AI-powered PR review with intent checking against Azure Boards and Jira work items, Confidence-scored Findings, and an interactive Approval Loop.
1 on GitHubby Unic AGv2.1.12LGPL-3.0-or-laterupdated 11 days ago
/
Plugin unic-pr-review from marketplace unic-agent-plugins in unic/unic-agents-plugins. Commands read from the repo's manifest — 9 days ago.
What it installs
Slash commands (5)
/unic-pr-review:doctor
Verify unic-pr-review prerequisites — az CLI, Azure DevOps extension, login state, Confluence, and Jira
/unic-pr-review:review-pr
Review a pull request or your local branch. Pass an Azure DevOps PR URL to review an open PR; add --post to enter the Approval Loop and write to ADO; omit the URL to review your local branch against its upstream base (Pre-PR mode).
/unic-pr-review:setup-azure
Interactive setup wizard — writes ~/.unic-azure.json with Azure DevOps credentials
/unic-pr-review:setup-confluence
Interactive setup wizard — writes ~/.unic-confluence.json with Confluence credentials
/unic-pr-review:setup-jira
Interactive setup wizard — adds jiraUrl to ~/.unic-confluence.json
Subagents (11)
ado-fetcher
ADO Fetcher — reads all PR data from Azure DevOps via az devops invoke. Fetches PR metadata, Revisions, Threads, and the changed-file list. Computes a checkout-free merge-base diff (commonRefCommit→sourceRefCommit) for first-review modes via git fetch + git diff; falls back to diffUnavailable when n
ado-writer
ADO Writer — posts approved Findings as inline Review Threads and the Review Summary as a General Comment Thread to Azure DevOps via az devops invoke.
code-reviewer
Code Reviewer — analyses the diff for correctness, style, and maintainability issues. Emits structured Findings with Confidence Scores.
code-simplifier
Code Simplifier — identifies opportunities to reduce complexity and eliminate unnecessary code in the diff. Emits structured Findings with Confidence Scores.
comment-analyzer
Comment Analyzer — scans the diff for comment rot, inaccurate documentation, and over-documentation antipatterns. Emits structured Findings with Confidence Scores.
intent-assessor
Intent Assessor — assesses each Acceptance Criterion for diff coverage and returns the intentCheck skeleton with verdicts filled in. NOT a Review Aspect — spawned by intent presence (intentBrief defined + skeleton non-empty), not changed-file categories. Never add to SPAWN_TABLE.
intent-checker
Intent Checker — fetches and synthesises Work Item intent from pasted Jira and Confluence URLs. Emits a structured Intent Brief plus per-AC verdicts.
pr-test-analyzer
PR Test Analyzer — checks the diff for test-coverage gaps, missing edge cases, and shallow assertions. Emits structured Findings with Confidence Scores.
re-review-coordinator
Re-review Coordinator — classifies existing PR Threads into addressed/disputed/pending/obsolete/regressed, decides reply vs resolve vs reopen vs new-Thread per Finding, and emits a structured plan for the ADO Writer.
silent-failure-hunter
Silent Failure Hunter — scans the diff for swallowed errors, empty catch blocks, and inappropriate fallback patterns. Emits structured Findings with Confidence Scores.
type-design-analyzer
Type Design Analyzer — inspects the diff for weakly-encapsulated types and missing invariants. Emits structured Findings with Confidence Scores.
Adversarial review of web specifications across Confluence pages & comments, Figma designs, and the live system. Produces Confidence-scored, Six-Hats-tagged Findings with an interactive Approval Loop that posts selected comments back to Confluence.
[DEPRECATED — being replaced, see #677] A config-driven Harness for the SDLC — the AI development lifecycle — as an installable Claude Code plugin — main line /specs → /tickets → /build → /pr-review → /qa, plus /triage, /explore, /improve-architecture, and /cleanup. Four Archon workflows (build, qa, pr-review, explore) with human approval gates; the rest are commands/skills composing Matt Pocock's methods.
Collects what a team learns about its repository from Claude Code sessions. A Stop hook says when learnings are due; a read-only subagent drafts .claude/rules/learned.md, and nothing is written without your yes.
Keeps Claude concise. Ships the Terse Orwell output style, generic behaviour rules for every session and subagent, a per-turn language reminder, and /setup to write your Personal layer.
Airtable is the database and operations layer for your agents — whether running product, marketing, sales, ops, HR, or a custom business app. It combines structured data with multiplayer visual surfaces (grid, kanban, calendar, gallery, timeline) humans and agents share — plus sync integrations to Jira, Salesforce, Zendesk, Google Drive, Databricks, and the rest of your stack, all backed by enterprise governance. This plugin makes Claude fluent in Airtable: creating bases and schema, working with records, and sharing UI for collaboration. Bundles the official Airtable MCP server.
Require a human approval signal before an AI agent can post PR reviews, comments, merges, or writes to CI configuration. Joins protect-mcp and signed-audit-trails in the governance category; composes with protect-mcp for runtime enforcement.
Airtable is the database and operations layer for your agents — whether running product, marketing, sales, ops, HR, or a custom business app. It combines structured data with multiplayer visual surfaces (grid, kanban, calendar, gallery, timeline) humans and agents share — plus sync integrations to Jira, Salesforce, Zendesk, Google Drive, and Databricks. Makes Claude fluent in Airtable: bases and schema, records, and collaboration UI. Bundles the official Airtable MCP server.
Turns OpenAI Codex into a pool of parallel agent workers inside Claude Code. Each worker runs in its own git worktree so parallel edits don't collide. Resumable sessions via Codex's native thread_id, dual-model PR review with materialized detached worktree, and role-based specialization (implementer/reviewer/planner/generic).
Generate production-grade engineering design diagrams (architecture, sequence, delta, drift) as self-contained dark-themed HTML files with accessible inline SVG. Grounds every diagram in real repo topology via DCI — package manifests, docker-compose, k8s, terraform, import graph. Four modes: generate, diff, trace, watch. Semantic OKLCH palette; Mermaid fallback for large graphs.
1 skill
jeremylongshore/tons-of-skills-marketplace 2,917
Maintain unic-pr-review?
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.