Stores and inspection reference

Store-related symbols

Name	Kind	Use when	Key constraints / notes
InMemoryRunStore	Store implementation	The run is local, short-lived, and does not need reopen support	No cross-process persistence
SqliteRunStore	Store implementation	You want the default persistent local backend	Good default for resume, report, compare, and export
RunStore	Store protocol	You are implementing or typing against the storage abstraction	Use for custom backends or shared interfaces
sqlite_store	Store factory helper	You want a concise way to create a SQLite-backed store	Wraps the persistent SQLite backend

Inspection and export helpers

Name	Kind	Use when	Key constraints / notes
get_run_snapshot(store, run_id)	Inspection helper	You want identity and provenance details for a stored run	Reads the compiled snapshot only
get_execution_state(store, run_id)	Inspection helper	You want stage progress, counts, and failure state	Best for resume decisions
get_evaluation_execution(store, run_id, case_id, metric_id)	Inspection helper	You want workflow-backed evaluation details for one case and metric	Applies to judge-backed metrics
Reporter	Reporting API	You want JSON, Markdown, CSV, or LaTeX output from stored runs	Works from persisted projections
snapshot_report	Summary helper	You want a concise snapshot-oriented report in Python	Smaller surface than full Reporter
quickcheck	Status helper	You want a compact run summary	Good for operational checks
Generation and evaluation bundle export/import helpers	Artifact portability helpers	You want portable generation or evaluation artifacts	Also exposed in the CLI for common handoff paths
Reduction, parse, and score bundle export/import helpers	Python-only artifact helpers	You want intermediate-stage handoff beyond the CLI boundary	Currently Python-only

Persistence boundaries

Option	Best for	Persistence / runtime behavior	Caveats
Persistent stores	Resume, reporting, comparison, export, imported-artifact replay, and cache-aware incremental reuse	Preserve artifacts across processes and later sessions	Usually the right choice for real runs
InMemoryRunStore	Short local runs and deterministic smoke tests	Keeps artifacts only in the current process	No cross-run stage cache behavior
Cross-run cache reuse	Reusing stored upstream work over time	Depends on a persistent backend	Not available with memory-only storage

persistent stores are required for cross-run cache reuse. InMemoryRunStore does not provide cross-run stage cache behavior.

Use persistent stores whenever the workflow needs resume, reporting, comparison, export, or later inspection from another process.