AI/Tasks/ExecutionUpgrades/Execution_Phase04_ObservabilityTelemetryRegressionCoverage_Implementation.txt
|
PHASE 04 IMPLEMENTATION PROMPT - OBSERVABILITY, TELEMETRY, AND REGRESSION COVERAGE
Goal: Implement stable, redacted observability contracts and regression coverage for execution-boundary behavior. Inputs: - AI/Tasks/ExecutionUpgrades/Execution_Phase04_ObservabilityTelemetryRegressionCoverage_Strict.txt - Outputs from Phases 01-03 Required code work: - Implement structured error categories and stable error codes with explicit naming conventions. - Implement deterministic precedence for multi-error scenarios. - Implement telemetry event contracts with required fields, correlation identifiers, duration units, and cardinality limits. - Implement runspace lifecycle telemetry for identity, reset status, quarantine state, and process timeout events. - Define and implement redaction policy for paths, arguments, headers, environment values, identifiers, and tool output fragments. - Ensure observability failures are non-fatal unless required safety evidence would be lost. - Implement mandatory regression gates for core safety boundaries and optional extended gates for broader scenarios. Implementation constraints: - Observability components may observe but may not change policy or execution decisions. - Avoid unbounded metric labels or high-cardinality event dimensions. Minimum tests to add/update: - Error-category and code stability tests. - Multi-error precedence tests. - Redaction absence tests (sensitive values must not appear). - Telemetry schema and cardinality tests. - Regression tests for malicious tool names, invalid args, contamination, atomic writes, DNS rebinding, NOAA restrictions, and process-tree containment. Validation commands: - dotnet build src/TechToolbox.Agent/TechToolbox.Agent.csproj -c Release - dotnet test src/TechToolbox.Agent/Tests/TechToolbox.Agent.Tests.csproj -c Release --filter "Telemetry|Error|Category|Code|Redaction|Regression|Containment" Deliverable: Return error taxonomy, telemetry schema, redaction matrix, regression-gate matrix, and validation evidence. |