Role in the Ecosystem
Learning MCP is the Continuous Improvement Layer of the VISHKAR ecosystem. It captures implementation learnings from AI agents during development, enables human review, and feeds approved insights back into WAMA (agent knowledge base) to improve future work.
- ✓ Log learnings with epic, category & severity
- ✓ Query by project, category, or review status
- ✓ Human approval workflow via /approve UI
- ✓ Statistics & insights by category and status
Learning Flow
Agent logs learning→
Stored in DB→
Human reviews→
Updates WAMA
3 MCP Tools
log_learning
Log a learning from implementation work with epic, category, severity, and action taken.
epic: e.g. V2-22
category: process / code / tooling
severity: low / medium / high
get_learnings
Query logged learnings by epic, category, or review status. Returns structured list for human review.
status: pending / approved / rejected
category: optional filter
get_learning_stats
Get aggregate statistics — counts by category, status, and severity across all projects.
Useful for team retrospectives
Identify recurring pain points
Usage Example
Log a Learning
curl -X POST https://learning-mcp.qiplabs.ai/api/mcp \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"name": "log_learning",
"arguments": {
"epic": "V2-22",
"category": "process",
"learning": "PR thread resolution requires GraphQL mutation",
"action_taken": "Used resolveReviewThread mutation instead of REST",
"severity": "high"
}
},
"id": 1
}'