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/drbaher/contract-ops-mcp/agents-mdgit clone --depth 1 https://github.com/DrBaher/contract-ops-mcpWrote 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/drbaher/contract-ops-mcp/agents-md)<a href="https://agentmods.dev/instructions/drbaher/contract-ops-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/drbaher/contract-ops-mcp/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 | $0.00814 | $0.00814 |
| Opus 5 | $0.00407 | $0.00407 |
| Sonnet 5 | $0.00163 | $0.00163 |
| Haiku 4.5 | $0.00081 | $0.00081 |
Grade A, and why
contract-ops-mcp 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 3d 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 — 24 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agents
contract-ops-mcp is itself an agent surface: an MCP (stdio) server that exposes the nine contract-ops CLIs as tools. This file is the contract for an agent (or its operator) driving it.
Discovery
- Call
suite_statusfirst — it reports which CLIs are installed, their versions, and the install command for any that are missing. A tool whose CLI isn't installed returns a clear install hint (not a crash). - Beyond the curated tools,
catalog(cli)returns that CLI's full<bin> --catalog json({name, bin, version, description, commands|flags, exitCodes}), andrun(cli, args)executes any command. Don't hardcode the long tail — discover it.
Output contract
- Every tool returns a single text block of JSON:
{ exitCode, result | output, stderr? }.resultis the parsed CLI JSON when the underlying command emitted JSON; otherwiseoutputis the text. - Branch on
exitCode, not on prose. Exit codes are NOT uniform across the suite — e.g.compare_versionsuses0clean /2substantive drift /3cosmetic /4moved;lint_contractuses0clean /1findings;extract_contract1= low-signal (still valid JSON). A non-zero exit is often a finding/gate result, not a failure — the JSON is still returned. Tool errors (bad input, missing CLI, path escape) setisError: true. - Abnormal termination is
exitCode: null, never a fake number. If a command is killed (180s timeout, signal, or 32 MiBmaxBufferoverflow) the result carriesexitCode: nullplus a distinct marker —timedOut: true,killed: "<signal>", ormaxBufferExceeded: true— andisError: true. Don't treat these as a real exit code (they used to collapse to1, indistinguishable from a genuine I/O error).
Safety boundaries (important)
- Signing is human-gated and cannot be performed through this server. Only read/verify ops are exposed (
verify_signature,verify_receipt,audit_show). Even therun/catalogescape hatches enforce this — asignsubcommand outside the read-only allowlist (request-create/send/sign/approve, key rotation, TSA anchoring, …) is refused, not executed. To request or apply a signature, use sign-cli's own MCP server, which gates the human gesture behind a per-signer, single-use approval token the agent never holds. - File access is confined to
CONTRACT_OPS_MCP_BASE_DIR(default: cwd). Curated tools reject paths that escape it. Therunescape hatch passes args verbatim (no shell) but is the operator's responsibility. - All CLIs are local-first and run on the host with the user's permissions — no network on the default paths (opt-in
--llmtiers excepted, via the shared~/.config/contract-ops/llm.json).
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.
- 3d ago First seen · 24 lines · 814 tokens per session scan A 7859a3dd9f16
contract-ops-mcp AGENTS.md is an instructions file published in the GitHub repository DrBaher/contract-ops-mcp (0 stars, last pushed 1mo ago), licensed MIT. It adds 814 tokens to every session, about $0.0041 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
mcp-nsa AGENTS.md
AGENTS.md instructions for matematicsolutions/mcp-nsa, covering agents.md - mcp-nsa, project goal, matematic context (hard constraints), mcp tools (tools contract) and build and test.
mcp-legal-it CLAUDE.md
Instructions for capazme/mcp-legal-it, covering mcp-legal-it — project context, confini del progetto — leggere prima, stack, struttura and tool disponibili (33 moduli, 221 tool).
mcp-saos AGENTS.md
Instructions for matematicsolutions/mcp-saos, covering agents.md - mcp-saos, project goal, matematic context (hard constraints), mcp tools (tools contract) and build and test.
mcp-krs AGENTS.md
AGENTS.md instructions for matematicsolutions/mcp-krs, covering agents.md - mcp-krs, project goal, matematic context (hard constraints), mcp tools (tools contract) and build and test.
sap-api-policy-skill CLAUDE.md
Instructions for marianfoo/sap-api-policy-skill, covering claude.md — sap-api-policy-skill, what this repo is, the one invariant — never break this, layout and editing rules.
mcp-saos CLAUDE.md
Instructions for matematicsolutions/mcp-saos: This file exists for compatibility with Claude Code. The canonical instructions for AI agents working with this repo live in AGENTS.md - the agents.md standard (Linux Foundation / Agentic AI Foundation), supported by 20+ tools.