Tessl • The management layer for AI agent skills

SkillPublic

best-of-n

Generate a small set of independent candidate solutions in worktrees, judge them against one explicit rubric, and apply the winner only after PASS verification.

Quality

76%

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 ./crates/tui/assets/skills/best-of-n/SKILL.md

Quality

Content

85%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 lean, highly actionable body with an explicit validation-gated workflow that fully covers the batch/parallel-write risk profile. The main defect is structural: it cites a checked-in workflow recipe that is absent from the bundle and gives no path to it.

Suggestions

Ship the referenced operate_best_of_n.workflow.js in scripts/ (or references/) and cite it with a resolvable path, e.g. 'See [scripts/operate_best_of_n.workflow.js](scripts/operate_best_of_n.workflow.js)'.

Add a one-line invocation example for the workflow recipe's 'strategy: "search"' option so the 2–16 candidate path is executable, not just described.

Remove or soften the 'checked-in' claim until the file actually exists in the bundle, since a dangling reference undermines navigation.

Dimension	Reasoning	Score
Conciseness	Imperative and dense throughout ('Never run parallel writers in the parent checkout', 'Do not ask candidates to vote for themselves') with zero padding and no explanations of concepts Claude already knows — every token earns its place.	5 / 5
Actionability	Provides a copy-paste-ready JSON agent action block plus concrete builder settings ('worktree: true', 'write_authority: "worktree_write"') and specific ranges (N 2–4, 2–16). Minor gaps: the 'checked-in operate_best_of_n.workflow.js' recipe has no invocation example and the file is not present in the bundle.	4 / 5
Workflow Clarity	Four clearly sequenced phases (set tournament → generate → judge → integrate) with explicit validation gates: replay on a clean baseline, the runtime (not the worker) runs hard gates, stop-early on invalidating constraints, and 'NONE is valid when every candidate fails'. Feedback loops for this batch/parallel-write operation are fully specified.	5 / 5
Progressive Disclosure	Sections are well organized and the single external reference is one level deep, but that reference is a bare filename ('operate_best_of_n.workflow.js') with no path or link, and no scripts/ directory exists in the bundle — it is a dangling reference the reader cannot locate.	3 / 5
	Total	17 / 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 states a clear, specific 'what' with three concrete actions but omits any 'when to use this' trigger guidance, capping completeness at 3. Adding an explicit 'Use when...' clause with natural user phrases would lift both completeness and trigger quality.

Suggestions

Append a 'Use when...' clause with natural trigger phrases, e.g. 'Use when a task has several plausible approaches worth comparing (best-of-n, tournament-style evaluation) or the user asks to compare candidate solutions.'

Add natural synonyms users would say — 'compare options/approaches', 'pick the best solution', 'evaluate candidates' — to improve trigger term coverage.

Mention the negative-result outcome ('NONE is valid when every candidate fails') to sharpen distinctiveness from generic review skills.

Dimension	Reasoning	Score
Specificity	Lists three concrete actions — 'Generate a small set of independent candidate solutions in worktrees', 'judge them against one explicit rubric', 'apply the winner only after PASS verification' — but coverage has minor gaps (no mention of candidate diversity, choosing N, or the integration/review workflow).	4 / 5
Completeness	The 'what' is clear and concrete, but there is no 'Use when...' clause or equivalent explicit trigger guidance, which caps completeness at 3 per the rubric guidelines.	3 / 5
Trigger Term Quality	Mechanism keywords like 'candidate solutions', 'worktrees', and 'rubric' are present, but natural user phrasings ('compare options', 'best of n', 'evaluate approaches') and their synonyms are missing.	3 / 5
Distinctiveness Conflict Risk	'worktrees', 'rubric', and 'PASS verification' carve a distinct ensemble-comparison niche with minimal overlap, though it could still trigger for general review/evaluation tasks since natural triggers are absent.	4 / 5
	Total	14 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: Hmbown/Codewhale
Path: crates/tui/assets/skills/best-of-n/SKILL.md
Revision: ff466b2

Reviewed: 29 days ago

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.

best-of-n

Table of Contents

Is this your skill?