kaddo scan

kaddo scan

Detects language, framework, package manager, code dirs, migration dirs, contract files, infra and test dirs. Suggests domains for human confirmation — never assumes.

Generated artifacts

kaddo scan persists a reusable baseline of the project’s technical state:

• .kaddo/scan.json — structured, machine-readable. Used by the CLI and future context-pack commands. Always regenerated on each scan.
• knowledge/inventory.md — human-readable inventory you can paste into an LLM chat. Overwrite is confirmed if the file already exists.

This baseline is the first input for the Knowledge Driven Development flow: the CLI prepares deterministic signals, and your LLM agents turn them into understanding.

Scan does not interpret your system. It detects signals and asks confirmation questions — it never claims to know your business capabilities or architecture.

State-aware next step

If .kaddo/config.yml exists, kaddo scan reads the project state recorded by kaddo init and prints a next step tailored to it:

• new → define initial knowledge, create your first work item, grow the roadmap gradually.
• pre-ai → use this baseline to create a context pack and understand the system with LLM agents.
• legacy → use this baseline to identify risks, unknowns and safe modernization candidates.

This is a pre-AI monorepo.
Next: Use this baseline to create a context pack and understand the existing system with LLM agents.

Scan Signals (v3.48.0)

Starting with v3.48.0, kaddo scan detects actionable signals across 14 categories:

Category	Examples
auth	NextAuth, Supabase Auth, Clerk, Passport
payments	Stripe, Mercado Pago, PayPal, Wompi
webhooks	Webhook route directories
storage	AWS S3, Cloudinary, Uploadthing
background_jobs	BullMQ, Supabase Edge Functions, cron dirs
email	Resend, SendGrid, Nodemailer
database	Prisma, Drizzle, Supabase, Mongoose
migrations	Migration directories detected by scan
api_routes	Next.js API routes, Express routers
tests	Vitest, Jest, test directories (warns if missing)
security	Helmet, CORS, RLS policies
infrastructure	Docker, Amplify, Vercel, GitHub Actions
external_integrations	Sentry, Redis, Algolia, Firebase
environment	Environment variable names from .env.example

Each signal includes a confidence level (high / medium / low) and evidence (dependency names, file paths). Some signals include a recommended_review suggestion.

Signals appear in:

• Console output after each scan
• knowledge/inventory.md under “Detected Signals”
• .kaddo/scan.json in the signals field
• Context pack and explain output
• MCP via kaddo://scan-signals

Security: environment detection reads variable names only — values are never stored, logged, or exposed.

Example knowledge/inventory.md:

# Project Inventory

## Detected Stack

- Language: typescript
- Framework: next
- Package manager: pnpm

## Possible Domains

- auth
- payments

## Open Questions

- Confirm whether the 'payments' domain reflects a real bounded context.