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/boazbk/codex_assistant/agents-mdgit clone --depth 1 https://github.com/boazbk/codex_assistantWhat 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 | $0.00547 | $0.00547 |
| Opus 5 | $0.00273 | $0.00273 |
| Sonnet 5 | $0.00109 | $0.00109 |
| Haiku 4.5 | $0.00055 | $0.00055 |
Grade A, and why
codex_assistant 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 2d 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.
What it actually says
Email Handling Instructions
- For any email task in this repo, first read and follow
skills/email-attention-flow/SKILL.md. Also useemail_templates.mdandlog.mdwhen relevant. - If the user asks you to scan for emails that are meant for them, be diligent about checking all relevant addresses and aliases. Surface messages that are personal to the user or clearly require action, and filter out marketing, list mail, and generic announcements unless the user asks for them.
- Do not surface generic emails that are not personal to the user or do not require action unless the user explicitly asks for them.
- Filter out emails that
log.mdshows were already handled. If the user tells you to mark off an email, record it inlog.mdand do not surface it again. If you send an email for the user, add it tolog.md. - By default, if the user asks you to respond, draft, or reply and has not explicitly authorized immediate sending, do this for each email:
- show an
EMAILblock with the relevant incoming email - show a
DRAFTblock with the proposed response - ask for confirmation before sending
- instructions such as
say,reply,respond,forward,decline,tell them, orcome up with a responseare drafting instructions only and are not permission to send - only explicit send authorization such as
send,email them,forward it now,do it, orsend all standard templatesallows immediate sending - if a mixed request contains some items with explicit send authorization and others without it, send only the explicitly authorized subset and draft the rest
- show an
- If the initial scan finds emails that match one of the categories in
email_templates.md, group them together and propose using the standard template. If the user tells you to do that, send those standard-template replies directly; for those cases you do not need to show drafts first. - If you send emails, always include the Gmail link to each email you sent.
- For recall-sensitive scans such as "surface all emails requiring my attention from the last X days/weeks", do not rely on a single Gmail search. Use a recall-first process that checks all relevant addresses, covers the whole timeframe, verifies against
log.md, and explicitly calls out encrypted or ambiguous personal emails instead of silently dropping them. - If the user responds only about some of the surfaced emails, do not lose track of the others. Keep a visible residual queue of actionable emails until they are handled, deferred, or marked off.
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.
- 2d ago First seen · 18 lines · 547 tokens per session scan A ac7ea16c42d9
codex_assistant AGENTS.md is an instructions file published in the GitHub repository boazbk/codex_assistant (5 stars, last pushed 4mo ago), licensed MIT. It adds 547 tokens to every session, about $0.0027 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
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
next.js AGENTS.md
Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.