autoresearch

Run bounded automated experiment iterations by recording baselines, applying hypothesis patches, comparing metrics, protecting regression guards, and deciding keep, discard, rollback, or block. Use when automated research is requested or a repo/skill needs evidence-backed research, metric tracking, or safe optimisation loops.

Quality

85%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

—

The risk profile of this skill

Quality

Content

85%Scale 1-3

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

This is a well-structured, highly actionable skill with clear workflow sequencing, explicit validation checkpoints, and proper progressive disclosure to reference files. Its main weakness is moderate verbosity—some sections overlap (Constraints vs Execution Boundaries, repeated guard/discard logic across Decision Language, Gotchas, and Acceptance Criteria) and could be consolidated for better token efficiency. Overall it is a strong skill that provides concrete, executable guidance for a complex multi-step process.

Suggestions

Consolidate overlapping sections: merge 'Execution Boundaries' into 'Constraints', and deduplicate guard/discard logic that appears in 'Decision Language', 'Gotchas', and 'Acceptance Criteria' to reduce redundancy and improve token efficiency.

Dimension	Reasoning	Score
Conciseness	The skill is fairly dense and avoids explaining basic concepts, but some sections are verbose or redundant (e.g., 'Execution Boundaries' and 'Constraints' overlap, 'Decision Language' repeats guard/discard logic found elsewhere). Several bullet points could be tightened without losing meaning.	2 / 3
Actionability	The skill provides concrete executable examples (shell commands, YAML ledger entries, specific command outputs), clear decision rules (baseline - candidate >= min_delta), and copy-paste-ready verification commands. The iteration example and ledger entry are fully concrete and actionable.	3 / 3
Workflow Clarity	The workflow is clearly sequenced (9 numbered steps) with explicit validation checkpoints (baseline first, verify then guard, fail-fast at first failed gate), feedback loops (repair smallest failing hypothesis, rerun gate before broad validation), and clear decision criteria (keep/discard/block with evidence). The repair/failure behavior section adds a proper error recovery loop.	3 / 3
Progressive Disclosure	The skill ends with a clear 'Progressive Disclosure' section pointing to one-level-deep references (references/autoresearch-project.md, references/contract.yaml, references/evals.yaml, references/task-profile.json). The main body serves as an effective overview without inlining content that belongs in those reference files. The discovery interview also references a separate file appropriately.	3 / 3
	Total	11 / 12 Passed

Description

85%Scale 1-3

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

This is a strong description that clearly articulates a specific experimental iteration workflow with concrete actions and an explicit 'Use when' clause. Its main weakness is that some trigger terms are technical jargon that users might not naturally use when requesting this capability, though the 'Use when' clause helps compensate with more accessible terms like 'automated research' and 'metric tracking'.

Suggestions

Add more natural user-facing trigger terms such as 'A/B testing', 'benchmarking', 'performance comparison', or 'try different approaches' to improve discoverability when users phrase requests colloquially.

Dimension	Reasoning	Score
Specificity	Lists multiple specific concrete actions: 'recording baselines', 'applying hypothesis patches', 'comparing metrics', 'protecting regression guards', and 'deciding keep, discard, rollback, or block'. These are detailed, actionable steps.	3 / 3
Completeness	Clearly answers both 'what' (run bounded automated experiment iterations with specific steps) and 'when' (explicit 'Use when' clause covering automated research requests, evidence-backed research, metric tracking, or safe optimisation loops).	3 / 3
Trigger Term Quality	Includes some relevant terms like 'automated research', 'metric tracking', 'optimisation loops', and 'experiment iterations', but uses somewhat technical jargon ('hypothesis patches', 'regression guards') that users may not naturally say. Missing more common variations like 'A/B testing', 'benchmarking', or 'performance testing'.	2 / 3
Distinctiveness Conflict Risk	The combination of bounded experiment iterations, hypothesis patches, regression guards, and keep/discard/rollback decisions creates a very distinct niche. This is unlikely to conflict with general coding, testing, or research skills due to its specific experimental loop focus.	3 / 3
	Total	11 / 12 Passed

Validation

90%

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 — 10 / 11 Passed

Validation for skill structure

Criteria	Description	Result
metadata_field	'metadata' should map string keys to string values	Warning

	Total	10 / 11 Passed

Repository: jscraik/Agent-Skills
Path: Skills/agent-ops/autoresearch/SKILL.md
Commit: d933d80

Reviewed: 20 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.