Description

Executive summary: A configurable AI-assisted code review system that standardizes PR quality and reduces reviewer fatigue across multiple languages.

What’s inside (modules):
– Language Rulebooks (Python, JS/TS, Go, Java, C#): style, complexity, testability, readability, error paths, concurrency hazards.
– Risk Heuristics Engine (prompts + JSON policies): flags secret handling, unsafe deserialization, high cyclomatic complexity, flaky test patterns.
– PR Summarization Workflows: CI jobs produce changeset synopsis, module impact, blast radius, and test deltas.
– Reviewer Playbooks: Blocker vs. Warning matrices; checklists for refactors, features, and hotfixes.
– Editor Toolkit: VS Code tasks, quick-fix prompts; optional pre-commit hooks for early feedback.

Technical specifications:
– Artifacts: .md rulebooks, .yaml CI jobs, .json policies, .code-snippets (VS Code).
– LLM-agnostic: env-parameterized prompts; optional tool/function calls.
– Idempotent CI: non-blocking summary job + optional blocking risk-threshold gate.

Setup & integration:
– Add /.github/workflows/review.yml or .gitlab-ci.yml.
– Configure provider secrets and risk thresholds per repo.
– Install VS Code snippets; enable pre-commit hooks.

Security & compliance:
– PII/secret redaction prompts; optional local-only/air-gapped mode.
– Hash-based anonymization for snippets included in LLM calls.

KPIs & ROI:
– 25–40% faster reviews; 10–20% fewer hotfixes; reduced variance between reviewers.

Included files:
/policies/*.json, /workflows/*.yml, /prompts/*.md, /editors/vscode/*.code-snippets, /docs/*