Agent skills

Skills matching “autoresearch”

Read straight from the source repositories, not from submitted listings. Every skill shows what it does, what is inside, where it came from — and whether attention around its source is actually growing.

Toolclaude-code 29,140codex 4,755cursor 3,111copilot 976windsurf 55cline 34
CategoryWorkflow & Productivity 4,979AI & Agents 3,037Data & Analytics 2,345Code Review & Quality 1,376Backend & API 1,244Security 1,194Design & Presentation 1,154Documentation 965Content & Marketing 916Testing & QA 777DevOps & Cloud 576Databases 550Frontend 469Business & Finance 328Media & Video 257Other 9,833
17 found
117 · page 1 / 1
lab:autoresearchSelf-improving loop for plugin skills. Reads program.md, proposes one mutation per iteration, evaluates against deterministic…oliver-kriskascriptsautoresearchAutonomous iteration loop: modify, verify, keep/discard against any metricuditgoenkaautoresearchOrchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment…OpenRaisertournament-autoresearchUse when the user wants an autonomous ML research loop that pressure-tests competing ideas before spending compute — several…gaashercodex-autoresearchRun or resume a measured improvement loop in a local project. Use for benchmark-driven optimization, qualitative quality-gap…TheGreenCedarautoresearchOrchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment…brycewang-stanfordml-autoresearchUse when the user wants an autonomous ML research loop that does more than blindly try changes. After every training run the…gaasherdueling-autoresearchUse when the user wants two approaches raced head-to-head on a single shared metric — e.g. a classical/algorithmic lane vs an…gaasherautoresearch-agentAutonomous experiment loop that optimizes any file by a measurable metric. Inspired by Karpathy's autoresearch. The agent edits a…majiayu000autoresearchRun a bounded, source-grounded research loop, draft a cited dossier, and optionally propose a separately reviewed canonical vault…AgriciDanielautoresearch-mlAutonomous LLM training optimization with GPU support. Runs 5-minute training experiments, measures val_bpb, keeps improvements…majiayu000autoresearchALWAYS activate when user types /autoresearch, $autoresearch plan, $autoresearch debug, $autoresearch fix, $autoresearch…majiayu000autoresearchThis skill should be used when the user asks to "run autoresearch", "optimize X in a loop", "set up autonomous experiments"…majiayu000autoresearchAutonomous experiment loop: edit code, commit, run benchmark, extract metrics, keep improvements or revert, repeat forever. Use…majiayu000autoresearchScaffold and run Karpathy-style autoresearch loops in any git repo. This skill should be used when setting up autonomous code…majiayu000exploratory-autoresearchUse when the user wants an autonomous ML research loop that explores the space broadly rather than hill-climbing one approach. A…gaasherautoresearchAutonomous iterative experimentation loop for any programming task. Guides the user through defining goals, measurable metrics…github
How the catalog works
What is an agent skill?

A folder with a SKILL.md inside — instructions, and often scripts and assets, that an AI agent loads when the task matches. Claude Code, Codex, Cursor and Copilot all read the same format, so one skill usually works across them.

Where does this catalog come from?

We read 660 source repositories straight from their file trees rather than from submitted listings — what you see is what is actually published. 98 repositories were rejected because they advertise skills but contain none: link lists, not folders.

Why is there no install counter?

Because install counts live in the registry that serves `npx skills add`, and that is not ours — publishing a number we cannot verify would be worse than showing none. Instead we show where a skill comes from and whether attention around its source is actually growing, measured from our own weekly snapshots.

Do you deduplicate?

Yes, and it matters more than expected. Aggregator repositories republish the same skill in several places — one source carried 6,317 SKILL.md files for 2,001 actual skills. We collapse by folder name and keep the canonical copy, so the catalog counts things, not copies.

Keep going