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/cunhaax/ai-workflow/agents-mdgit clone --depth 1 https://github.com/cunhaax/ai-workflowWrote 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/cunhaax/ai-workflow/agents-md)<a href="https://agentmods.dev/instructions/cunhaax/ai-workflow/agents-md"><img src="https://agentmods.dev/badge/instructions/cunhaax/ai-workflow/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 | $0.02602 | $0.02602 |
| Opus 5 | $0.01301 | $0.01301 |
| Sonnet 5 | $0.00520 | $0.00520 |
| Haiku 4.5 | $0.00260 | $0.00260 |
Grade A, and why
ai-workflow 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 — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Workflow Template
Guidance for AI coding agents working in this repository. This file
(AGENTS.md) is the canonical copy; CLAUDE.md imports it via @AGENTS.md so
Claude Code (and its sub-agents) load it, while other tools that read
AGENTS.md directly still see one source of truth. Keep it lean — as the
instruction count grows, instruction-following quality degrades across all
instructions. Deep detail belongs in docs/ and the code — this file is the
map, not the territory.
Rules — non-negotiable
These apply to every agent — the main agent and any sub-agents it invokes. Nothing enforces them automatically; following them is your responsibility.
-
Use the project's canonical commands. Build, run, stop, and test only through the documented commands (see Commands). If a wrapper exists (
make/just/an npm script/…), do not invoke the underlying build tool directly — the wrapper may add serialisation, environment setup, or safety the raw tool skips. If a needed capability has no command, STOP and report the gap (Rule 2) rather than substituting raw shell. -
If a command or tool fails — or succeeds without doing what it should — STOP and report it; never silently work around it. Give the user the exact step (command or tool call) and its full output, then wait. Do not retry with different flags/ports/tools or skip the step. Surfacing the problem is expected, not a failure on your part.
-
Stay on the branch you were started on. Commit directly onto the current branch. Never create, switch, or check out branches, and never
git worktree add(restoring a file withgit checkout -- <path>is fine). If you believe the branch must change, STOP and ask first. Feature work assumes the human starts the session on a fresh feature branch — the code review diffs against the default branch and the PR targets it, neither of which works from the default branch itself. If you start on the default branch, STOP and ask before planning.
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 · 188 lines · 2,602 tokens per session scan A 3a60623815ce
ai-workflow AGENTS.md is an instructions file published in the GitHub repository cunhaax/ai-workflow (5 stars, last pushed 8d ago), licensed MIT. It adds 2,602 tokens to every session, about $0.0130 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
Skill_Seekers CLAUDE.md
Instructions for yusufkaraaslan/Skill_Seekers, covering claude.md, project overview, essential commands, required before running tests or cli (src/ layout) and run all tests (never skip - all must pass before commits).
octocode CLAUDE.md
Claude Code instructions for bgauryy/octocode: Read AGENTS.md. Discover repository skills under .agents/skills and activate only those matching the task.
che-ical-mcp CLAUDE.md
Claude Code instructions for PsychQuant/che-ical-mcp, covering spectra instructions, use /spectra: skills when, workflow, parked changes and mcp 功能驗證的表面選擇（.claude/rules/mcp-binary-version-testing.md）.
substack-mcp-plus CLAUDE.md
Instructions for ty13r/substack-mcp-plus, covering claude.md, quick command reference (use these!), always activate virtual environment first, run tests (use python3!) and code quality.
claude-code-skill-factory CLAUDE.md
Instructions for alirezarezvani/claude-code-skill-factory, covering claude.md - skills factory orchestration, repository purpose, modular claude.md architecture, navigation map and repository structure.
access-to CLAUDE.md
Claude Code instructions for dougdevitre/access-to, covering claude.md — instructions for claude code, project overview, repository structure, commands and validate config files.