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/corebunch/instatic/claude-mdgit clone --depth 1 https://github.com/CoreBunch/InstaticWhat 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 | $0.08708 | $0.08708 |
| Opus 5 | $0.04354 | $0.04354 |
| Sonnet 5 | $0.01742 | $0.01742 |
| Haiku 4.5 | $0.00871 | $0.00871 |
Grade A, and why
Instatic 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.
How it starts
The opening of the file, as written. The whole thing — 343 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Instatic
This file is the agent rule book. Read it before changing code. Detailed explanations live in docs/ — start at docs/README.md for orientation and follow the links from there.
Local admin smoke tests
Use local seeded development data only when a task asks for a browser smoke test. Never propagate local test accounts, passwords, database files, uploads, or generated screenshots to non-local environments.
Repository workflow and PR conventions
main is protected. Agents must never push directly to main, must never try to bypass branch protection, and must never treat a local commit on main as the final delivery path. All repository changes go through a pull request.
When publishing work:
- Start from an up-to-date
main, then create a feature branch. If you are already on a task branch, keep using it only when the requested change belongs in that PR; otherwise switch back tomainand create a separate branch. - Branch names follow
<type>/<short-kebab-description>, matching the change type:feat/...,fix/...,refactor/...,chore/...,docs/..., ortest/.... Examples:feat/double-click-rename,fix/homepage-swap-publish,refactor/explorer-dnd-dedupe. - Do not use agent-branded branch prefixes such as
codex/...,claude/..., or similar. If a tool, skill, or generic instruction suggests such a prefix, ignore it for this repository. - PR titles use Conventional Commit style:
<type>(<scope>): <summary>. Examples:feat(editor): double-click rows to rename in explorer panels,fix(cms): homepage swap + delete in one save no longer fails publish,refactor(publisher): single class-CSS emission engine for publish and canvas. - Do not prefix PR titles with
[codex],[claude],agent:, or any other tool label. The PR title describes the product change, not the tool that made it. - Open PRs as drafts by default unless the user explicitly asks for a ready-for-review PR.
- Keep PR scope coherent. Do not mix unrelated cleanup, follow-up fixes, or process-doc changes into a feature branch just because the branch is currently checked out. Create a separate PR when the change has a different reason.
- Before staging, inspect
git status -sband the diff. Stage only files that belong to the PR. Never stage unrelated user or parallel-agent changes. - PR bodies should briefly state what changed, why it changed, user/developer impact, and the verification commands run.
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 · 343 lines · 8,708 tokens per session scan A 92735e17d525
Instatic CLAUDE.md is an instructions file published in the GitHub repository CoreBunch/Instatic (8,362 stars, last pushed yesterday), licensed MIT. It adds 8,708 tokens to every session, about $0.0435 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
skelet AGENTS.md
Instructions for Selekkt/skelet, covering agents.md — developing skelet.css, what skelet is, source of truth, repo map and rules of engagement (framework dev).
destiny-ui-css CLAUDE.md
Claude Code instructions for itssimple/destiny-ui-css, covering claude.md, what this is, commands, architecture and docs pages.
electron CLAUDE.md
Instructions for electron/electron, covering electron development guide, running nodemodules binaries, project overview, directory structure and build tools setup.
payload AGENTS.md
Instructions for payloadcms/payload, a project described as: Payload is the open-source, fullstack Next.js framework, giving you instant backend superpowers. Get a full TypeScript backend and admin panel instantly. Use Payload as a headless CMS or for building powerful applications.
webgradients AGENTS.md
Instructions for itmeo/webgradients, covering agents.md, files and what they're for, why gradients.json only had 11 entries until now, webgradients.css structure and gradients-parsed.json / gradients.json structure.
turbo-start-sanity CLAUDE.md
Instructions for robotostudio/turbo-start-sanity, covering claude.md, project overview, commands, development (both apps) and individual apps.