content-calibrator

内容校准预测循环——打分+盲预测合一 → 发布 → 记录 → T+3d 复盘 → 进化 rubric。本技能负责打分+预测（blind sub-agent + score-only.sh + commit-prediction.sh + 阈值门）与校准闭环；发布记录与数据采集由 published-track 统一管理。

Quality

68%

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 ./crews/main/skills/content-calibrator/SKILL.md

Quality

Content

77%

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 its workflows are clearly sequenced with validation gates and retry loops. Its weaknesses are repetition of script examples across sections and a monolithic structure with no reference files to offload detailed material.

Suggestions

De-duplicate the script invocations: keep full examples once in 脚本 and reference them from 流程1A (or vice versa) to tighten conciseness toward 3.

Extract the stable reference material (the 7-dimension rubric table + composite formula, the platform ID table, the file-structure tree) into a references/ file (e.g. RUBRIC.md) referenced one level deep, so SKILL.md stays a lean overview and progressive disclosure reaches 3.

Consider folding the duplicated cal-toggle.sh / score-only.sh usage blocks into a single canonical example each.

Dimension	Reasoning	Score
Conciseness	The body is mostly operational and assumes Claude's competence (no 'what is a PDF' padding), but it repeats the same script invocations in 流程1A and again in 脚本 (score-only.sh, commit-prediction.sh, cal-toggle.sh each shown twice) and could be tightened; this matches 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the lean level 3.	2 / 3
Actionability	It provides fully executable, copy-paste-ready bash commands with concrete flags and file paths (score-only.sh --cal-er 3 ..., commit-prediction.sh, cal-toggle.sh, sqlite3 queries) backed by real scripts in ./scripts, matching 'fully executable code/commands; specific examples; copy-paste ready'.	3 / 3
Workflow Clarity	Multi-step flows are explicitly numbered (流程1A steps 1-6, Retro, Bump) with validation checkpoints (阈值门 passed/failing_dims) and a real feedback loop (passed=false → 改稿 → 重新spawn → 再判门, 最多2轮 → 上报用户), matching 'clear sequence with explicit validation steps; feedback loops for error recovery'.	3 / 3
Progressive Disclosure	Scripts are referenced one level deep and clearly signaled (good), and sections are well organized, but SKILL.md is a ~416-line monolithic doc with no references/ files offloading detailed material (rubric table, per-flow steps), fitting 'some structure but content that should be separate is inline' rather than the split, navigable level-3 pattern.	2 / 3
	Total	10 / 12 Passed

Description

60%

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 concrete and names specific scripts and actions, but it lacks an explicit 'when to use' trigger clause and leans heavily on internal jargon, capping completeness and trigger-term quality at 2. It is distinguishable but shares trigger surface with sibling skills like published-track.

Suggestions

Add an explicit 'Use when ...' trigger clause in plain user language (e.g. 'Use when scoring an article/video draft before publishing, or when running a T+3d retro / bumping the rubric') so completeness can reach 3.

Lead with natural trigger terms (打分、盲预测、复盘、校准 rubric) before the internal mechanism names (blind sub-agent, score-only.sh) to improve trigger-term quality.

Sharpen distinctiveness by contrasting the trigger surface against published-track and viral-chaser up front (e.g. '打分+预测用本技能；落库/抓数据用 published-track').

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions (打分、盲预测、阈值门、记录、T+3d 复盘、进化 rubric) plus specific tools (blind sub-agent, score-only.sh, commit-prediction.sh), matching the 'lists multiple specific concrete actions' anchor rather than the vague 'Names domain and some actions' level 2.	3 / 3
Completeness	It states clearly what the skill does (打分+预测+校准闭环) but provides no 'Use when...' clause or equivalent explicit trigger guidance, which per the judging guidelines caps completeness at 2 ('Has what, but when is missing or only implied').	2 / 3
Trigger Term Quality	Natural terms like 打分/预测/复盘/校准 are present, but they are buried under internal jargon (blind sub-agent, score-only.sh, commit-prediction.sh, 阈值门, published-track) rather than the common variations a user would say; not the clean coverage of level 3.	2 / 3
Distinctiveness Conflict Risk	It carves out a niche and explicitly demarcates from published-track ('发布记录与数据采集由 published-track 统一管理'), but trigger terms like 打分/复盘/校准 could overlap with neighboring content/publishing skills, so it is 'somewhat specific but could still overlap' rather than a clear level-3 niche.	2 / 3
	Total	9 / 12 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning

	Total	14 / 16 Passed

Repository: TeamWiseFlow/xiaobei
Path: crews/main/skills/content-calibrator/SKILL.md
Commit: bcef823

Reviewed: 3 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.