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/avikalpa/codex-litellm/agents-mdgit clone --depth 1 https://github.com/avikalpa/codex-litellmWrote 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/avikalpa/codex-litellm/agents-md)<a href="https://agentmods.dev/instructions/avikalpa/codex-litellm/agents-md"><img src="https://agentmods.dev/badge/instructions/avikalpa/codex-litellm/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 | $0.01905 | $0.01905 |
| Opus 5 | $0.00953 | $0.00953 |
| Sonnet 5 | $0.00381 | $0.00381 |
| Haiku 4.5 | $0.00191 | $0.00191 |
Grade C, and why
codex-litellm AGENTS.md scanned grade C with 1 finding 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- `rm -rf test-workspace && ./setup-test-env.sh` How it starts
The opening of the file, as written. The whole thing — 125 lines — stays where its author put it; the contents beside it link to each section on GitHub.
codex-litellm
Purpose
codex-litellmis upstreamopenai/codexplus a maintained patchset so the CLI can work directly against a LiteLLM backend.- The goal is not to fork Codex permanently. The goal is to keep a reproducible diff that can be carried forward to newer upstream
rust-v*tags. - Build direction comes from upstream. Our job is to preserve LiteLLM compatibility, observability, and model usability without drifting from Codex more than necessary.
What Is Not Obvious
- LiteLLM compatibility is not just an endpoint swap. Different providers and models diverge on tool-calling, reasoning output, timeout behavior, usage reporting, and whether they ever emit a clean final assistant reply.
codex-litellmtherefore needs narrow runtime logic around provider setup, request shaping, tool compatibility,/modelsdiscovery, and release model curation.- The right answer is usually evidence first, patch second. Do not guess which layer is broken until telemetry or direct backend probes show it.
Non-Negotiable Rules
latest upstreammeans the latest stable upstreamrust-v*tag fromopenai/codex, excluding prereleases.- If the user asks to update to latest upstream, do not release or publish anything from an older base afterward.
- Never cut a release unless
main,package.json,package-lock.json,stable-tag.patch, and the checked-outcodex/baseline all agree on the same upstream version. build.shis release automation. Do not use it for development work insidecodex/; it resets the upstream checkout.- After every non-release milestone commit, explicitly ask the user whether to publish to npm before doing release actions.
- Exception: when the user asks to update to latest upstream, the default workstream is to smoke test, push, publish, and babysit the GitHub Actions and npm release path through completion unless the user explicitly says not to publish.
- Block release on live model failures. A local build is not enough.
- Release builds happen on GitHub Actions. Do not treat local release artifacts as publishable outputs.
- Keep the repo surface clean enough for fast pacing. Local test/build clutter should not become normal working state.
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 · 125 lines · 1,905 tokens per session scan C b3bc09f081de
codex-litellm AGENTS.md is an instructions file published in the GitHub repository avikalpa/codex-litellm (13 stars, last pushed 21d ago), licensed Apache-2.0. It adds 1,905 tokens to every session, about $0.0095 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
apm python.instructions.md
Python development guidelines.
CodexPotter AGENTS.md
AGENTS.md instructions for breezewish/CodexPotter, covering repository guidelines, workflow principles, engineering rules, core principles: simplicity & readability and better maintainability.
OpenMicro AGENTS.md
Instructions for stephenleo/OpenMicro, covering openmicro — project conventions and releases.
squarebox CLAUDE.md
Instructions for SquareWaveSystems/squarebox, covering claude.md, agent skills, issue tracker, triage labels and domain docs.
codmate AGENTS.md
Instructions for loocor/codmate, a project described as: CodMate is a macOS SwiftUI app for managing CLI AI sessions: browse, search, organize, resume, and review work produced by Codex, Claude Code, and Gemini CLI. It focuses on speed, a compact three-column UI, and “ship it” workflows like Project Review (Git…
CodexBar AGENTS.md
AGENTS.md instructions for bob-zebedy/CodexBar, covering repository guidelines, 项目结构与模块组织, 构建、测试与开发命令, 写作风格 and 代码和注释.