cheat-score-blind

INTERNAL sub-agent for blind 7-dim rubric scoring. **NOT a user-facing skill — do NOT invoke from main conversation.** Called via Task tool by cheat-score / cheat-predict / cheat-bump to get a context-isolated score on a script. Receives ONLY script_path + rubric_notes_path; refuses any other input. Outputs strict JSON: 9 dimensions × {score 0-5, confidence enum, one-line reason}. **Hard refuses to Read** .cheat-state.json, predictions/*, retro 段, or anything that could leak post-publish data. This is channel B in the 3-channel calibration model (A=main, B=blind sub, C=cross-model).

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 ./skills/cheat-score-blind/SKILL.md

Quality

Content

81%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.

The body is highly actionable and well-sequenced with strong validation checkpoints, but is held back by notable redundancy of the contamination/refusal theme and lengthy background that could be tightened. No bundle files exist, yet the single-file structure is clearly navigable.

Suggestions

Consolidate the contamination/refusal guidance into one authoritative section and cross-reference it instead of restating it in Inputs, Refusals, and 调用契约 — this is the main conciseness drain.

Move the detailed path-blacklist table and the full JSON schema into a reference file (e.g. references/refusal-codes.md) and link to it, improving progressive_disclosure toward 5.

Trim the 'Why this exists' rationale to the essential contamination problem; the three-channel model table already conveys the rest.

Dimension	Reasoning	Score
Conciseness	Mostly purposeful but the contamination/refusal constraint is restated across Inputs, 禁止读取, Refusals, and 调用契约 sections and the 'Why this exists' background is lengthy, fitting 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the lean 4.	3 / 5
Actionability	Provides a fully specified JSON output schema, executable grep commands ('grep -E \'\d+\s*[wWmMkK万]\|播放\|实绩\|实际\''), enumerated refusal codes, and a copy-paste Task prompt template — fully executable guidance covering the common cases.	5 / 5
Workflow Clarity	Phases 0–3 are clearly sequenced with explicit validation checkpoints (path whitelist check, contamination self-check grep, input_status/self_check fields) and a feedback loop (detect contamination → flag refusal → lower confidence → still output), satisfying the explicit-validation anchor despite batch use.	5 / 5
Progressive Disclosure	Well-organized into clearly headed sections (Why this exists, three channel model, Inputs, 禁止读取, Workflow, 调用契约, Refusals, Known limitations, Integration) with no bundle files present; a few long blocks (the refusal path table, JSON schema, call-contract template) could be split into reference files, keeping it at good-structure rather than fully optimized.	4 / 5
	Total	17 / 20 Passed

Description

71%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 is specific, distinctive, and covers both what and when via an explicit delegation trigger, though it is jargon-heavy and not oriented to natural user phrasing. It uses appropriate third-person voice with no first/second-person slippage.

Suggestions

Add a short, user-natural trigger phrase even though it is internal (e.g., 'Use when cheat-score/cheat-predict/cheat-bump need an uncontaminated score') to lift trigger_term_quality above 3.

Resolve the 7-dim vs 9-dimensions inconsistency in the same sentence ('blind 7-dim rubric scoring' then '9 dimensions × ...') to avoid ambiguity.

Trim the embedded path blacklist from the description — it duplicates body content and pads the field without adding trigger value.

Dimension	Reasoning	Score
Specificity	Names several concrete actions — 'blind 7-dim rubric scoring', 'Outputs strict JSON: 9 dimensions × {score 0-5, confidence enum, one-line reason}', 'Hard refuses to Read .cheat-state.json, predictions/*' — with only minor coverage gaps, fitting the 'several specific actions' anchor rather than the fully comprehensive 5.	4 / 5
Completeness	Clearly states what it does (blind scoring → strict JSON) and gives equivalent trigger guidance ('Called via Task tool by cheat-score / cheat-predict / cheat-bump'), so both what and when are present with the when reasonably explicit but not maximally trigger-phrased.	4 / 5
Trigger Term Quality	Contains relevant domain keywords ('rubric scoring', 'script', 'rubric_notes', 'JSON', 'Task tool') but they are technical jargon rather than natural user phrases, and the skill is explicitly not user-invoked, so it lands at 'some relevant keywords but missing common variations' rather than good coverage.	3 / 5
Distinctiveness Conflict Risk	Occupies a clear niche ('channel B in the 3-channel calibration model', context-isolated blind scorer) with distinct triggers tied to three named skills, giving minimal conflict risk with other skills.	5 / 5
	Total	16 / 20 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
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	13 / 16 Passed

Repository: XBuilderLAB/cheat-on-content
Path: skills/cheat-score-blind/SKILL.md
Commit: 86823e1

Reviewed: 4 days 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.