agent-watch AGENTS.md

agent-watch AGENTS.md is an instructions file for Codex, OpenCode from surnr/agent-watch. It costs 3,137 tokens per session, scanned A, original, MIT.

Project instructions for Agent Watch, a command-line tool that collects lessons from AI coding chats and updates shared instruction files.

In plain words
What is it for?
Use them when building, testing, or maintaining Agent Watch and its instruction-file update process.
Why use it?
They keep coding conventions and project knowledge in sync across Claude, GitHub Copilot, and other AI tools.

Instructions file for CodexOpenCode

Install

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.

agentmods
npx agentmods add instructions/surnr/agent-watch/agents-md
Clone the repo
git clone --depth 1 https://github.com/surnr/agent-watch

Made for: Codex, OpenCode.

Wrote 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.

agentmods badge for agent-watch AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/surnr/agent-watch/agents-md.svg)](https://agentmods.dev/instructions/surnr/agent-watch/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/surnr/agent-watch/agents-md"><img src="https://agentmods.dev/badge/instructions/surnr/agent-watch/agents-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 3,137 This file is loaded in full into every session.
When invoked 3,137 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.03137 $0.03137
Opus 5 $0.01569 $0.01569
Sonnet 5 $0.00627 $0.00627
Haiku 4.5 $0.00314 $0.00314

Measured 3d ago against content hash fe7fc5cedab7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-watch 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.

AGENTS.md · 381 lines

How it starts

The opening of the file, as written. The whole thing — 381 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Watch: Development Guide

This document provides guidance for AI agents working on the Agent Watch codebase. It includes project structure, code conventions, testing practices, and key architectural patterns.


Project Overview

Agent Watch is a CLI tool that automatically extracts knowledge from AI chat sessions (Claude Code, GitHub Copilot Chat, Copilot CLI) and updates agent configuration files (CLAUDE.md, AGENTS.md, .github/copilot-instructions.md) to keep them in sync with evolving codebase conventions.

Key responsibility: Keep developer knowledge about project patterns and conventions captured in version-controlled agent instruction files, accessible to all AI tools that assist the team.


Tech Stack & Build

Technologies

  • Language: TypeScript (strict mode)
  • Runtime: Node.js 20+ (ESM)
  • Package Manager: pnpm 10.8.0
  • Build System: tsdown (ESM) + tsc (type checking)
  • Formatter/Linter: Biome 1.9.4 (tabs, line width 120, no semicolons)
  • Testing: Vitest 3.1.1 with coverage
  • CLI Framework: Commander 13.1.0
  • Git Hooks: Lefthook (auto-detected, can fall back to Husky or direct hooks)

Build & Test Commands

# Build: transpile TypeScript → ESM/CJS, generate type definitions
pnpm build
# Output: dist/index.js, dist/index.cjs, dist/index.d.ts, dist/cli.js

# Run full CI pipeline (lint + type check + unit tests + publint)
pnpm test:ci

# Run just unit tests (Vitest)
pnpm test:unit

# Run linter
pnpm check

# Fix linting issues
pnpm check:fix

# Type checking
pnpm test:types

# Check for unused exports (knip)
pnpm test:unused

# Export validation (any-type-wrong)
pnpm test:exports

Development workflow:

git clone https://github.com/surnr/agent-watch.git
cd agent-watch
pnpm install
pnpm build
pnpm link --global  # Link for testing

Code Conventions & Patterns

Naming & Structure

Category Convention Examples
Files kebab-case claude-code.ts, copilot-cli.ts, session-state.ts
Variables/Functions camelCase processAllSessions, enabledTools, summarizeSession
Types/Interfaces PascalCase SessionExtractor, AgentWatchConfig, AgentFilePattern
Constants UPPER_SNAKE_CASE SUPPORTED_AI_AGENTS, KNOWN_AGENT_FILES, CONFIG_FILE_NAME

Read the full file on GitHub · 381 lines

Changes

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.

  1. 3d ago First seen · 381 lines · 3,137 tokens per session scan A fe7fc5cedab7

Subscribe to this mod's changes

agent-watch AGENTS.md is an instructions file published in the GitHub repository surnr/agent-watch (2 stars, last pushed 6mo ago), licensed MIT. It adds 3,137 tokens to every session, about $0.0157 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.