Kanon documentation
Kanon turns your web application's codebase into a living, verified knowledge base. It extracts what your product actually does — business rules, data flows, edge cases, security posture, and operational details — directly from the source code, and keeps it current as the code changes.
Who this is for
- Founders and executives who need to understand what their product actually does without reading code
- Product managers who want to know every edge case, threshold, and business rule
- Engineers who need a reliable system reference when onboarding, debugging, or planning changes
- QA teams who need comprehensive test plans grounded in actual code behavior
What Kanon produces
- Feature guides — comprehensive, grounded documentation for each product feature: business narrative, design principles, lifecycle flows, data model, routes, jobs, flags, events, security findings, test plans, and more
- Tracking plan — analytics event inventory reconciled against the code, with gap detection
- Flag plan — feature flag lifecycle management with staleness detection
- Experiments page — A/B test inventory with variant, hypothesis, and status tracking
- Security audit — static analysis findings mapped to OWASP categories
- Testing plan — E2E test scenarios generated from the knowledge base
- AI chat — ask questions about your product and get grounded answers with Linear ticket drafting
How it works
- Install the Claude Code plugin — the Kanon plugin runs inside Claude Code
- Discover — analyze your codebase structure (routes, navigation, schemas, guards) to build a taxonomy of domains and features
- Approve — review the proposed taxonomy in the web app and approve/reject
- Scan — deep-scan each feature: extract claims, synthesize guides, collect facts
- Use — read the knowledge base, ask the AI chat, manage tracking plans, work on changes
- Execute — keep a worker daemon running and pending changes become PRs straight from the UI — no terminal needed