Document prompt cache control behavior

This commit is contained in:
2026-07-02 23:10:24 +00:00
parent 5dcb3cd4fc
commit 4d4bb7a121
7 changed files with 113 additions and 4 deletions

View File

@@ -108,9 +108,11 @@ Validation content failures are not run errors:
- only the environment-variable name is retained; secret value is never returned
7. resolve output contract and structured-output schema payload when `json_schema` mode is active.
8. read input artifacts.
9. render prompt messages.
9. render prompt messages, including any normalized message cache-control metadata.
10. compute prompt/input/render hashes and return `PreparedRun`.
`rendered_prompt_hash` includes cache-control metadata when present because it affects the outbound provider request. Prompts without cache control keep the role/content hash behavior.
`Prepare` does not call the LLM.
## Run Flow
@@ -123,7 +125,7 @@ Validation content failures are not run errors:
4. build output artifact content type from output format.
5. validate output.
6. optionally attempt bounded repair when repairer is injected and contract allows it.
7. return `RunResult` with artifact, raw output, validation, hashes, profile/model metadata, usage, and timestamps.
7. return `RunResult` with artifact, raw output, validation, hashes, profile/model metadata, token/cache usage, and timestamps.
## Repair Hook Boundary