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/vincentkoc/agentrinse/agents-mdgit clone --depth 1 https://github.com/vincentkoc/agentrinseWhat 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 | $0.00356 | $0.00356 |
| Opus 5 | $0.00178 | $0.00178 |
| Sonnet 5 | $0.00071 | $0.00071 |
| Haiku 4.5 | $0.00036 | $0.00036 |
Grade A, and why
agentrinse 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 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.
What it actually says
AGENTS.md
Product Boundary
AgentRinse is a safety-critical local cleanup planner. Unknown state is protected. A missing path, old timestamp, or large directory is never enough evidence for deletion.
Hard Safety Rules
- Never run AgentRinse against the developer's real home during development.
- Never point tests at
$HOME,/, a repository parent, or a non-temporary directory. - Use synthetic fixtures under a test-owned temporary root.
- Do not add a mutating action without plan, revalidation, journal, and recovery tests.
- Do not kill Codex, Claude, Cursor, Copilot, Zed, OpenCode, Grok, Docker, tmux, terminal, SSH, or test processes.
- Do not delete transcripts, session databases, credentials, configuration, skills, memories, plugins, Docker volumes, Git branches, or stashes.
- Do not follow symlinks during size measurement or cleanup.
- Do not execute project-provided hooks or configuration as code.
- Never add a generic
--forceoverride.
Engineering
- TypeScript ESM with exact
.jssuffixes in relative imports. - Keep collectors read-only.
- Keep policy separate from discovery.
- Machine output is a versioned contract.
- Human output must not be parsed by other modules.
- Use owner APIs and structured output before filesystem inference.
- Tests mirror source ownership and use explicit fake homes.
- Prefer one focused commit per contract, adapter, or behavior.
Verification
Before commit:
pnpm format:check
pnpm lint
pnpm typecheck
pnpm test
Before publish:
pnpm pack:check
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 · 52 lines · 356 tokens per session scan A 09164268a8e4
agentrinse AGENTS.md is an instructions file published in the GitHub repository vincentkoc/agentrinse (20 stars, last pushed 8d ago), licensed MIT. It adds 356 tokens to every session, about $0.0018 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
TreeMap-Disk-Visualizer AGENTS.md
Instructions for Prithvi-Web/TreeMap-Disk-Visualizer, covering treemap for agents, two ways in, the core workflow: scan → inspect → dry-run → act, the safety model (enforced server-side, not advisory) and mcp specifics.
claudette CLAUDE.md
Instructions for utensils/claudette, covering claude.md, claudette, ai assistant configuration, documentation discipline and build & test commands.
codehydra CLAUDE.md
Instructions for stefanhoelzl/codehydra, covering claude.md, critical rules, no ignore comments, api/ipc interface changes and new boundary interfaces.
claudette copilot-instructions.md
Instructions for utensils/claudette: Claudette is a cross-platform desktop orchestrator for parallel Claude Code agents. It uses a Tauri 2 Rust backend, a React/TypeScript frontend, SQLite via rusqlite, Tokio process management, git worktrees, xterm.js terminals, Lua plugins, and a local CLI that talks to the running…
claudette rust.instructions.md
Instructions for utensils/claudette: Rust code uses edition 2024, default rustfmt, and clippy with warnings denied in CI. Prefer clear ownership, explicit error handling, and small focused modules over large multi-purpose functions.
claudette frontend.instructions.md
Instructions for utensils/claudette: Frontend code is React, TypeScript strict mode, Zustand, CSS modules, Vite, Bun, and Vitest. Do not use any; read the actual type definitions before constructing fixtures or Tauri payloads.