Skip to main content

Analyze

kindling analyze is the first step in the kindling journey. Run it before kindling generate to check your project's readiness.

kindling analyze
Dashboard

You can also run analysis from the dashboard: Setup → Analyze & Generate, or press ⌘K and type "analyze". See Dashboard for details.

What it checks​

Dockerfiles​

Finds all Dockerfiles in your repo and checks Kaniko compatibility:

  • No BuildKit-only features (TARGETARCH, BUILDPLATFORM, etc.)
  • No RUN --mount syntax (safe to include, but caching is ignored)
  • Go builds should use -buildvcs=false (no .git directory in Kaniko)
  • Poetry builds should use --no-root
  • npm builds should redirect cache: ENV npm_config_cache=/tmp/.npm

Dependencies​

Detects backing services from source code, config files, and Dockerfiles:

  • Databases: Postgres, MySQL, MongoDB
  • Caches: Redis, Memcached
  • Message queues: RabbitMQ, Kafka, NATS
  • Search: Elasticsearch
  • Storage: MinIO (S3-compatible)

Secrets​

Scans for external credentials your app references:

  • *_API_KEY, *_SECRET, *_TOKEN, *_DSN
  • Suggests kindling secrets set for each one

Agent architecture​

Detects AI/ML frameworks and multi-agent patterns:

  • Frameworks: LangChain, CrewAI, LangGraph, OpenAI Agents SDK
  • MCP servers: Model Context Protocol server detection
  • Inter-service calls: HTTP communication between services
  • Concurrent agents: Multiple agent processes that need shared infrastructure

Build context alignment​

Verifies that COPY/ADD paths in each Dockerfile align with the expected build context. Catches mismatches before they cause Kaniko build failures.

Example output​

📁 Repository: /path/to/your-app
──────────────────────────────

🐳 Dockerfiles
✅ ./Dockerfile (Kaniko-compatible)
✅ ./agent/Dockerfile (Kaniko-compatible)

📦 Dependencies
• postgres 16
• redis (latest)

🔑 Secrets
⚠️ OPENAI_API_KEY — run: kindling secrets set OPENAI_API_KEY <value>
⚠️ STRIPE_KEY — run: kindling secrets set STRIPE_KEY <value>

🤖 Agent Frameworks
• LangChain detected in agent/

✅ Ready for 'kindling generate'

Flags​

FlagShortDefaultDescription
--repo-path-r.Path to the local repository to analyze
--verbose-vfalseShow additional detail

After analyze​

Once your project passes analysis:

  1. Set any secrets it detected:

    kindling secrets set OPENAI_API_KEY sk-...
  2. Generate a workflow:

    kindling generate -k <api-key> -r .
  3. Push and deploy:

    git push origin main

Coming soon: scaffold​

For new projects that don't have Dockerfiles yet, kindling scaffold will generate them automatically. Stay tuned.