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/amondnet/vercel-action/agents-mdgit clone --depth 1 https://github.com/amondnet/vercel-actionWhat 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 | $0.00856 | $0.00856 |
| Opus 5 | $0.00428 | $0.00428 |
| Sonnet 5 | $0.00171 | $0.00171 |
| Haiku 4.5 | $0.00086 | $0.00086 |
Grade A, and why
vercel-action 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 — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Problem definition → small, safe change → change review → refactor — repeat the loop.
Mandatory Rules
- Before changing anything, read the relevant files end to end, including all call/reference paths.
- Keep tasks, commits, and PRs small.
- If you make assumptions, record them in the Issue/PR/ADR.
- Never commit or log secrets; validate all inputs and encode/normalize outputs.
- Avoid premature abstraction and use intention-revealing names.
- Compare at least two options before deciding.
Mindset
- Think like a senior engineer.
- Don’t jump in on guesses or rush to conclusions.
- Always evaluate multiple approaches; write one line each for pros/cons/risks, then choose the simplest solution.
Code & File Reference Rules
- Read files thoroughly from start to finish (no partial reads).
- Before changing code, locate and read definitions, references, call sites, related tests, docs/config/flags.
- Do not change code without having read the entire file.
- Before modifying a symbol, run a global search to understand pre/postconditions and leave a 1–3 line impact note.
Required Coding Rules
- Before coding, write a Problem 1-Pager: Context / Problem / Goal / Non-Goals / Constraints.
- Enforce limits: file ≤ 300 LOC, function ≤ 50 LOC, parameters ≤ 5, cyclomatic complexity ≤ 10. If exceeded, split/refactor.
- Prefer explicit code; no hidden “magic.”
- Follow DRY, but avoid premature abstraction.
- Isolate side effects (I/O, network, global state) at the boundary layer.
- Catch only specific exceptions and present clear user-facing messages.
- Use structured logging and do not log sensitive data (propagate request/correlation IDs when possible).
- Account for time zones and DST.
Testing Rules
- New code requires new tests; bug fixes must include a regression test (write it to fail first).
- Tests must be deterministic and independent; replace external systems with fakes/contract tests.
- Include ≥1 happy path and ≥1 failure path in e2e tests.
- Proactively assess risks from concurrency/locks/retries (duplication, deadlocks, etc.).
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 · 85 lines · 856 tokens per session scan A 052ddd2f6ad2
vercel-action AGENTS.md is an instructions file published in the GitHub repository amondnet/vercel-action (765 stars, last pushed 1mo ago), licensed MIT. It adds 856 tokens to every session, about $0.0043 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 2026-08-30.
Other instructions, from other repositories
workflow AGENTS.md
AGENTS.md instructions for vercel/workflow, covering agent instructions, architecture, core components, workflow execution model and development commands.
run-gemini-cli GEMINI.md
Instructions for google-github-actions/run-gemini-cli, covering guidelines for developing this github action, core principles for this action and your role in development.
otter AGENTS.md
Instructions for AmadeusITGroup/otter, covering agents.md, quick start, context files, always relevant and task-specific context.
agents-shipgate CLAUDE.md
Instructions for ThreeMoonsLab/agents-shipgate, covering claude code instructions, running the cli in this repository, permissions, output handling and slash command.
Repolis AGENTS.md
Instructions for hyeonsangjeon/Repolis, covering ⏱️ 60-second quickstart (no build, no install), 🗺️ repository map (what lives where), ▶️ run & rebuild, 1) run the city (static) and 🚀 deploy.
ConsultMe CLAUDE.md
Instructions for Tarek-Bohdima/ConsultMe, covering claude.md, project context, common commands, module graph and conventions enforced by tooling.