Not in the top 1,000 by GitHub stars
Muminur/autoresearch-skill-Andrej-Karpathy
Claude Code skill for autonomous, goal-directed iteration. /autoresearch <goal> builds a real-data benchmark harness, captures a baseline, and iterates with a regression gate until the goal is hit. Inspired by Andrej Karpathy's autoresearch.
History
Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.
View weekly changes
SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
760eeaafae8f → 760eeaafae8f
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 760eeaafae8f
No earlier version was observed for comparison.
20GitHub stars
0stars in 7 days
1skill in this repository
0commits in 30 days
Heat
0.3- Star growth, last 7 daysNot available
- Star growth, last 24 hoursNot available
- Rank movement, last 7 daysNot available
- Commits, last 30 daysNot available
- Accounts asking about it through MCP, last 7 daysNot available
- Hot votes this weekNot available
Heat never changes the order by stars, and sponsorship never touches it. How heat works
Stars over time
30-day change not yet available: not enough history.
Individual skills in this repository
How it works
autoresearch
When your agent uses it
Autonomous Goal-directed Iteration. Apply Karpathy's autoresearch principles to ANY task. Loops autonomously — modify, verify, keep/discard, repeat. Supports optional loop count via Claude Code's /loop command. Invoking /autoresearch <free-form goal> builds a real-data benchmark harness, captures a baseline, and iterates with a regression gate until the goal is hit.
What it loads into context
Frontmatter
93 tokensAlways available to the agent
SKILL.md
3,455 tokensWhen the skill is activated
Reference documents
20,104 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Subcommands
- When to Activate
- Optional: Controlled Loop Count
- Setup Phase (Do Once)
- The Loop
- Critical Rules
- Principles Reference
- Adapting to Different Domains
Files
.gitignoreasset789 bytesLICENSEasset1,064 bytesREADME.mddoc17,945 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesSKILL.mdcore14,452 bytesreferences/autonomous-loop-protocol.mddoc5,585 bytesreferences/benchmark-harness.mddoc7,694 bytesreferences/core-principles.mddoc3,911 bytesreferences/plan-workflow.mddoc8,133 bytesreferences/results-logging.mddoc2,168 bytesreferences/security-workflow.mddoc35,472 bytes
Read SKILL.md
# Claude Autoresearch — Autonomous Goal-directed Iteration
Inspired by [Karpathy's autoresearch](https://github.com/karpathy/autoresearch). Applies constraint-driven autonomous iteration to ANY work — not just ML research.
**Core idea:** You are an autonomous agent. Modify → Verify → Keep/Discard → Repeat.
## Subcommands
| Subcommand | Purpose |
|------------|---------|
| `/autoresearch <goal>` | **Default path** — parse free-form goal, build harness, capture baseline, loop until goal met |
| `/autoresearch` | Run the autonomous loop (default) |
| `/autoresearch:plan` | Interactive wizard to build Scope, Metric, Direction & Verify from a Goal |
| `/autoresearch:security` | Autonomous security audit: STRIDE threat model + OWASP Top 10 + red-team (4 adversarial personas) |Skills included
- autoresearch
Repository facts
- License
- MIT
- Created
- Apr 19, 2026
- Last pushed
- Apr 19, 2026
- Forks
- 2
- Open issues
- 0
- Project activity
- Slowing
Before you install
Network or broad shellAutomated signals, not a guarantee
Checked Nov 7, 2026
Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.
Show 7 notes
SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedSKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzed
- No broad tool permissions
- No dangerous shell commands
- 31 network referencesin 4 files: README.md:7 references/security-workflow.md:468 SKILL.md:9 and 1 more
- No hidden or obfuscated text
- No instruction overrides
- No executable scripts
- 3 references to secretsin 2 files: references/security-workflow.md:358 .gitignore:23
- No requests to vote in Hot
Show all 34 signals by file
.gitignore
- Access to secretsline 23
[Credential-related source omitted]
README.md
- Network accessline 7
*Inspired by [Andrej Karpathy's autoresearch](https://github.com/karpathy/autoresearch) — extended into a universal, real-data benchmark-driven workflow for any engineering task.* - Network accessline 9
[](./SKILL.md) - Network accessline 10
[](https://docs.claude.com/en/docs/claude-code) - Network accessline 11
[](./LICENSE) - Network accessline 12
[](https://github.com/karpathy/autoresearch) - Network accessline 13
[](#) - Network accessline 103
# Clone into your Claude Code skills directory git clone https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy.git \ - Network accessline 104
git clone https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy.git \ - Network accessline 110
git clone https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy.git ` - Network accessline 143
git clone https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy.git ~/autoresearch-skill - Network accessline 152
[Codex CLI](https://github.com/openai/codex) loads custom prompts from `~/.codex/prompts/`. - Network accessline 195
[OpenCode](https://github.com/sst/opencode) keeps custom commands at `~/.config/opencode/command/`. - Network accessline 267
For [Continue](https://continue.dev): - Network accessline 516
- **[Andrej Karpathy](https://github.com/karpathy/autoresearch)** — for the original autoresearch pattern: *single file, single metric, iterate*. - Network accessline 517
- **[Strix](https://github.com/usestrix/strix)** — adversarial AI security testing with PoC validation (inspiration for `/autoresearch:security`). - Network accessline 537
[Report an issue](https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy/issues) · [Open a PR](https://github.com/Muminur/autoresearch-skill-Andrej-Karpathy/pulls)
SKILL.md
- Network accessline 9
Inspired by [Karpathy's autoresearch](https://github.com/karpathy/autoresearch). Applies constraint-driven autonomous iteration to ANY work — not just ML research. - Network accessline 117
**Inspired by:** - [Strix](https://github.com/usestrix/strix) — AI-powered security testing with proof-of-concept validation - Network accessline 118
- [Strix](https://github.com/usestrix/strix) — AI-powered security testing with proof-of-concept validation
references/benchmark-harness.md
- Network accessline 22
1. Identify the corpus source from the goal (logs, exports, fixtures, API, scraped pages). 2. If a corpus directory already exists under `autoresearch/data/`, use it. 3. If not, SCRAPE it (Playwright / curl / `tail` on prod logs / database - Network accessline 23
2. If a corpus directory already exists under `autoresearch/data/`, use it. 3. If not, SCRAPE it (Playwright / curl / `tail` on prod logs / database export) into `autoresearch/data/<name>.jsonl` — one real case per line. 4. **NEVER synthesi - Network accessline 24
3. If not, SCRAPE it (Playwright / curl / `tail` on prod logs / database export) into `autoresearch/data/<name>.jsonl` — one real case per line.
