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/tonydzi/agent-leash/agents-mdgit clone --depth 1 https://github.com/tonydzi/agent-leashWrote 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/tonydzi/agent-leash/agents-md)<a href="https://agentmods.dev/instructions/tonydzi/agent-leash/agents-md"><img src="https://agentmods.dev/badge/instructions/tonydzi/agent-leash/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 | $0.00834 | $0.00834 |
| Opus 5 | $0.00417 | $0.00417 |
| Sonnet 5 | $0.00167 | $0.00167 |
| Haiku 4.5 | $0.00083 | $0.00083 |
Grade A, and why
agent-leash 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 4d 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 — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — working in this repo
Written for AI coding agents, and equally readable by a human contributor. Short on purpose.
What this repo is
LEASH-8: an 8-domain control model for AI agents that hold delegated authority. Patterns, not a product, and not our live control surfaces. There is no runnable code here — the deliverables are a scored worksheet, the control model, and two architecture notes.
If you are an agent reading this repo to apply it to your own operator's setup, start at
FOR-ROBOTS.md, then SCORECARD.md.
Layout
SCORECARD.md— 24 statements, scored 0/1/2, five minutes, produces a band.docs/leash-8.md— the eight domains, the minimal implementation of each, and what evidence to keep. The evidence column is the part that makes it auditable.docs/plan-vs-authorize.md— the core pattern: the model plans, a policy gate decides, an executor acts. Most of the repo is downstream of this separation.templates/approval-design-checklist.md— designing human approvals for irreversible actions.docs/a2a-agent-card.md+agent-card.json— reference A2A Agent Card.
How to verify a change
Nothing executes, except one file: agent-card.json must stay valid JSON and conform to the A2A
Agent Card shape described in docs/a2a-agent-card.md.
python -c "import json;json.load(open('agent-card.json'));print('agent-card.json OK')"
For everything else, verification is claim discipline (below). In the PR, say which claims you added or changed and what backs each one.
Conventions — the claim discipline
This repo's credibility is its only asset, so the rule is explicit:
- We claim: these controls reduce blast radius, raise attacker cost, and make agent actions reviewable — and we can show the implemented control, what it covers, and what stays with a human.
- We do not claim: "your agents will be secure", "prompt injection is solved", or any outcome guarantee. A PR that adds a sentence of that shape gets rejected however well written it is.
- Every external fact carries a source. Incident references, CVEs, benchmark results: link them, date them, and say what they do not show. A number without a source is the one defect that cannot be patched later — by then it has been quoted.
- Scoring must stay honest. If you change
SCORECARD.md, do not make a statement easier to score 2 on. The worksheet is useless the moment it flatters the reader. - No live control surfaces, no real infrastructure detail, no secrets. Patterns only.
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.
- 4d ago First seen · 67 lines · 834 tokens per session scan A b29deba0fe4b
agent-leash AGENTS.md is an instructions file published in the GitHub repository tonydzi/agent-leash (2 stars, last pushed 5d ago), licensed MIT. It adds 834 tokens to every session, about $0.0042 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 2026-08-31.
Other instructions, from other repositories
sint-protocol CLAUDE.md
Instructions for pshkv/sint-protocol, covering sint protocol — agent guide, what is this?, quick commands, monorepo layout and architecture rules.
sint-protocol AGENTS.md
Instructions for pshkv/sint-protocol, covering sint protocol — ai agent guide, what this repo does, key invariants (never violate these), quick orientation and package dependency graph.
Doberman-Core AGENTS.md
AGENTS.md instructions for DobermanCore/Doberman-Core, covering claude.md — doberman operating manual, 0. on startup (every session), 1. what this repository is, 2. architecture & extension points and 3. prime directives (non-negotiable).
agentfootprint AGENTS.md
AGENTS.md instructions for footprintjs/agentfootprint, covering agentfootprint — agent instructions (openai codex), core thesis, the mental model — three slots, six flavors, mock-first development (recommended workflow) and public api.
agentfootprint CLAUDE.md
Claude Code instructions for footprintjs/agentfootprint, covering agentfootprint — feature-work map, before you design it: it may already exist, module map, core state & flow and extension points.
openfirma AGENTS.md
AGENTS.md instructions for Firma-AI/openfirma, covering agents.md, key commands, formatting, linting rules and api stability.