompt-following failures as programmatic tests. Discover them in real use and fix the model-facing contract; tests may cover only deterministic parser, executor, result, or routing boundaries independently of model compliance.
- Skills and extensions must work for any user. Never ship local paths, personal names, machine assumptions, or private workflow details.
- Slash commands are for users; agents use tools. Prefer one routed entry command over several command names unless explicitly requested.
- Treat related package work from one session as one release unit: one PR or one directly, atomically merged stack. Installed users should not absorb serial cleanup releases.
- Changed published behavior or payload requires a changeset; ordinary docs and tests do not. Use concrete release language; never write “upcoming release”, “unreleased”, or speculative notes.
- Changeset bodies become changelog copy: lead with one concise outcome; for broad releases add short user-facing capability bullets. Never dump implementation, tests, review history, or a multi-feature release into one prose lump.
- Before a
dev→mainPR, fetch/prune, resetdevontoorigin/main, then cherry-pick only intended commits. Never mergemainintodev. bun run check:changedselects committed branch changes, not working edits; use focused package checks while editing. Prefer patch-autodetectingbun changeset -- "summary"; usebun changeset:rawonly for intentional non-patch bumps.- After behavior-preserving module splits, run
bun refactor:compare --base <pre-refactor> --entry <old path> --probe <ESM>; the probe calls old and replacement modules with identical inputs, asserts normalized outputs, and expands until no reachable old-side paths remain unexplained. Coverage is traversal, not proof; retain probes only for durable independent contracts. - Do not bump aggregate package versions or add their changesets manually; CI runs
bun run changeset:aggregates.
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.
- today Changed · +23 tokens per session b7764eea886c
- yesterday Changed · +1 lines · +37 tokens per session 938c5cdaa5d6
- 5d ago First seen · 20 lines · 644 tokens per session scan A dc79521566cf
howaboua-pi-stuff AGENTS.md is an instructions file published in the GitHub repository IgorWarzocha/howaboua-pi-stuff (346 stars, last pushed today), licensed MIT. It adds 704 tokens to every session, about $0.0035 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 instructions, from other repositories
planning-with-files AGENTS.md
AGENTS.md instructions for OthmanAdi/planning-with-files, covering agents.md — planning-with-files agent reference card, commit rules, release checklist (12 steps), version bump scope and changelog format.
openrouter-mcp-multimodal AGENTS.md
AGENTS.md instructions for stabgan/openrouter-mcp-multimodal, covering agent instructions, before you ship, releasing (read this before publishing), short version and version files (must all match package.json).
OpenMicro AGENTS.md
Instructions for stephenleo/OpenMicro, covering openmicro — project conventions and releases.
ai-usagebar CLAUDE.md
Claude Code instructions for akitaonrails/ai-usagebar, covering claude.md, release checklist — must do all of these, hard invariants — never break these, secret-discipline rules (learned the hard way) and live api smoke discipline.
living-manual CLAUDE.md
Instructions for dougstanford/living-manual, covering living-manual, start of every session, how work lands — non-negotiable, user's manual (living-manual) and releasing.
yandex360-mcp CLAUDE.md
Claude Code instructions for pa1ch/yandex360-mcp, covering claude.md — гайд для ассистента, как устроен релиз (всё автоматически по тегу), процедура релиза (шаги ассистента), гарды and архитектура (кратко).