Prompt Versioning & Runs
Track every prompt iteration and agent execution run. Quantify token savings, detect hallucination events, and analyze latency trends.
Prompt Lifecycle Management
| Tool Name | Purpose |
|---|---|
prompt_template_create | Registers a new system or agent prompt template. |
prompt_version_create | Creates an immutable semantic version of a prompt template. |
prompt_version_activate | Promotes a specific prompt version to active production status. |
prompt_version_history | Lists changelog history for a prompt template. |
prompt_version_compare | Computes side-by-side visual diffs between two prompt versions. |
prompt_version_get | Fetches the raw prompt text for a specific version. |
Run Telemetry & Hallucination Flagging
| Tool Name | Purpose |
|---|---|
run_record | Records execution trace, token consumption, and agent latency. |
run_feedback | Submits user evaluation ratings and flags factual drift. |
run_session_analytics | Computes aggregate latency, cost, and success metrics over time. |
run_prompt_stats | Analyzes performance metrics associated with a specific prompt template. |
run_hallucinations | Retrieves all execution runs flagged for hallucination or contradiction. |
run_get | Inspects full telemetry logs for an individual run ID. |