autoresearch

Orchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment iterations with clear optimization targets. The outer loop synthesizes results, identifies patterns, and steers research direction. Routes to domain-specific skills for execution, supports continuous agent operation via Claude Code /loop and OpenClaw heartbeat, and produces research presentations and papers. Use when starting a research project, running autonomous experiments, or managing a multi-hypothesis research effort.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with clear, validated workflows and a strong progressive-disclosure structure. Its main weaknesses are length/repetition and a broken templates/ reference that undercuts the otherwise clean file split.

Suggestions

Create the referenced templates/ directory (or point to references/ instead) so the line 58 link to research-state.yaml/findings.md templates resolves.

Consolidate repeated guidance — the 'Never stop' principle and agent-continuity/loop setup instructions each appear in multiple sections — to reduce the 411-line body.

Dimension	Reasoning	Score
Conciseness	The tone is lean and avoids explaining basics Claude knows, but at 411 lines it repeats guidance ('Never stop', agent-continuity/loop setup) across sections and could be tightened.	2 / 3
Actionability	Provides copy-paste-ready concrete artifacts: a workspace tree, a full /loop prompt, a complete cron.add JSON payload, git commit message patterns, and an experiment-trajectory JSON example.	3 / 3
Workflow Clarity	The inner and outer loops are numbered with an explicit sanity-check validation checkpoint before trusting results, plus cron verification via cron.list, matching a clear sequence with feedback loops.	3 / 3
Progressive Disclosure	Structure is well-organized with one-level-deep, clearly signaled real references (agent-continuity.md, progress-reporting.md, skill-routing.md), but the referenced templates/ directory does not exist, leaving a broken link.	2 / 3
	Total	10 / 12 Passed

Description

85%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, complete with an explicit 'Use when' trigger, and clearly distinct from other skills. Its main weakness is trigger-term breadth, which leans specialized rather than covering the full range of natural phrasings a user might say.

Suggestions

Broaden trigger terms to capture more natural user phrasings, e.g. 'when the user asks to run experiments overnight', 'iterate on a benchmark', or 'explore a research question autonomously'.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions (orchestrates end-to-end projects, runs inner-loop experiment iterations, synthesizes results, produces papers) rather than vague language.	3 / 3
Completeness	Explicitly states what it does (two-loop orchestration, synthesis, presentations/papers) and when to use it ('Use when starting a research project, running autonomous experiments, or managing a multi-hypothesis research effort').	3 / 3
Trigger Term Quality	Includes some natural terms users would say ('research project', 'autonomous experiments') but the coverage of common phrasings is fairly specialized and not comprehensive.	2 / 3
Distinctiveness Conflict Risk	The two-loop autonomous-research niche plus Claude Code /loop and OpenClaw heartbeat references give it distinct triggers unlikely to conflict with other skills.	3 / 3
	Total	11 / 12 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 missing	Warning
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	13 / 16 Passed

Repository: Orchestra-Research/AI-Research-SKILLs
Path: 0-autoresearch-skill/SKILL.md
Commit: 773a529

Reviewed: about 12 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.