--- title: "Get Run Details" method: GET path: "/api/v1/optimization/runs/{run_id}/details" tags: ["optimization"] --- # Get Run Details `GET /api/v1/optimization/runs/{run_id}/details` Get a detailed GEPA improvement report for a single optimization run. ## Path parameters - `run_id` string, required — Identifier of the optimization run to inspect. ## Response `200` Successful Response - OptimizationRunDetailResponse — Detailed GEPA run report for RLM improvement auditability. - `run` OptimizationRunResponse, required — A single optimization run record. - `id` string, required — Unique run identifier. - `status` string, required — Run status: running, completed, or failed. - `module_slug` string, nullable — Module slug when server-side dispatch was used. - `program_spec` string, required — DSPy program specification that was optimized. - `optimizer` string, required — Optimizer backend that was used. - `auto` string, nullable — Optimization intensity level. - `train_ratio` number — Train/validation split ratio. - `dataset_path` string, nullable — Path to the dataset used. - `reflection_profile_id` string, nullable — LLM provider profile id used for GEPA reflection/proposal, when selected. - `reflection_model_id` string, nullable — Model id used for GEPA reflection/proposal, when selected. - `raw_trace_export_path` string, nullable — Full raw trace export path, when present. - `distilled_trace_bundle_path` string, nullable — Distilled GEPA trace evidence bundle path, when present. - `prompt_snapshot_path` string, nullable — Prompt snapshot or diff artifact path, when present. - `train_examples` integer, nullable — Number of training examples used. - `validation_examples` integer, nullable — Number of validation examples used. - `validation_score` number, nullable — Validation score from the optimized program. - `output_path` string, nullable — Filesystem path where the optimized program was saved. - `manifest_path` string, nullable — Filesystem path to the optimization manifest. - `error` string, nullable — Error message when the run failed. - `phase` string, nullable — Current phase of the optimization run. - `started_at` string, required — ISO timestamp when the run started. - `completed_at` string, nullable — ISO timestamp when the run completed. - `manifest_available` boolean, required — Whether the manifest file was parsed. - `manifest` object, nullable — Parsed optimization manifest, when available. - `review_bundle` object, nullable — Parsed manifest review bundle, when available. - `typed_review_bundle` OptimizationReviewBundle — Typed subset of the manifest review bundle used by the optimization UI. - `version` integer — Review bundle schema version. - `holdout` OptimizationHoldoutSummary — Typed holdout validation summary from a GEPA review bundle. - `promotion_ready` boolean — Whether the run has external holdout validation suitable for promotion. - `external_validation_available` boolean — Whether a true holdout validation split was available. - `baseline_score` number, nullable — Baseline validation score. - `optimized_score` number, nullable — Optimized validation score. - `score_delta` number, nullable — Optimized minus baseline score. - `insights` OptimizationRunInsights — Normalized human-readable improvement insights for a GEPA run. - `selected_outcome` 'changed' | 'unchanged' | 'failed' | 'running' | 'unknown', required — Outcome of the selected GEPA artifact. - `summary` string, required — Short explanation of what GEPA did for this run. - `trace_driven_recommendations` string[] — Recommendations distilled from trace evidence. - `next_step` string, required — Recommended next optimization action. - `artifact_refs` OptimizationArtifactRef[], required — Important run artifact paths. - `label` string, required — Human-readable artifact label. - `path` string, required — Filesystem path to the artifact. - `kind` string, required — Artifact kind such as manifest, output, trace_bundle, or promotion_draft. - `exists` boolean — Whether the artifact exists on the local filesystem. - `score_summary` OptimizationScoreSummary, required — Score and split summary for a GEPA run. - `baseline_score` number, nullable — Baseline validation score, when available. - `optimized_score` number, nullable — Optimized validation score, when available. - `score_delta` number, nullable — Optimized minus baseline score, when available. - `train_examples` integer, nullable — Number of training examples. - `validation_examples` integer, nullable — Number of validation examples. - `train_ratio` number, nullable — Requested train/validation split ratio. - `split_strategy` string, nullable — Dataset split strategy recorded in the manifest. - `prompt_diffs` OptimizationPromptDiffItem[], required — Full before/after prompt snapshots. - `predictor_name` string, required — Predictor or skill component name. - `before_prompt` string — Prompt text before GEPA. - `after_prompt` string — Prompt text selected after GEPA. - `changed` boolean, required — Whether the selected prompt differs semantically from the original text. - `trace_evidence` OptimizationTraceEvidenceItem[], required — Distilled trace evidence records without raw spans. - `kind` string, required — Distilled bundle record kind. - `trace_id` string, nullable — Supporting MLflow trace id. - `session_id` string, nullable — MLflow/runtime session id. - `client_request_id` string, nullable — Client request id, when available. - `trace_count` integer, nullable — Trace count for summary records. - `span_count` integer, nullable — Number of spans in the supporting trace. - `failure_categories` string[] — Distilled failure categories. - `prompt_change_recommendations` string[] — Prompt-change recommendations distilled from trace evidence. - `candidate_decisions` OptimizationCandidateDecision[], required — Selected and rejected candidate decisions when available. - `candidate_id` string, required — Stable candidate identifier for display. - `status` string, required — Candidate status: selected, rejected, unavailable, or failed. - `summary` string, required — Human-readable decision summary. - `rationale` string, nullable — Why this candidate was selected or rejected. - `score` number, nullable — Candidate score, when available. - `score_delta` number, nullable — Candidate score delta, when available. - `artifact_path` string, nullable — Candidate artifact path, when available. - `missing_candidate_artifact` boolean — Whether the proposer generated ideas but no candidate artifact was persisted. - `insights` OptimizationRunInsights, required — Normalized human-readable improvement insights for a GEPA run. - `selected_outcome` 'changed' | 'unchanged' | 'failed' | 'running' | 'unknown', required — Outcome of the selected GEPA artifact. - `summary` string, required — Short explanation of what GEPA did for this run. - `trace_driven_recommendations` string[] — Recommendations distilled from trace evidence. - `next_step` string, required — Recommended next optimization action. - `optimized_artifact_text` string, nullable — Text content of the selected optimized artifact when it is safely readable. - `optimized_artifact_truncated` boolean — Whether optimized_artifact_text was truncated. ## Other responses - `401` — Authentication is required or the provided token is invalid. - `404` — Run not found. - `422` — Validation Error --- [API](https://skmtc.net/qredence/apis/fleet-rlm.md) · [All operations](https://skmtc.net/qredence/apis/fleet-rlm/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/qredence/fleet-rlm/versions/62e8c152aa18/schema)