Optional skill. Reconstruct a human-review-preparation file from an existing pull request, merge request, branch diff, or commit range in a repository the user trusts. Use when the user wants retrospective understanding of already-implemented changes, AI-side assessment and recommendations, and an optional provider-specific sharing variant written to a local file when needed.
48
51%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./skills/spec-loop-assess-pull-request/SKILL.mdQuality
Discovery
67%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 provides a reasonable 'what' and an explicit 'when' clause, which is its strongest aspect. However, the specificity of actions is moderate—terms like 'human-review-preparation file' and 'provider-specific sharing variant' are domain-specific jargon that may not resonate with users. Adding common shorthand trigger terms (PR, MR, code review) and more concrete action descriptions would improve discoverability and clarity.
Suggestions
Add common shorthand trigger terms like 'PR', 'MR', 'code review', 'review summary' to improve matching with natural user language.
Replace abstract phrases like 'human-review-preparation file' and 'provider-specific sharing variant' with more concrete descriptions of what is actually produced (e.g., 'generates a markdown summary of changes with inline comments and recommendations').
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description names a domain (pull requests, merge requests, diffs) and some actions (reconstruct a review-preparation file, AI-side assessment, recommendations, sharing variant), but the actions are somewhat abstract and not concretely enumerated. Terms like 'human-review-preparation file' and 'provider-specific sharing variant' are vague without further explanation. | 2 / 3 |
Completeness | The description answers both 'what' (reconstruct a review-preparation file with assessment and recommendations) and 'when' ('Use when the user wants retrospective understanding of already-implemented changes, AI-side assessment and recommendations, and an optional provider-specific sharing variant'). The 'Use when' clause is explicit. | 3 / 3 |
Trigger Term Quality | Includes relevant terms like 'pull request', 'merge request', 'branch diff', 'commit range', and 'repository', which users might naturally say. However, it misses common shorthand like 'PR', 'MR', 'code review', 'diff review', and the phrasing is somewhat formal rather than matching natural user language. | 2 / 3 |
Distinctiveness Conflict Risk | The focus on retrospective review of existing changes and generating a local review-preparation file is somewhat distinctive, but terms like 'pull request', 'merge request', and 'diff' could overlap with other code review or git-related skills. The concept of a 'human-review-preparation file' is unique but not clearly enough defined to fully prevent conflicts. | 2 / 3 |
Total | 9 / 12 Passed |
Implementation
35%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill functions primarily as an orchestration entry-point that delegates nearly all substantive guidance to external files. While the delegation strategy is reasonable, the SKILL.md itself provides almost no actionable, executable content—no example commands, no sample output structure, no concrete workflow steps. The result is a skill that tells Claude what to read rather than what to do, making it difficult to follow without the referenced files.
Suggestions
Add at least one concrete, executable example showing the actual gh/glab commands used to gather evidence (e.g., `gh pr diff 123`, `gh pr view 123 --json`).
Include a numbered workflow with explicit steps: 1) detect provider, 2) fetch diff, 3) reconstruct review artifact, 4) validate artifact structure, 5) optionally create sharing variant—with validation checkpoints.
Show a minimal skeleton of the expected review artifact output so Claude knows the target structure without needing to read review-guidance.md first.
Remove the redundant summary of what review-guidance.md covers ('Review outcome labels, Review Area behavior, assessment style...') since Claude will read that file directly.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is moderately efficient but includes some redundancy—e.g., restating what review-guidance.md covers in a bullet list and then again in a summary line. The repeated clarifications about trust boundaries and what not to apply add some bloat, though most content is relevant. | 2 / 3 |
Actionability | The skill provides no concrete commands, code examples, or executable steps. It is almost entirely meta-orchestration—telling Claude which files to read and which rules to follow—without showing what the actual review artifact looks like, what commands to run, or what output to produce. | 1 / 3 |
Workflow Clarity | There is a rough sequence (read prerequisites, detect provider, gather evidence, write artifact, optionally create sharing variant), but it lacks explicit numbered steps, validation checkpoints, or error-recovery loops. The workflow is implicit rather than clearly sequenced. | 2 / 3 |
Progressive Disclosure | The skill references several external files (review-guidance.md, common-task-guidance.md, task-file-path-guidance.md, example file) which is good progressive disclosure in principle. However, no bundle files were provided to verify these references, and the SKILL.md itself is somewhat unclear about what lives where—the delineation between SKILL.md and review-guidance.md responsibilities is stated but the navigation could be cleaner with a structured reference table. | 2 / 3 |
Total | 7 / 12 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 — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
90f996e
Table of Contents
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.