feat(skills): add bundled glm-review skill (cross-model code review via z.ai GLM)#151
Open
vikasiwp wants to merge 1 commit into
Open
feat(skills): add bundled glm-review skill (cross-model code review via z.ai GLM)#151vikasiwp wants to merge 1 commit into
vikasiwp wants to merge 1 commit into
Conversation
…ia z.ai GLM) A fast, one-shot cross-model review of the current repo diff (or a GitHub PR) using z.ai GLM models over their Anthropic-compatible API — an independent second opinion alongside the host reviewer. Self-contained: SKILL.md + glm-review.sh, needs only curl/jq + a ZAI_API_KEY env var (gh for --pr). Reviews uncommitted / --staged / --base <branch> / --pr <N>; prints bugs / security / edge-cases / simplifications most-severe-first. Description-triggered; auto-materialises to agents like any bundled skill. No platform behaviour change — reads a diff, calls an API; the gateway still governs any change made as a result. Live-verified against a planted-bug diff: caught SQL injection, null-deref, and an off-by-one with file:line + fixes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01HiUDC14XfeTQJZHf2wswma
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds a self-contained bundled skill
glm-review— a fast, one-shot cross-model code review using z.ai's GLM models via their Anthropic-compatible API. A different model has different blind spots, so it makes a cheap, independent second opinion alongside the host's own review (e.g./code-review).What it is
config/skills/glm-review/=SKILL.md+glm-review.sh. Fully self-contained:curl+jq(andghonly for--pr). No node/vendor.ZAI_API_KEYfrom the environment (documented; exits with a clear error if unset). OptionalZAI_ANTHROPIC_URLto point at another Anthropic-compatible base.git diff HEAD(default),--staged,--base <branch>(<branch>...HEAD), or--pr <N>. Caps the diff at 120KB. Prints concrete bugs / security / missed edge-cases / simplifications, most-severe-first withfile:line+ a one-line fix (or "no blocking issues");--modeloverrides (defaultglm-4.6),--jsonfor raw.Why bundle it
It's description-triggered ("review with GLM / z.ai / a different model", or before opening/merging a PR) and auto-materialises to agents like any other bundled skill, so a code agent can self-check its diff before a human review. Cross-model agreement raises confidence; disagreement flags a spot worth a closer look.
Safety
No platform behaviour change: the skill only reads a diff and calls an API. Any change an agent makes as a result still passes the gateway.
SKILL.mdtells the agent to treat the output as a second opinion to verify, not a verdict.Verification
Live-tested against a planted-bug diff — it caught the SQL injection, the null-dereference, and the off-by-one, each with
file:line+ a fix. A clean diff returns "no blocking issues". All three GLM models (glm-4.6/4.7/5.2) return 200 on the endpoint.🤖 Generated with Claude Code
https://claude.ai/code/session_01HiUDC14XfeTQJZHf2wswma