CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

factory-review

Adversarially review one ready-for-review ticket task in a read-only Pi subagent using a model different from the worker, then return a structured markdown findings list and a verdict. Use only for the review stage; it never edits code or applies its own findings.

Quality

84%

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

88%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 exemplary orchestration skill body: a lean three-phase workflow with concrete blocked conditions, exact template and tool-call names, and well-signaled references to real bundle files. The only deductions are minor — slight repetition of the read-only rules and a second-hop reference (review-contract.md) that is not linked directly from the overview.

DimensionReasoningScore

Conciseness

The body is lean and directive with no concept explanations — "Determine whether the implementation solves the stated problem. Be adversarial about evidence." — and it assumes competence throughout. Not 5 because the read-only rule is restated ("Remain read-only. Never edit code, tests, plans, or state." followed shortly by "Never edit or write.") and the output-format rules duplicate reviewer-prompt.md's Output section, so a few tokens could be trimmed.

4 / 5

Actionability

For an instruction-only skill the guidance is fully executable: exact decision rules ("If the model matches the work model, return `blocked`", "If the packet or diff file is missing and cannot be read, return `blocked` and name the missing input"), an exact artifact name ("the `factory.review.v5` template"), and exact tool-call semantics for `subagent_done` and `caller_ping` including the shutdown-reply stop condition. A subagent can follow this without guessing.

5 / 5

Workflow Clarity

A clear three-phase sequence (Isolate the session → Review the task → Return the verdict) with explicit validation checkpoints and error-recovery paths: the blocked conditions are checked before review begins and require naming the missing input, and "If it replies that the session is shutting down, stop. Do not call it again" handles the terminal failure case. The operation is read-only, so the destructive/batch validation cap does not apply.

5 / 5

Progressive Disclosure

All three referenced files exist and are well-signaled one level deep with purpose statements ("Read [references/reviewer-prompt.md] and follow it", "apply [references/code-quality.md]", "the `factory.review.v5` template in [references/output.md]"). Not 5 because review-contract.md is only reachable via a second hop through reviewer-prompt.md and is never linked from SKILL.md, and the output rules are duplicated between SKILL.md and reviewer-prompt.md rather than split cleanly.

4 / 5

Total

18

/

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.

A strong, specific description that clearly states what the skill does, when to use it, and what it must never do, in third person with no fluff. Its main weakness is trigger coverage: it relies on narrow internal jargon and a single trigger condition rather than natural variations a requester might use.

Suggestions

Add natural trigger variations such as "code review", "review the diff", or "review the ticket" so the description matches how a requester actually phrases the need.

Expand the when clause with a concrete trigger condition, e.g. "Use when a ticket task is marked ready-for-review and needs an independent review pass", to make the trigger context more explicit.

DimensionReasoningScore

Specificity

Names several concrete actions — "Adversarially review one ready-for-review ticket task", "return a structured markdown findings list and a verdict", "it never edits code or applies its own findings" — with only minor gaps. Not 5 because what the review actually examines (acceptance criteria, code quality, security) is not covered in the description.

4 / 5

Completeness

Both parts are explicit: a clear what ("Adversarially review... in a read-only Pi subagent using a model different from the worker, then return a structured markdown findings list and a verdict") and an explicit when ("Use only for the review stage"). Not 5 because the when clause is a single scoping statement without the varied concrete trigger phrases of the 5 anchor.

4 / 5

Trigger Term Quality

Relevant keywords like "review", "ready-for-review ticket task", "findings", "verdict", and "review stage" are present, but the phrasing leans on internal jargon ("Pi subagent", "the worker") and offers no natural variations or synonyms (e.g. "code review", "review the diff"). This matches anchor 3 (some relevant keywords, missing common variations) better than anchor 4's good-but-not-full coverage.

3 / 5

Distinctiveness Conflict Risk

"read-only Pi subagent using a model different from the worker" and "Use only for the review stage; it never edits code or applies its own findings" carve out a clear niche with distinct triggers and explicit non-goals, giving minimal overlap risk with general code-review or editing skills.

5 / 5

Total

16

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

geut/factory-skills

Path

skills/factory-review/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.