autoresearch

Bounded research experiment loop that tries hypotheses, measures benchmark evidence, keeps what works, and records what fails. Use when the user asks to optimize a research metric, run an experiment loop, improve model/retrieval/evaluation performance iteratively, or benchmark a research hypothesis.

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/autoresearch/SKILL.md

Quality

Content

76%Weight 40%Scale 1-5

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

A token-efficient delegation skill that correctly points Claude to the /autoresearch workflow and names the relevant tools and session files. Its main weakness is workflow clarity: the body neither surfaces nor points to validation/feedback steps for the iterative benchmark loop it orchestrates.

Suggestions

Add a one-line validation/feedback checkpoint to the body (e.g., 'After each run, compare benchmark result to prior best in autoresearch.jsonl before keeping or discarding the change') so the iterative loop has an explicit verify->fix->retry step.

Specify the fallback recording format briefly (e.g., one JSONL line per experiment with metric, value, evidence, decision) so the no-tools path is fully actionable.

Add lightweight section headers (## Workflow, ## Tools, ## Session files) to lift progressive_disclosure to a clear 5 for this short skill.

Dimension	Reasoning	Score
Conciseness	The ~6-line body is lean and efficient: it delegates to the slash command, names optional tools, and lists session files with no padding or explanation of concepts Claude already knows, matching the 5-anchor 'every token earns its place'.	5 / 5
Actionability	Gives concrete directives ('Run the /autoresearch workflow', 'do not try to read a relative prompt-template path', named tools, fallback to 'record benchmark result, evidence, and decision in the session files'); as an instruction-only skill the lack of code is acceptable, but the fallback path omits specifics like record format, leaving minor gaps vs the 5-anchor.	4 / 5
Workflow Clarity	The body delegates the real sequence to the slash command and only sketches 'run workflow, use tools, record results' without any validation checkpoints; per the batch-operation cap, an iterative experiment loop lacking explicit validation/feedback steps cannot exceed 3 even though delegation is intentional.	3 / 5
Progressive Disclosure	Under 50 lines with no external reference bundle, and content is split into clearly labeled concerns (delegation, optional tools, session files); not a 5 because the paragraphs lack explicit section headers and the session-file list is informational rather than navigational.	4 / 5
	Total	16 / 20 Passed

Description

83%Weight 40%Scale 1-5

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

A strong, third-person description that clearly states what the skill does and when to use it, with concrete experiment-loop actions and natural trigger phrases. Minor gains are possible by sharpening the most abstract verbs and adding a few synonyms for trigger coverage.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions ('tries hypotheses, measures benchmark evidence, keeps what works, and records what fails') covering the experiment-loop phases, with only minor abstraction gaps; not a 5 because 'keeps what works'/'records what fails' are less crisply concrete than the 5-anchor examples.	4 / 5
Completeness	Explicitly answers both 'what' (bounded experiment loop with the four listed actions) and 'when' via a concrete 'Use when the user asks to...' clause with multiple trigger phrases, matching the 5-anchor.	5 / 5
Trigger Term Quality	Natural user-facing phrases appear ('optimize a research metric', 'run an experiment loop', 'improve model/retrieval/evaluation performance', 'benchmark a research hypothesis'); a few common synonyms or shorthand variants are missing, so it sits just below the comprehensive 5-anchor.	4 / 5
Distinctiveness Conflict Risk	The 'bounded research experiment loop' framing with benchmark-specific triggers carves a clear niche with minimal conflict risk, but the broad 'improve model performance' phrasing could slightly overlap with general coding or evaluation skills, keeping it just below 5.	4 / 5
	Total	17 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: companion-inc/feynman
Path: skills/autoresearch/SKILL.md
Commit: 95c51d9

Reviewed: about 11 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.