Autoresearch toolfinder

Skill jajupmochi/agent-harness/skills/general/autoresearch-toolfinder

Linlin's curated AI agent harness configuration: workflow rules, skills, hooks, plugin recommendations, tooling preferences, and project templates. Loadable into any new project so a fresh /init can pick the relevant subset.

Install

npx -y skills add jajupmochi/agent-harness --skill autoresearch-toolfinder

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Find the right autonomous-research / autoresearch tool, framework, port, or skill for a research or engineering task. Searches a local cached index of two curated awesome-autoresearch lists (alvinreal + yibie, 550+ entries) and returns only the few matching tools, never loading the whole catalog into context. Use when the user wants to pick, compare, or set up an autoresearch loop, an AI-scientist / research-agent system, a domain or hardware port (Apple Silicon, RTX, RL, trading, materials, bio, vision, kernels...), or an evaluation harness, or asks "is there an autoresearch tool for X".

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.3 KB, as published. Nobody here has run it

autoresearch-toolfinder

Recommends tools from two curated awesome-autoresearch catalogs (550+ entries) WITHOUT reading the whole list into context. You run a search script and read back only the top matches.

How to use (token-efficient — follow this; do NOT cat the index)

The catalog is large. Never read data/index.json directly (that defeats the purpose). Run the query script from the skill directory; it returns only the top candidates:

python3 scripts/query.py "<keywords from the user's task>"
# options:  --source alvinreal|yibie   --category "<substring>"   --limit 8   --json
python3 scripts/query.py --list-categories      # see sections + counts first

Examples:

Apple-Silicon / MLX port: python3 scripts/query.py "apple silicon mlx mac metal"
End-to-end AI scientist: python3 scripts/query.py "ai scientist paper literature review" --source alvinreal
RL post-training loop: python3 scripts/query.py "reinforcement learning grpo post-training"
Trading strategy search: python3 scripts/query.py "trading strategy backtest" --source yibie
Browse a whole section: python3 scripts/query.py "" --category "Evaluation"

Then: read the handful of name + url + one-liner results, pick the best 1-3 for the user's actual context, and say why. Open a specific repo URL (WebFetch) only if the user wants depth.

When to activate (auto)

Activate when the user is choosing / comparing / setting up: an autoresearch or self-improvement loop; an AI-scientist or research-agent system; a hardware/platform port; a domain adaptation (bio, materials, finance, vision, RL, kernels, robotics...); or an eval harness — or asks "what should I use for autonomous research / overnight experiments on X".

Not this skill: to actually run a full autonomous research project end-to-end, use the sibling autoresearch orchestration skill. This skill is the catalog / finder only.

Keeping it current (update tracking)

data/state.json stores each upstream repo's commit SHA + sync time.

python3 scripts/check_updates.py     # cheap: 1 API call/repo, compares SHA, exits 1 if stale
python3 scripts/update_index.py      # re-fetch + re-parse both repos, rewrite the index

A weekly user systemd timer (systemd/autoresearch-index.timer) refreshes automatically; query.py also prints a hint when the local index is older than 30 days.