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/barleviatias/toolkit-ai/agents-mdgit clone --depth 1 https://github.com/barleviatias/toolkit-aiWrote 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/barleviatias/toolkit-ai/agents-md)<a href="https://agentmods.dev/instructions/barleviatias/toolkit-ai/agents-md"><img src="https://agentmods.dev/badge/instructions/barleviatias/toolkit-ai/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 | $0.01961 | $0.01961 |
| Opus 5 | $0.00981 | $0.00981 |
| Sonnet 5 | $0.00392 | $0.00392 |
| Haiku 4.5 | $0.00196 | $0.00196 |
Grade C, and why
toolkit-ai AGENTS.md scanned grade C 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 6d 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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
The install policy is **alert, never block** — scanner findings surface to the user (TUI consent dialog, CLI log output) but do not refuse the install. The CLI `--strict` flag (`InstallOptions.strict`) opts in to hard-fa Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
- **No shell strings** — use `spawnSync(bin, [args], ...)` with array arguments; never `{shell: true}` How it starts
The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Guidelines
Conventions for contributors and AI coding agents (Claude Code, Codex, Copilot, Cursor). Pairs with CLAUDE.md (architecture) and README.md (user docs).
Project Structure & Module Organization
src/ contains all application code for the CLI and Ink TUI:
src/commands/— command entrypoints (headless.ts,init.ts)src/core/— install / update / remove / catalog / scanner / sources / platform logicsrc/components/— shared UI pieces (ItemRow,ItemList,DetailView,TypeFilter, …)src/tabs/— top-level TUI screens (CatalogTab,InstalledTab,SourcesTab)src/hooks/— React hooks (useCatalog,useFilteredItems)
Static data lives in resources/ (default sources manifest); repo-level utilities live in scripts/. Build output is written to bin/ai-toolkit.mjs — do not hand-edit generated files.
Build, Test, and Development Commands
| Command | Purpose |
|---|---|
npm install |
Install dev dependencies (ink, react, tsup, typescript) |
npm run build |
Production bundle into bin/ai-toolkit.mjs (publish only) |
npm run build:dev |
Local verification build — stamps a git build number into the UI / --version; use this to verify changes manually |
npm run dev |
tsup --watch for iterative development |
npm test |
tsc --noEmit typecheck, then run all tests/*.test.mjs via node:test |
npm link |
Link the CLI globally for local testing (toolkit, ai-toolkit, toolkit-ai) |
Tests are authored against the compiled build in .test-dist/ (see tests/run.mjs). Fixtures live in tests/fixtures/*.mjs and communicate results as JSON via stdout.
Manual / TUI verification must use npm run build:dev, never npm run build. The dev build stamps a build number (git short SHA, +-dirty) into the Logo and --version, e.g. toolkit-ai v2.1.13 dev build · adb1edd. This matters because the global toolkit command may resolve to the published npm copy in node_modules, not your repo build — both print the same version, so a stale global silently runs old code. Run npm link first so the global commands point at the repo build, then confirm the build number in the UI before trusting any manual verification.
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.
- 6d ago First seen · 124 lines · 1,961 tokens per session scan C 328554d9aef5
toolkit-ai AGENTS.md is an instructions file published in the GitHub repository barleviatias/toolkit-ai (12 stars, last pushed 5d ago), licensed MIT. It adds 1,961 tokens to every session, about $0.0098 per session on Opus 5. A static security scan graded it C with 2 findings (tells the agent never to refuse, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
eigent AGENTS.md
AGENTS.md instructions for eigent-ai/eigent, covering eigent agent instructions, ui work, product terminology and validation.
LeAgent AGENTS.md
Instructions for vixues/LeAgent, covering leagent development guidelines, project overview, architecture, execution topology (one kernel, many ingresses) and surface map (where things live).
pm-pilot CLAUDE.md
Claude Code instructions for mshadmanrahman/pm-pilot, covering pm pilot, skill loading, context flywheel, session protocol and agent dispatch.
vant AGENTS.md
AGENTS.md instructions for dhaupin/vant, covering agents.md — vant agent guide, what is vant?, your job, brain router interface and middleware chain.
claude-sdlc CLAUDE.md
Claude Code instructions for lantisprime/claude-sdlc, covering claude.md, what this repo is, design intent — read this before "improving" anything, eat your own dog food and hook strictness philosophy.
caddis-plugin fastapi.instructions.md
FastAPI development standards for backend services.