failure_analysis
Analyze permanently failed tasks to generate defense rules, training cases, and improvement proposals, turning failures into team memory that helps prevent repeat errors.
Instructions
Analyze failed tasks, distill defense rules + training cases + improvement proposals (failure alchemy).
When a task permanently fails (exceeds retry limit), call this tool for deep failure analysis. Automatically generates three learning artifacts saved to team memory:
Antibody: Defensive rule suggestions to prevent similar failures
Vaccine: Structured failure case for new Agents to reference and learn from
Catalyst: System improvement proposals to drive process optimization
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"task_id": "<task_id>",
"team_id": "<team_id>"
},
"name": "failure_analysis"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| task_id | Yes | ID of the failed task | |
| team_id | Yes | ID of the owning team |
Output Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||