CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

critical

Adversarial pre-mortem of a plan, code change, or bug diagnosis — cited findings over a fixed failure taxonomy plus a mandatory steelman; never scores or fixes, one pass per run. Modes plan (default), code, analysis; `deep` runs 3–5 parallel persona lenses; triggers on "critical", "challenge this", "pre-mortem", "red-team this", "deep critical review", "review from every angle", "/critical".

Quality

85%

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

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.

An exceptionally well-written body: terse, concrete, with a fixed output contract, mandatory grounding, and an explicit validation rule. Its one serious defect is that the bundle is incomplete — the taxonomy files it instructs Claude to read on every run and the entire deep-mode procedure file are missing, leaving the skill only partially executable as shipped.

Suggestions

Ship the missing bundle files (rules/deep-mode.md and rules/taxonomy-plan.md, taxonomy-code.md, taxonomy-analysis.md) or inline their content — as bundled, "Read the mode's file on every run" and all of deep mode are unexecutable, and by the skill's own grounding rule the unresolvable paths would be must-fix findings against itself.

State the run order once as a short numbered procedure (parse args → detect mode → announce → persona → grounding → taxonomy walk → steelman → emit report) so the sequence is explicit rather than inferred from section order.

Inline a one-line definition per taxonomy row so baseline plan/code/analysis runs remain actionable even when the rule files are not present.

DimensionReasoningScore

Conciseness

The body is lean and directive throughout — a mode-detection table, a five-line persona blockquote, a copy-paste output template, and terse hard rules — with zero explanation of concepts Claude already knows and no padding. It matches the 5 anchor (every token earns its place) rather than 4, since no section could be trimmed without losing load-bearing instruction.

5 / 5

Actionability

Baseline guidance is concrete (exact output structure, steelman template, per-mode grounding commands like "grep for at least two referenced files/symbols"), but the core per-mode content is deferred to files absent from the bundle: "Read the mode's file on every run" points to rules/taxonomy-plan|code|analysis.md, none of which exist, and the entire deep-mode procedure lives only in the missing rules/deep-mode.md. That is more than the "minor gaps" of a 4 — key execution details are unresolvable as bundled — but better than a 2 because the inlined taxonomy row names and templates still make a baseline run partially executable.

3 / 5

Workflow Clarity

The sequence is coherent across ordered sections (parse arguments → detect and announce mode → persona → grounding → taxonomy walk → steelman → fixed output → hard rules) with an explicit validation checkpoint ("A referenced file, symbol, or path that does not resolve is itself a must-fix finding") and a hard-rules checklist. It is not 5 because the ordering is implied by section placement rather than stated as one procedure, and the deep-mode branch cannot be verified since its rule file is missing; it is well above 3 because validation and error paths are explicit.

4 / 5

Progressive Disclosure

Structure and signaling are good — a Contents TOC, clearly labeled one-level references, taxonomy detail split out to per-mode files — but scored against the actual bundle, 4 of the 5 referenced paths (rules/deep-mode.md and the three rules/taxonomy-*.md files) do not exist; only references/why-single-pass.md resolves. Broken reference paths are a bigger defect than the "minor organization gaps" of a 4, but the 2 anchor (inlined bulk content, buried references) does not fit.

3 / 5

Total

15

/

20

Passed

Description

96%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 dense, high-signal description that explicitly states what the skill does, when to invoke it via concrete trigger phrases, its modes, and its non-goals. The only weakness is that the standalone trigger word "critical" is ambiguous enough to create minor conflict risk with general review skills.

Suggestions

Qualify the bare "critical" trigger (e.g., "critical review" or "be critical of this") to reduce false triggers on unrelated uses of the word like "critical bug".

DimensionReasoningScore

Specificity

"Adversarial pre-mortem of a plan, code change, or bug diagnosis — cited findings over a fixed failure taxonomy plus a mandatory steelman; never scores or fixes, one pass per run" names multiple concrete behaviors, the three modes ("plan (default), code, analysis"), and the deep variant ("runs 3–5 parallel persona lenses") — comprehensive coverage of what the skill does. It is not a 4 because no meaningful behavior is left unspecified: mechanism, modes, depth modifier, and constraints are all explicit.

5 / 5

Completeness

Both questions are answered explicitly: what ("cited findings over a fixed failure taxonomy plus a mandatory steelman; never scores or fixes") and when ("triggers on 'critical', 'challenge this', 'pre-mortem', 'red-team this', ..."). This is equivalent explicit trigger guidance, so the missing-'Use when' cap does not apply, and it matches the 5 anchor (both what and when with concrete trigger phrases).

5 / 5

Trigger Term Quality

Explicit natural triggers — "critical", "challenge this", "pre-mortem", "red-team this", "deep critical review", "review from every angle", "/critical" — cover the synonym space users would actually say. It fits the 5 anchor (comprehensive natural terms including synonyms) rather than 4, since the variations span informal ("challenge this"), formal ("pre-mortem"), and command forms.

5 / 5

Distinctiveness Conflict Risk

"never scores or fixes, one pass per run" carves a clear niche apart from scoring/fixing skills, and the adversarial framing is distinct. It is not 5 because the bare trigger "critical" is a common word that could fire in unrelated contexts (e.g., "fix this critical bug"), and "review from every angle" slightly overlaps the code-review skill family — minor overlap risk matching the 4 anchor.

4 / 5

Total

19

/

20

Passed

Validation

81%

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

CriteriaDescriptionResult

metadata_field

'metadata' should map string keys to string values

Warning

frontmatter_unknown_keys

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

Warning

relative_links

Relative link issues: 5 missing

Warning

Total

13

/

16

Passed

Repository

mthines/agent-skills

Path

skills/quality/critical/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.