autoresearch

autoresearch is a skill for Claude Code, Codex from jimezsa/opencolab. It costs 49 tokens per session (1,148 once invoked), scanned A, original, MIT.

An iterative experiment workflow for repeatedly changing one configured file in a repository, running an experiment, and keeping or discarding the change based on a metric. A metric is a measurable result used to judge whether the experiment improved.

In plain words
What is it for?
Use it to run controlled optimization or machine-learning experiments while recording whether each change improves the chosen metric.
Why use it?
It prevents unclear experiment loops by requiring the repository, editable file, run command, and success rule to be specified first.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to run controlled optimization or machine-learning experiments while recording whether each change improves the chosen metric.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/jimezsa/opencolab/autoresearch
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.

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

Made for: Claude Code, Codex.

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/jimezsa/opencolab/autoresearch/github.svg)](https://agentmods.dev/skills/jimezsa/opencolab/autoresearch)
Your own site
<a href="https://agentmods.dev/skills/jimezsa/opencolab/autoresearch"><img src="https://agentmods.dev/badge/skills/jimezsa/opencolab/autoresearch/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for autoresearch

Your own site · 80×15
<a href="https://agentmods.dev/skills/jimezsa/opencolab/autoresearch"><img src="https://agentmods.dev/badge/skills/jimezsa/opencolab/autoresearch.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 49 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,148 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00049 $0.01148
Opus 5 $0.00024 $0.00574
Sonnet 5 $0.00010 $0.00230
Haiku 4.5 $0.00005 $0.00115

Measured 8d ago against content hash e4cc0d26e4fe, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, 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 8d ago.

The scan reads SKILL.md. This mod also ships 1 executable file (scripts/plot_progress.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

projects/SKILLS/autoresearch/SKILL.md · 147 lines

How it starts

The opening of the file, as written. The whole thing — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Autoresearch Skill

Use this skill for iterative experiment loops over one explicitly configured repo.

This is a normal shared OpenColab skill. Any agent may use it when it is the right tool, but the built-in autoresearch specialist is the default owner for sustained experiment-loop work.

Required Repo Contract

Do not start the loop until these are explicit:

  • repo_path: local path to the experiment repo
  • editable_file_path: the only file you may modify
  • run_command: the command that launches one experiment run
  • metric_rule: how to extract the metric and whether higher or lower is better

Optional:

  • results_file: repo-local results table or summary log
  • branch_prefix: disposable experiment branch prefix

Default failure rule:

  • a run fails when the command exits non-zero or the metric cannot be extracted

Do not assume:

  • the editable file is train.py
  • the run command is uv run train.py

Those are upstream examples, not OpenColab requirements.

Core Rules

  • Work only inside the configured repo.
  • Edit only the configured editable file unless the human explicitly changes the repo contract.
  • Treat the configured metric rule as the source of truth for keep/discard decisions.
  • Keep changes narrow and reviewable.
  • Run one bounded experiment at a time.
  • Keep artifacts in the experiment repo unless the user asks for a different handoff.
  • If remote GPU execution is needed, combine this workflow with projects/SKILLS/runpod-job/SKILL.md.

Git Safety Boundary

Use a dedicated disposable branch or worktree for this loop.

Inside that dedicated experiment branch or worktree:

  • keep or discard decisions may use branch rewinds
  • discard operations are pre-approved for this workflow

Outside that dedicated experiment branch or worktree:

  • normal OpenColab safety rules still apply
  • do not perform destructive git actions without approval

Minimal Loop

  1. Confirm the repo contract.
  2. Enter the configured repo and inspect the current baseline.
  3. Ensure you are working in the dedicated disposable experiment branch or worktree.
  4. Make one narrow change in the configured editable file.
  5. Run the configured experiment command.
  6. Extract the metric using the configured metric rule.
  7. Decide keep or discard.
  8. Record the outcome in the configured results file when one exists.
  9. Summarize the result and next useful step.

Read the full file on GitHub · 147 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 8d ago First seen · 147 lines · 49 tokens per session scan A e4cc0d26e4fe

Subscribe to this mod's changes

autoresearch is a skill published in the GitHub repository jimezsa/opencolab (11 stars, last pushed 26d ago), licensed MIT. It adds 49 tokens to every session and 1,148 once invoked, about $0.0002 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