CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

veomni-review

Pre-PR code review gate. Run before opening a pull request, and again before pushing a substantive update to an open one — not per commit. Required when the PR's branch diff touches Python under veomni/, tasks/ or tests/, or CI workflows, pyproject.toml, uv.lock, docker/ or configs/. Also trigger proactively for runtime or configuration changes that span multiple files, touch shared infrastructure (BaseTrainer, distributed, model loading, data pipeline, ops dispatch), or have uncertain safety. Docs, comments and .agents/-only changes use the self-check below. The review launches a subagent that checks implementation quality, multi-file consistency, and known constraint violations, then rates the change as safe/needs-attention/risky.

Quality

88%

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

81%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 highly actionable — concrete git commands, a complete copy-paste subagent prompt, and an explicit verdict-to-action table with real validation checkpoints — and the workflow is clearly sequenced. Its main weakness is token efficiency: the anti-bloat rules appear twice and several passages over-explain git behavior Claude already knows.

Suggestions

State the anti-bloat rules once — inside the subagent prompt's 'Anti-Bloat & Proportionality' checklist — and reduce the outer 'Anti-Pattern' section to a two-line pointer to it, cutting roughly 30 duplicated lines.

Trim the git mechanics explanations (three-dot merge-base semantics, `git add -N` intent-to-add behavior, `.gitignore` interaction) to one-line tips; Claude already knows git and only needs the repo's conventions.

Consider moving the VeOmni test-workflow ownership details (which workflow runs which `tests/` paths, file-by-file listing rules) into a reference file under a `references/` directory and linking it from the checklist, shortening the subagent prompt.

DimensionReasoningScore

Conciseness

Quotes: "Use three dots: it diffs from the merge base, so commits that landed on `<base>` after you branched do not show up as your changes" and "`git add -N` is what makes a brand-new module or test visible at all, and it respects `.gitignore`" tutor git mechanics Claude already knows, and the anti-bloat rules are stated twice — once in the body's "Anti-Pattern" section and again nearly verbatim in the subagent prompt's "Anti-Bloat & Proportionality" checklist. Not 4 because the duplication and git tutoring are more than minor trim instances; not 2 because nearly all content is repo-specific, non-obvious knowledge rather than generic padding.

3 / 5

Actionability

Quotes: "git diff <base>...HEAD", "git add -N .", "Run `make quality` before pushing", plus a complete copy-paste subagent prompt and a verdict-to-action table ("risky | Output the report, do NOT open the PR, wait for the user"). Fully executable guidance covering the common cases; no gaps.

5 / 5

Workflow Clarity

Numbered steps 1-5 in clear sequence with explicit validation checkpoints and feedback loops: "Address the listed issues, then open or update the PR", "do NOT open the PR, wait for the user", and the pre-push "Run `make quality`" gate. Not 4 because validation and escalation are explicit decision points, not implied.

5 / 5

Progressive Disclosure

Well-organized sections with one-level-deep, clearly signaled references ("Read `.agents/knowledge/constraints.md`", "See `.agents/knowledge/testing.md`", "The full requirements are in 'Documenting an extension point' in `.agents/skills/veomni-develop/SKILL.md`"); no bundle files exist to verify. Not 5 because the ~60-line Anti-Pattern section duplicates rules already embedded in the subagent prompt — content that could be consolidated or split into a reference file; not 3 because the references that do exist are clearly signaled and navigation is easy.

4 / 5

Total

17

/

20

Passed

Description

91%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: it explicitly answers both what the skill does and when to run it, with concrete repo-specific triggers, skip conditions, and named output verdicts. Its only weaknesses are minor — a leading phrase ('Pre-PR code review gate') that could collide with generic code-review skills, and describing a single capability rather than multiple distinct actions.

DimensionReasoningScore

Specificity

Quotes: "launches a subagent that checks implementation quality, multi-file consistency, and known constraint violations, then rates the change as safe/needs-attention/risky" — several concrete, named actions with explicit output states. Not 5 because it describes one core capability (the review gate) rather than a comprehensive set of distinct actions; not 3 because it goes well beyond 1-2 generic actions.

4 / 5

Completeness

Explicit what ("Pre-PR code review gate... rates the change as safe/needs-attention/risky") and explicit when with concrete triggers ("Run before opening a pull request, and again before pushing a substantive update to an open one"; "Required when the PR's branch diff touches Python under veomni/..."). Not 4 because both what and when are explicit, with exclusion criteria on top; the rubric's cap for a missing 'Use when' clause does not apply.

5 / 5

Trigger Term Quality

Quotes: "pull request", "PR", "push", "branch diff", "CI workflows", "pyproject.toml", "uv.lock", "docker/" — covers natural synonyms (PR / pull request) and file-level triggers, matching the comprehensive anchor pattern ("PDF files, PDFs, forms, document extraction, .pdf"). Not 4 because no common natural variation a user would say is obviously missing.

5 / 5

Distinctiveness Conflict Risk

Repo-scoped triggers ("Python under veomni/, tasks/ or tests/", "BaseTrainer, distributed, model loading") establish a clear niche. Not 5 because the opening "Pre-PR code review gate" broadly overlaps generic code-review skills; not 3 because the required-trigger list is tightly scoped to this repository.

4 / 5

Total

18

/

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

ByteDance-Seed/VeOmni

Path

.agents/skills/veomni-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.