odin-codex-plugin AGENTS.md

A set of instructions for using the OutlineDriven/odin-codex-plugin project with a coding agent.

In plain words
What is it for?
Use it when making or reviewing changes in that project, especially when you need its required workflow, file-operation rules, or English-only responses.
Why use it?
It gives the agent consistent rules for planning changes, searching files, using command-line tools, and keeping work within scope.

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/outlinedriven/odin-codex-plugin/agents-md
Clone the repo
git clone --depth 1 https://github.com/OutlineDriven/odin-codex-plugin

Made for: Codex, OpenCode.

Per session 10,608 This file is loaded in full into every session.
When invoked 10,608 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 $0.10608 $0.10608
Opus 5 $0.05304 $0.05304
Sonnet 5 $0.02122 $0.02122
Haiku 4.5 $0.01061 $0.01061

Measured 2d ago against content hash fd6ec20ef703, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

odin-codex-plugin 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.

AGENTS.md · 390 lines

How it starts

The opening of the file, as written. The whole thing — 390 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Codex Coding Agent Adherences

Method: principle-first minimalism (delete > edit > add), data-first design, plan-before-change, ask-with-evidence, delegate intentionally, verify continuously, scope discipline, simplicity bias, workspace hygiene (.outline/, /tmp).

Language [MANDATORY—HARD ENFORCEMENT]: ALWAYS think, reason, act, and respond in English regardless of user's language. Translate ALL non-English inputs to English BEFORE reasoning or acting. No exceptions — internal reasoning, code comments, commit messages, documentation, agent communication, tool output interpretation: ALL must be English. May write multilingual docs ONLY when explicitly and specifically requested by the user. Violation = CRITICAL FAILURE; formal-logic reasoning uses ASCII operators only — connectives ! & | ^ -> <->, quantifiers forall exists exists!, turnstiles |- |=, relations = != < > <= >= = :=, set ops in notin subset subseteq union intersect \ empty, type/lambda \x. : :: |-> -> <:, proof/inference => :. s.t. iff QED induction, modal/temporal [] <> G F X U R W A E | — not Unicode glyphs; source code syntax, identifiers, and locale-specific strings follow the task's target language — code comments, non-locale documentation, and commit messages remain English unless multilingual docs are explicitly requested.

<verbalized_sampling> Sample multiple intent hypotheses, weight each (0–1), and name the falsifier per hypothesis. Scale depth to ambiguity/risk; broaden until edge cases stop changing the decision. Synthesize surviving hypotheses into one direction. Output: intent summary, assumptions, focused questions. No non-trivial change without visible VS. </verbalized_sampling>

<tool_selection> Tool Selection Matrix: ast-grep → code structure/refactoring/bulk transforms | srgn → grammar-scoped regex replacement | ripgrep → text/comments/strings/non-code | awk → column extraction/line ranges/text regex | tokei → scope assessment | Combined → multi-stage via fd/rg/xargs pipelines Scope (tokei-driven): Run tokei <target> --output json | jq '.Total.code' before editing. Micro (<500 LOC): Direct edit/single-file focus/minimal verification | Small (500-2K LOC): Progressive refinement/2-3 file scope/standard verification | Medium (2K-10K LOC): Multi-agent parallel/dependency mapping/staged rollout | Large (10K-50K LOC): Research-first/architecture review/incremental checkpoints | Massive (>50K LOC): Decompose to subsystems/formal planning/multi-phase execution </tool_selection>

Query Language (Revsets):

  • Draft/Stack: draft() | stack() | branches()
  • Author/Message: author.name("Alice") | message("fix bug")
  • Paths: paths.changed("src/*.rs")
  • Relations: ancestors(<rev>) | descendants(<rev>) | children(<rev>) | parents(<rev>)
  • Operations: <set1> | <set2> (union) | <set1> & <set2> (intersection) | <set1> - <set2> (difference) | <set1> % <set2> (only)
  • Tests: tests.passed() | tests.failed("<cmd>")
  • Shortcuts: :<rev> (ancestors) | <rev>: (descendants)
  • Usage: git query '<revset>' | git smartlog '<revset>' | git sync '<revset>'

Read the full file on GitHub · 390 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. 2d ago First seen · 390 lines · 10,608 tokens per session scan A fd6ec20ef703

Subscribe to this mod's changes

odin-codex-plugin AGENTS.md is an instructions file published in the GitHub repository OutlineDriven/odin-codex-plugin (15 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 10,608 tokens to every session, about $0.0530 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.