Knowledge Base
Store project documentation in the cloud. The AI retrieves relevant chunks during generation, ensuring output aligns with your team's conventions.
Cloud Documents
Documents are chunked and indexed for semantic search. Each document:
- Belongs to a team and optionally a project
- Has a name, content, and metadata
- Is searched automatically during generation based on spec + query context
Operations (via Remote MCP)
| Operation | MCP Tool |
|---|---|
| Upload | remote_ingest_cloud_document |
| Search | remote_cloud_search |
| List | remote_list_cloud_documents |
| Get | remote_get_cloud_document |
| Update | remote_update_cloud_document |
| Delete | remote_delete_cloud_document |
Remote MCP
curl -X POST "https://api.aexol.ai/mcp" \
-H "Authorization: Bearer sk-aexol-team-..." \
-d '{
"jsonrpc": "2.0", "id": 1, "method": "tools/call",
"params": {
"name": "remote_ingest_cloud_document",
"arguments": {
"name": "api-design.md",
"content": "# API Design Guidelines\n\nAll endpoints use REST..."
}
}
}'AI Retrieval Flow
Specification
Your .aexol file
Parse to AST
Syntax tree
AI Generation
Model + KB context
Output Code
TS, Python, Rust, Go, JS
Generated code follows your team's conventions, API patterns, and naming standards without explicit prompting.
Permissions
Admin
- Full control
- Settings & billing
- Manage members
- All projects
- Full KB access
Moderator
- Manage projects/members
- Read + Write KB
- Can't change settings
- Can't manage billing
Member
- Access shared projects
- Generate code
- Read-only KB
- No project management
What to Upload
Good knowledge documents:
- API specifications, database schemas
- Coding standards, naming conventions
- Product requirements, domain knowledge
- Plain text or Markdown works best
Next Steps
- Code Generation — How KB feeds into generation
- Refinement — Iterative improvement with KB context