autoresearch

autoresearch is a skill for Claude Code, Codex from darkroomengineering/cc-settings. It costs 50 tokens per session (4,337 once invoked), scanned A, original, MIT.

An experimental loop for changing and scoring a skill’s SKILL.md instructions, keeping better versions and discarding worse ones. It is designed for a separate Claude-based setup and is unsupported in standalone Codex.

In plain words
What is it for?
Use it only in the supported Claude workflow to optimize a skill’s instructions against a defined checklist or score.
Why use it?
It provides a repeatable way to test prompt changes, but cannot be used as intended in this environment.

Skill for Claude CodeCodex

Part of the darkroom plugin — 37 skills, 10 agents, 4 hooks, 1 MCP server shipped together

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 skills/darkroomengineering/cc-settings/autoresearch
Any agent
npx skills add darkroomengineering/cc-settings --skill autoresearch
Clone the repo
git clone --depth 1 https://github.com/darkroomengineering/cc-settings

Made for: Claude Code, Codex.

Or install darkroom, the plugin that ships this one along with the rest of its 37 skills, 10 agents, 4 hooks, 1 MCP server.

Wrote 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.

agentmods badge for autoresearch

README.md
[![agentmods](https://agentmods.dev/badge/skills/darkroomengineering/cc-settings/autoresearch.svg)](https://agentmods.dev/skills/darkroomengineering/cc-settings/autoresearch)
Your own site
<a href="https://agentmods.dev/skills/darkroomengineering/cc-settings/autoresearch"><img src="https://agentmods.dev/badge/skills/darkroomengineering/cc-settings/autoresearch.svg" alt="Measured on agentmods" height="20"></a>
Per session 50 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,337 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 $0.00050 $0.04337
Opus 5 $0.00025 $0.02168
Sonnet 5 $0.00010 $0.00867
Haiku 4.5 $0.00005 $0.00434

Measured today against content hash 2e19fbc20755, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

autoresearch 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 today.

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.

skills/autoresearch/SKILL.md · 414 lines

How it starts

The opening of the file, as written. The whole thing — 414 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AutoResearch

Standalone Codex: unsupported

Stop here in standalone Codex. This workflow depends on a measured Claude subprocess isolation and model-pinning loop. Do not substitute native agents or claim comparable scores until a Codex equivalent has its own controlled, repeatable measurements. The remaining workflow is Claude-only.

Autonomous skill optimization. You modify a skill's prompt, test it, keep improvements, revert failures. Repeat forever.

Adapted from Karpathy's autoresearch. Same method: single editable file, single metric, git-based keep/revert, autonomous loop. The only difference: SKILL.md replaces train.py, checklist pass rate replaces val_bpb.

NEVER STOP. Once the loop begins, do NOT pause to ask the human if you should continue. The human might be away and expects you to work indefinitely until manually interrupted. If you run out of ideas, think harder — re-read failing outputs, try combining near-misses, try more radical prompt rewrites. The loop runs until the human interrupts you, period.


Setup

Work with the user to configure, then go autonomous.

  1. Parse target skill: Get <skill-name> from $ARGUMENTS. Validate skills/<skill-name>/SKILL.md exists.

  2. Load or create RESEARCH.md: Check for skills/<skill-name>/RESEARCH.md. If it exists, read it — a skill born from /harvest arrives with a seeded RESEARCH.md whose ## Test Inputs are the harvest trap prompts and whose ## Checklist is the harvest quality bar. If not, generate one:

    • Read the target SKILL.md
    • Derive 3 test inputs from its description and use cases
    • Derive 5-7 checklist items from its workflow steps and output format
    • Write the generated RESEARCH.md and show it to the user for confirmation

    Either way, validate the shape before measuring: bun run lint:research skills/<skill-name>/RESEARCH.md (required sections present, ≥2 test inputs, 3-7 checklist items, numeric settings). A seed that fails this parses wrong in the loop below.

Read the full file on GitHub · 414 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. today Changed · +1 lines 2e19fbc20755
  2. 4d ago First seen · 413 lines · 50 tokens per session scan A 06d32729b85b

Subscribe to this mod's changes

autoresearch is a skill published in the GitHub repository darkroomengineering/cc-settings (43 stars, last pushed 2d ago), licensed MIT. It adds 50 tokens to every session and 4,337 once invoked, about $0.0003 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.

Related

Other skills, from other repositories

merge-pr

Merge a pull request with status checks, squash merge, and branch cleanup. Handles worktree contexts.

joshukraine/dotfiles · 23 tokens

po-translate

Orchestrate English→Japanese translation of po/ja.po — classify, delegate translation/review to subagents, iterate until clean.

natsukium/dotfiles · 30 tokens

gui-debug

Verify a rendering or window-chrome change in any macOS GUI app unattended — find its CGWindowID via JXA, capture that window alone (with per-pixel alpha) using screencapture -l, and read exact RGBA via NSBitmapImageRep. Use for blur/opacity/corner-radius/glyph/color rendering bugs where a screenshot or pixel value is…

natsukium/dotfiles · 89 tokens

bootstrap-prd

Set up PRD-driven development infrastructure for a new project, including directory structure, templates, and roadmap.

joshukraine/dotfiles · 25 tokens

taskgo

Maintains a private Git-backed personal project and task control repository using concise current Markdown, derived Git history, ADRs, and status synchronization. Use when tracking tasks, updating project status (STATUS.md), managing ADRs, synchronizing tracker state, delegating tasks to external agents, or generating…

ithinkihaveacat/dotfiles · 73 tokens

vibe

Delegate a coding task to a cheap AI model (Mistral Vibe by default, but any provider Vibe knows about — DeepSeek, Gemini Flash, etc.) and supervise the result via git diff. Claude orchestrates, the cheap model codes. Claude consumes 500-1500 tokens per delegation regardless of how many file reads the delegate does…

sebastienrousseau/dotfiles · 137 tokens