codex

A structured research-and-improvement process for Codex projects, added through an AGENTS.md instruction file. It supports narrow, broad, and sweep modes with session logs.

In plain words
What is it for?
Use it to investigate a specific target, test several possible improvements on separate branches, or run a general code-quality improvement loop.
Why use it?
It gives open-ended coding work a recorded goal, measurable checks, and a repeatable sequence, so progress can be resumed and assessed.

Agent

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 agents/ehmo/autoresearch-skill/codex
Clone the repo
git clone --depth 1 https://github.com/ehmo/autoresearch-skill
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,554 The whole file, excluding the scripts and references it only reads on demand.
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 $0.00000 $0.01554
Opus 5 $0.00000 $0.00777
Sonnet 5 $0.00000 $0.00311
Haiku 4.5 $0.00000 $0.00155

Measured yesterday against content hash 5004fbd3795c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

codex 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 yesterday.

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/codex.md · 163 lines

How it starts

The opening of the file, as written. The whole thing — 163 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Autoresearch for Codex

Add this to your project's AGENTS.md or equivalent instruction file.

Current protocol version: 2.0.0 (semver). Record this version in every session log you create so a resumed run can detect protocol changes. Changelog at skills/autoresearch/CHANGELOG.md in the autoresearch repo.

Note: Codex runs as a single agent, so the clean-room separation between phases is weaker than in multi-agent setups like Claude Code. You'll have full context of what you found when you start fixing things. The protocol still works because the phased approach forces structured thinking, but the independent-perspective benefit is reduced.


Three modes

Pick one before starting:

  • narrow — the user has a specific measurable goal. Work ranked angles in order.
  • broad — the user has an aspiration. Generate 3–5 diverse hypotheses, run each on its own branch.
  • sweep — general quality improvement loop (legacy default).

Record the mode in a session log so a resumed run knows which path to take.


Narrow mode protocol

Goal Gate (before anything else)

Collect and record:

  1. Metric name
  2. Measurement command (prints a parseable value)
  3. Baseline value (run the command, record the result)
  4. Target value with comparator (≥, ≤, =)

Refuse to start cycles without all four.

Collect angles (strategies to hit the goal). Each angle is a name + one-sentence hypothesis. If more than one angle is proposed, get an explicit priority ranking (1, 2, 3) from the user before starting. Do not guess.

Execution

Work angle 1 first. Run the cycle below repeatedly on angle 1 until:

  • Goal met (measurement ≥/≤/= target) → stop the whole session
  • Angle exhausted (two consecutive cycles with zero productive commits) → advance to angle 2
  • Three consecutive regressions → revert each, mark angle exhausted, advance

Then angle 2, then angle 3.

Cycle (on current angle)

  1. Find problems scoped to the current angle's hypothesis. Skip unrelated issues; save them for later in an ideas.md backlog.
  2. Fix one at a time. Test. Commit (fix(narrow/a<N>): ...) or revert.
  3. Re-run the measurement command. Record the value. Stop if target met.

Read the full file on GitHub · 163 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. yesterday First seen · 163 lines · 0 tokens per session scan A 5004fbd3795c

Subscribe to this mod's changes

codex is an agent published in the GitHub repository ehmo/autoresearch-skill (53 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,554 tokens. 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.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens