skills.rest
sanjanb
sanjanbCOMMUNITY
Explore all skills by this author →
GitHub

review

Reviews pull request diffs for SQL safety, LLM trust boundaries, and structural issues.

1|Updated Jul 23, 2026
Search Tags:#git#diff analysis#code review#pull request#gstack#sql safety#pre-landing
One-click install
npx skills add https://github.com/sanjanb/my-agent-harness --skill review-sanjanb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/sanjanb/my-agent-harness/tree/main/skills/gstack-review
Command: npx skills add https://github.com/sanjanb/my-agent-harness --skill review-sanjanb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Catching structural defects before code lands is hard under time pressure. This Skill performs a pre-landing review of your diff against the base branch, flagging SQL safety risks, LLM trust boundary violations, conditional side effects, and other structural problems before merge. ## Core Features & Use Cases - Pre-Landing Diff Analysis: Analyzes the current branch diff against the base branch for SQL injection risks, unsafe handling of LLM-generated content, and hidden conditional side effects. - Session-Aware Workflow: Detects interactive, headless, spawned, and Conductor sessions, adapting prompts and decision briefs so reviews work in CI, orchestrators, and local terminals. - Decision Briefs with Recommendations: Presents review findings as structured decision briefs with ELI10 explanations, completeness scores, and clear recommendations. - Use Case: Before merging a feature branch, ask for a review and receive a structured report of structural issues in the diff, with recommended fixes, so nothing unsafe ships. ## Quick Start Ask the agent to review this PR before merging and address any flagged issues in the diff.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a PR diff before merging?▼

Invoke the review skill on your feature branch and it analyzes the diff against the base branch. It flags SQL safety problems, LLM trust boundary violations, and conditional side effects, then presents findings as decision briefs with recommendations.

What structural issues does automated pre-landing review catch?▼

The review targets SQL safety risks, trust boundary violations where LLM output is used unsafely, and conditional side effects hidden in diffs. Findings are reported with ELI10 explanations and completeness-scored options for each decision.

Can this code review run in headless CI or spawned agent sessions?▼

Yes. The preamble detects session kind (interactive, headless, spawned) and Conductor environments. In headless mode it blocks on unavailable prompts; in spawned sessions it auto-chooses recommended options and reports results as prose.

What happens when AskUserQuestion is unavailable during review?▼

The skill follows a documented fallback: headless sessions stop as BLOCKED, spawned sessions auto-select the recommended option, and interactive sessions render the decision brief as prose with completeness scores and a recommendation, then wait for a typed reply.

Does the review skill work outside the gstack installation?▼

No. It depends on gstack binaries such as gstack-config, gstack-telemetry-log, and gstack-repo-mode located under the gstack skills root, plus git for branch and diff detection. Without gstack installed, the preamble commands fail.

Related Skills

Explore More Software Engineering Skills→
benmatselbybenmatselby
68

bms:draft-pr

Create draft pull requests with summaries and PR templates via GitHub CLI.

Community
Advanced
devs6186devs6186

land-and-deploy

Automate pull request merge, production deployment, and canary health verification.

Community
Advanced
CristoVenturesCristoVentures
2

bug-reproduction-validator

Reproduce bug reports and classify issues as Confirmed Bug, Cannot Reproduce, or User Error.

Official
Advanced
Pacha-ePacha-e

plan-eng-review

Review architecture, data flow, edge cases, test coverage, and performance before coding.

Community
Advanced