CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

skill-authoring

Design, improve, and evaluate reusable agent skills with high-quality SKILL.md files, precise trigger descriptions, progressive disclosure, and testable behavior. This skill should be used when users ask to create a new skill, rewrite or review an existing skill, audit a skill collection such as `config/source/skills` for redundancy or overlap, improve skill trigger quality, organize skill references, or evaluate whether a skill should trigger and behave correctly.

Quality

81%

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

SKILL.md
Quality
Evals
Security

Quality

Content

67%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 a well-structured overview with accurate routing to real reference files and concrete commands, but it suffers from redundancy: a security rule stated three times, two overlapping workflow sections, and CloudBase guardrail detail inlined that duplicates its own reference file. Consolidating the workflows and moving the CloudBase block fully into its reference would raise quality across dimensions.

Suggestions

State the remote-skill-fetch prohibition once (in the guardrail list) and reference it from the workflow and self-check instead of repeating it three times, cutting roughly three duplicated passages.

Merge "How to use this skill (for a coding agent)" into the single "Quick workflow" section to remove the overlapping six-step vs eight-step dual sequence.

Compress the inline "Repo-managed CloudBase skill review" guardrail block to a two-line pointer to references/cloudbase-skill-review.md, since the routing table already routes CloudBase review there and the reference covers the same standards in more depth.

DimensionReasoningScore

Conciseness

The body is mostly efficient but includes repeated and tighten-able material, matching anchor 3 ("mostly efficient but includes some unnecessary explanation or could be tightened"). The remote-skill-fetch prohibition is stated three times (guardrail bullet, Quick workflow step 4, and the self-check), "How to use this skill (for a coding agent)" substantially overlaps the "Quick workflow" section, and filler like "This section is the repo-managed CloudBase skill review baseline for this repository." adds no instruction. It is above 2 because there is no explanation of concepts Claude already knows, but below 4 due to the triple-stated rule and dual overlapping workflows.

3 / 5

Actionability

Guidance is mostly executable: a routing table with exact reference paths, concrete commands (`npm run build:skill-manifest`), and specific file paths (`plugin/cloudbase/skill-metadata.json`, `config/source/skills`). It does not reach 5 because key steps stay high-level — "Create should-trigger and should-not-trigger prompts. Run them, review the results" never shows what a prompt looks like or how to run it, and the reference files are pointed to rather than exemplified. It is above 3 because the routing and workflow steps are concrete and specific rather than pseudocode-level hints.

4 / 5

Workflow Clarity

Both workflow sections are clearly numbered and include an evaluation feedback loop ("Run evaluation prompts and revise until trigger quality and behavior are stable") plus a "Minimum self-check" checklist, matching anchor 4 ("clear sequence with most checkpoints present; minor validation gaps"). It does not reach 5 because two overlapping sequences ("How to use this skill" and "Quick workflow") create ambiguity about which to follow, and no explicit iterate-on-failure guidance exists for the evaluation step. No destructive/batch-operation cap applies.

4 / 5

Progressive Disclosure

The routing table cleanly signals seven real, one-level-deep reference files (all exist in references/) and the body stays a navigable overview, matching anchor 4. It falls short of 5 because the ~11-bullet "Repo-managed CloudBase skill review" guardrail block inlines detail that belongs in (and largely duplicates) references/cloudbase-skill-review.md, which the routing table already routes to. It is well above 3 because references are clearly signaled, shallow, and accurately targeted.

4 / 5

Total

15

/

20

Passed

Description

87%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.

A strong description with explicit what-and-when structure, natural trigger phrases, and a well-defined skill-authoring niche. Its only weakness is mild qualifier padding ("high-quality", "precise", "testable") that adds tokens without adding concrete actions or trigger terms.

DimensionReasoningScore

Specificity

The description lists several concrete actions ("create a new skill", "rewrite or review an existing skill", "audit a skill collection ... for redundancy or overlap", "improve skill trigger quality", "organize skill references", "evaluate whether a skill should trigger"), which sits between anchor 4 and 5. It does not fully reach 5 because qualifiers like "high-quality SKILL.md files", "precise trigger descriptions", and "testable behavior" are aspirational adjectives rather than concrete actions; it is well above 3 because coverage spans the full create/review/audit/evaluate lifecycle, not just 1-2 actions.

4 / 5

Completeness

It explicitly answers both questions: the "what" ("Design, improve, and evaluate reusable agent skills with ... progressive disclosure, and testable behavior") and the "when" ("This skill should be used when users ask to create a new skill, ... or evaluate whether a skill should trigger and behave correctly") with concrete trigger phrases, exactly matching the anchor 5 exemplar pattern. It is not 4 because the "when" clause is already fully explicit with enumerated trigger conditions rather than merely improvable.

5 / 5

Trigger Term Quality

Natural phrases users would actually say are present ("create a new skill", "rewrite or review an existing skill", "audit a skill collection", "organize skill references"), matching anchor 4's "good keyword coverage; a few natural terms missing". It falls short of 5 because common variations like "write a skill", "improve my skills", or "SKILL.md" phrasing users might use are not covered, but it clearly exceeds anchor 3's generic keyword level.

4 / 5

Distinctiveness Conflict Risk

The skill has a clear niche — authoring/auditing agent skills — and every trigger phrase is scoped to skills ("create a new skill", "audit a skill collection", "skill trigger quality"), giving minimal conflict risk with unrelated skills. It is not 4 because no meaningful overlap exists: generic documentation or code-review requests are excluded by the skill-specific trigger language. Voice is third-person/declarative ("Design, improve, and evaluate"), so no voice penalty applies.

5 / 5

Total

18

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

frontmatter_unknown_keys

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

Warning

Total

15

/

16

Passed

Repository

TencentCloudBase/CloudBase-AI-Toolkit

Path

skills/skill-authoring/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.