classify_documents
Classifies project repo documents by semantic similarity to generate a migration plan for moving them into the knowledge base.
Instructions
Classify project repo documents for migration into the knowledge base. Read-only.
Does not write any files. The agent reads project files locally and
passes their content here; the Docker container cannot access the
project source repo directly. Flaiwheel classifies each file by
semantic similarity and returns a migration plan.
Trigger: user says "This is the Way" or "42".
Step 1 of the migration workflow — after classification, use the
suggested write_*() tool for each file to push it into the knowledge base.
Use analyze_knowledge_repo() instead when files are already inside the
knowledge repo and need reorganisation.
Args:
files: JSON array of {"path": "...", "content": "..."} objects.
Send the first ~2000 characters of each file as content.
Example: [{"path": "docs/auth.md", "content": "# Auth..."}]
project: Target project name (optional)
Returns:
Per-file classification (category, suggested write_*() tool),
duplicate detection, and a step-by-step migration plan.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| files | Yes | ||
| project | No |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |