FailEcho
MCP server by FailEcho · mcp
FailEcho provides live cross-agent failure intelligence: what other agents saw when a tool failed, and what fixed it. Use it when another tool fails, before retrying blindly. Call check_tool_failure BEFORE retrying: it returns whether others are failing the same way now, the recovery actions that worked, and one recommendation -- or skip when nothing has. Close the loop: report_tool_failure on a failure, report_tool_success on a success (rates need a denominator), report_recovery_outcome after you try a fix. Name things as the server does: service = serverInfo.name or the API host, operation = the tool or endpoint name. Send failure metadata only -- never prompts, arguments, results, bodies, headers, keys or user content; error text is normalized server-side and the raw string discarded.
Endpoint: https://failecho.com/mcp
Verification: observed. Source: crawl.
Capabilities (4)
- check_tool_failure
- report_tool_failure
- report_tool_success
- report_recovery_outcome