references/security-workflow.md
- Access to secretsline 358
[Credential-related source omitted] - Network accessline 468
# Test for missing rate limiting for i in {1..100}; do curl -s -o /dev/null -w "%{http_code}" https://app/api/login; done - Network accessline 469
for i in {1..100}; do curl -s -o /dev/null -w "%{http_code}" https://app/api/login; done - Network accessline 471
[Credential-related source omitted] - Network accessline 472
[Credential-related source omitted] - Network accessline 474
# Test for XSS curl https://app/search?q=%3Cscript%3Ealert(1)%3C/script%3E - Network accessline 475
curl https://app/search?q=%3Cscript%3Ealert(1)%3C/script%3E - Network accessline 670
- name: Install Autoresearch Skill run: | git clone https://github.com/uditgoenka/autoresearch.git /tmp/autoresearch - Network accessline 671
run: | git clone https://github.com/uditgoenka/autoresearch.git /tmp/autoresearch cp -r /tmp/autoresearch/skills/autoresearch .claude/skills/autoresearch - Network accessline 672
git clone https://github.com/uditgoenka/autoresearch.git /tmp/autoresearch - Access to secretsline 677
[Credential-related source omitted]
How the signals fit together
Facts from a static read of the files at the pinned commit. They describe what the code does, not why.
Some files could not be read; the facts below still hold.
Signal confidence: 33 facts
- Network access in prose of documentationREADME.md:7Low confidence
- Network access in prose of documentationREADME.md:9Low confidence
- Network access in prose of documentationREADME.md:10Low confidence
- Network access in prose of documentationREADME.md:11Low confidence
- Network access in prose of documentationREADME.md:12Low confidence
- Network access in prose of documentationREADME.md:13Low confidence
- Network access in code of documentationREADME.md:103Low confidence
- Network access in code of documentationREADME.md:104Low confidence
- Network access in code of documentationREADME.md:110Low confidence
- Network access in code of documentationREADME.md:143Low confidence
- Network access in prose of documentationREADME.md:152Low confidence
- Network access in prose of documentationREADME.md:195Low confidence
- Network access in prose of documentationREADME.md:267Low confidence
- Network access in prose of documentationREADME.md:516Low confidence
- Network access in prose of documentationREADME.md:517Low confidence
- Network access in prose of documentationREADME.md:537Low confidence
- Network access in prose of the SKILL.mdSKILL.md:9Low confidence
- Network access in prose of the SKILL.mdSKILL.md:117Low confidence
- Network access in prose of the SKILL.mdSKILL.md:118Low confidence
- Network access in prose of documentationreferences/benchmark-harness.md:22Low confidence
- Network access in prose of documentationreferences/benchmark-harness.md:23Low confidence
- Network access in prose of documentationreferences/benchmark-harness.md:24Low confidence
- Access to secrets in code of documentationreferences/security-workflow.md:358Low confidence
- Network access in code of documentationreferences/security-workflow.md:468Low confidence
- Network access in code of documentationreferences/security-workflow.md:469Low confidence
- Network access in code of documentationreferences/security-workflow.md:471Low confidence
- Network access in code of documentationreferences/security-workflow.md:472Low confidence
- Network access in code of documentationreferences/security-workflow.md:474Low confidence
- Network access in code of documentationreferences/security-workflow.md:475Low confidence
- Network access in code of documentationreferences/security-workflow.md:670Low confidence
- Network access in code of documentationreferences/security-workflow.md:671Low confidence
- Network access in code of documentationreferences/security-workflow.md:672Low confidence
- Access to secrets in code of documentation: a variable, placeholder or call, not a stored valuereferences/security-workflow.md:677Low confidence
Works with
| Agent | Status | Evidence | Install to |
|---|---|---|---|
| Claude Code | Compatible | Standard SKILL.md SKILL.md | .claude/skills/<skill-name>/ |
| Codex | Compatible | Standard SKILL.md SKILL.md | .agents/skills/<skill-name>/ |
| Cursor | Compatible | Standard SKILL.md SKILL.md | .cursor/skills/<skill-name>/ |
| Gemini CLI | Compatible | Standard SKILL.md SKILL.md | .gemini/skills/<skill-name>/ |
| GitHub Copilot | Compatible | Standard SKILL.md SKILL.md | .github/skills/<skill-name>/ |
| OpenCode | Compatible | Standard SKILL.md SKILL.md | .opencode/skills/<skill-name>/ |
- Native The repository ships a manifest, folder or frontmatter field for this agent.
- Compatible A standard SKILL.md that this agent can read.
- Unknown No evidence either way.
Derived automatically from paths, manifests and frontmatter. Not tested on each agent.
