CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

autoresearch

Autonomous improvement loop: scan codebase metrics, scaffold experiment files, run agent-driven iterations until metric improves

56

Quality

70%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./examples/skills/autoresearch/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

71%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 highly actionable, well-sequenced command spec whose four modes each carry executable bash and explicit validation loops. Its weaknesses are a dangling runner.sh reference, some redundant prose sections, and zero progressive disclosure across files.

Suggestions

Define or scaffold runner.sh (it is referenced in --scaffold's next steps but never created), or replace the reference with the inline loop instructions already present in Mode 3.

Move 'The pattern (background)' section and the program.md minimal-structure template into a references/ file (e.g. references/pattern.md) to slim SKILL.md to the operational core.

Add an explicit error-handling step in the Mode 3 cycle for measure.sh failing or returning a non-integer (it runs with `set -euo pipefail`).

DimensionReasoningScore

Conciseness

The body is dominated by executable commands and templates, but 'The pattern (background)' comparison table, the 'Key insight' paragraph, and the closing Usage section (which duplicates the mode headers and $ARGUMENTS) are trimmable padding.

4 / 5

Actionability

Every mode ships copy-paste-ready bash (metric greps, loop detection, status checks) and concrete file templates. The gap: `--scaffold`'s next steps instruct `bash scripts/autoresearch/runner.sh {loop-name} 0 15`, but runner.sh is never scaffolded or defined anywhere.

4 / 5

Workflow Clarity

Each mode is numbered with prerequisite checks, and Mode 3 has an explicit feedback loop (measure -> apply -> re-measure -> KEEP/REVERT) that validates every destructive change. Minor gaps remain: no handling if measure.sh fails mid-loop (it uses `set -euo pipefail`) and no post-scaffold verification.

4 / 5

Progressive Disclosure

A single monolithic ~290-line SKILL.md with no bundle files at all. Sections are clearly headed, but the background/pattern section and the program.md template are inline content that belongs in a separate reference file.

3 / 5

Total

15

/

20

Passed

Description

58%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.

The description clearly and concretely states what the skill does in third person, but it lacks any 'when to use' trigger guidance and its keywords are more technical than natural. It is distinguishable from most skills, with only minor overlap risk against general code-quality tooling.

Suggestions

Append a trigger clause such as: 'Use when the user wants to autonomously improve code quality metrics, run improvement experiments, or mentions autoresearch loops.'

Add natural user-facing synonyms to the keyword set (e.g., 'code quality', 'refactor', 'improvement experiments') so the skill triggers on phrasings users actually say.

Mention the --status mode so the description's action list covers the skill's full scope.

DimensionReasoningScore

Specificity

Lists several concrete third-person actions — 'scan codebase metrics', 'scaffold experiment files', 'run agent-driven iterations until metric improves' — but omits the --status mode and program.md authoring, leaving minor gaps in coverage.

4 / 5

Completeness

The 'what' is clear and specific, but there is no 'Use when...' clause or any equivalent trigger guidance; 'when' is entirely absent, capping this dimension at 3.

3 / 5

Trigger Term Quality

Contains relevant keywords ('improvement loop', 'codebase metrics', 'iterations') but misses natural phrases a user would say, such as 'code quality', 'refactor', 'experiments', or 'autonomous'.

3 / 5

Distinctiveness Conflict Risk

'Autonomous improvement loop' with agent-driven keep/revert iterations is a distinct niche, but 'scan codebase metrics' is generic enough to overlap with code-quality-analysis and refactoring skills.

4 / 5

Total

14

/

20

Passed

Validation

87%

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

Validation — 14 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

referenced_paths_exist

Referenced path issues: 12 missing, 4 deeper-than-1-level

Warning

Total

14

/

16

Passed

Repository

FlorianBruniaux/claude-code-ultimate-guide

Path

examples/skills/autoresearch/SKILL.md

Reviewed

Table of Contents

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.