Skills
autoresearch-universal
by balukosuri
Self-improving prompt optimization using the Karpathy autoresearch pattern.
Our verdict
We have not reviewed autoresearch-universal yet
No one at Enterprise DNA has checked this listing against its sources. The description comes from public listings gathered automatically. Any fact below shows where it came from and when it was last checked.
From the listing
About autoresearch-universal
Self-improving prompt optimization using the Karpathy autoresearch pattern. Scans any repo, suggests optimization targets, auto-defines binary eval metrics, and runs an autonomous generate-eval-score-mutate loop to improve prompts over time. Use when the user asks to optimize, improve, or run autoresearch on anything in their codebase.
Best for
Developers seeking automated prompt improvement across existing codebases
Use cases
- Automatically identify and optimize prompt targets in a codebase
- Define binary eval metrics for prompt performance
- Run an autonomous loop to iteratively improve prompts
Notes
autoresearch-universal — evals from balukosuri.
209 stars on GitHub. Last updated 2026-03-23.
Indexed from github-skill-md. Verified against the live GitHub repo.
Pros
- Autonomous optimization reduces manual tuning effort
- Works on any repository, not tied to a specific framework
- Clear eval metrics provide measurable improvement
Cons
- Requires a codebase with identifiable prompt targets
- Binary metrics may oversimplify complex prompt quality
- Autonomous loop may need supervision to avoid regressions
Key facts
What we have checked
- Days since the last commit
- 236
- GitHub forks
- 33
- GitHub stars
- 209
- Open issues
- 0
- Releases in the last 90 days
- 0
Each value shows when we last confirmed it and where it came from. Values we have not confirmed are left out, never guessed.
Stars and usage
How autoresearch-universal is growing
208 to 209
Health and security
Is autoresearch-universal looked after, and what do its setup files show
Health score
19 out of 100
| Part | Points | What we saw |
|---|---|---|
| Recent activity | 11.6 of 30 | Last push 236 days ago |
| Releases | 0 of 15 | 0 releases in 90 days |
| Issue replies | Not counted | The latest check has no issue response reading (fewer than 3 measured issues, or issues switched off) |
| Maintainer | 0 of 15 | Owned by a single user |
| Official source | Not counted | MCP servers only |
Security check static checks
90 out of 100
- Medium Tells the agent to skip permission or approval prompts: SKILL.md, line 441.
- Note 1 of 1 packages are not pinned to one version, so only the malicious packages list was checked for them, not version advisories.
Alternatives
Alternatives to autoresearch-universal
| Record | Type | How we know | Checked |
|---|---|---|---|
| autoresearch-skillAlternative to this | Skill | Matched from the listing text |
Compare
Compare autoresearch-universal side by side
Sources
Where this comes from
- GitHubRegistry
A source shows what was published when we checked it. It does not certify security, performance or a working installation. Read our method.
Changelog
What changed
No changes recorded yet. Releases, price changes and deprecations appear here, with their dates, once we track them for this record.
Open-source & AI alternatives
Swap-in tools that solve the same job. Weigh the trade-offs before you commit.
Pairs with
Other entries in the index that connect to this one. Click through to see the chain.
eval-engineering
langchain-ai
Inspect an agent repository and optional traces, interview the user, write reviewed Task Specs, build and audit Harbor tasks, and bootstrap reusable project World Knowledge…
evaluating-llms-harness
Orchestra-Research
Comprehensive open-source library of AI research and engineering skills for any AI model. Package the skills and your claude code/codex/gemini agent will be an AI research agent wi
Get the free Developer’s Field Guide
A 27-page field guide to the AI coding workflow with Claude. Claude Code, MCP servers, the prompt patterns that work, and what to delegate. Free.
Enter your work email. We send it straight over, plus a few short notes worth knowing. Unsubscribe any time.