manage-assessments

You can store plan-level assessments in JSONL format by adding records with fields like hash_id, timestamp, file_path, certainty, and confidence. The system persists them deterministically, enabling consistent auditing and retrieval during plan execution.

Querying assessments by certainty and confidence ranges is done through CLI-based operations that filter JSONL records. You can specify exact certainty values or confidence ranges to retrieve matching entries for Q-Gate validation and analytics workflows.

JSONL assessment tracking works by capturing component evaluations with deterministic storage during plan execution. Agents contribute assessments containing certainty and confidence scores, which are appended as JSONL records for subsequent querying and artifact validation.

Yes, you can filter stored assessments by file path patterns using the query functionality. This allows you to retrieve specific JSONL records matching certain file paths, supporting targeted analysis across plans and projects.

Yes, the JSONL storage supports optional fields including agent, detail, and evidence alongside the required hash_id, timestamp, file_path, certainty, and confidence fields. This enables richer component evaluations for comprehensive auditing.

JSONL storage for component evaluations relies on CLI-based add, query, clear, and get operations without external dependencies. While it provides deterministic persistence, large-scale analytics may require exporting the records to more specialized data processing tools.