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

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

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.

A strong, highly actionable gate skill: concrete git commands, a complete subagent prompt, and a verdict-driven workflow with explicit checkpoints. The main flaws are a paste-vs-point contradiction between Step 3 and the subagent prompt template, some redundant rationale prose, and inlined testing-workflow detail that belongs in the referenced knowledge file.

Suggestions

Resolve the contradiction between Step 3 ("Point it at the diff command rather than pasting") and the subagent prompt's "<paste full git diff here>" placeholder — pick one mechanism and make the template match.

Move the per-path test-workflow enumeration ("tests/data/ runs wholesale in both unit workflows, tests/ops/ only in the GPU one...") out of the subagent prompt and into .agents/knowledge/testing.md, keeping only a pointer in the checklist.

Trim the opening rationale paragraphs in "When this gate applies" to the rule itself, since the "once per PR, not once per commit" principle is already clear from the heading and table.

Dimension	Reasoning	Score
Conciseness	The body is dense and much of it is non-obvious value (three-dot merge-base semantics, git add -N behavior), but the opening rationale ("the reviewer sees the whole change instead of a slice of it, and a branch of ten commits costs one review rather than ten") re-explains a rule already stated, matching level 4's 'minor instances of over-explanation'.	4 / 5
Actionability	Executable commands (git diff <base>...HEAD, git add -N ., git diff HEAD, make quality), a paste-ready subagent prompt, and a verdict-action table make the guidance mostly copy-paste ready, but Step 3 says "Point it at the diff command rather than pasting" while the prompt template says "<paste full git diff here>" — an internal contradiction and a minor gap.	4 / 5
Workflow Clarity	Five numbered steps with explicit validation checkpoints: the verdict table blocks risky changes ("do NOT open the PR, wait for the user"), needs-attention creates a fix-then-proceed feedback loop, and make quality runs before pushing — matching the level-5 anchor.	5 / 5
Progressive Disclosure	No bundle files exist; the body is well-sectioned with clearly signaled, one-level-deep pointers (.agents/knowledge/constraints.md, .agents/knowledge/testing.md). At ~125 lines with an inlined ~45-line subagent prompt, the testing-workflow enumeration (which workflows own which paths) duplicates content that belongs in the referenced testing.md, matching level 4's 'minor organization gaps'.	4 / 5
	Total	17 / 20 Passed

Description

92%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.

An excellent description that explicitly states what the skill does, when it is required, and when to skip it, with concrete path-level triggers and a clear verdict vocabulary. Its only weaknesses are verbosity for a frontmatter field and a "self-check below" clause that dangles without a body anchor in the description context.

Dimension	Reasoning	Score
Specificity	"launches a subagent that checks implementation quality, multi-file consistency, and known constraint violations, then rates the change as safe/needs-attention/risky" lists multiple specific concrete actions with comprehensive coverage of what the gate does, matching the level-5 anchor; level 4's 'minor gaps in coverage' does not apply.	5 / 5
Completeness	Both questions are answered explicitly: what ("a subagent that checks implementation quality, multi-file consistency, and known constraint violations... rates the change as safe/needs-attention/risky") and when ("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...").	5 / 5
Trigger Term Quality	Natural terms are well covered — "pull request"/"PR", "push", "branch diff", plus concrete file triggers (veomni/, pyproject.toml, uv.lock, docker/, configs/) — but the closing "use the self-check below" clause resolves to nothing below a frontmatter field and a few natural phrasings are missing, keeping it below the comprehensive-synonyms level-5 anchor.	4 / 5
Distinctiveness Conflict Risk	"Pre-PR code review gate" scoped to veomni/, tasks/, tests/, BaseTrainer, and .agents/ carves out a clear niche with distinct triggers and minimal conflict risk with other skills, matching the level-5 anchor.	5 / 5
	Total	19 / 20 Passed

Validation

100%

Warnings & errors only

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
Commit: 86c5ec1

Reviewed: 8 days ago

Table of Contents

Discovery Implementation Validation

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.