No actionable comments were generated in the recent review. 🎉

sequenceDiagram
  participant StorageFlowsE2E
  participant MemoryStorage
  participant CronDocuments
  participant FlowCatalogDocuments
  participant FlowState
  StorageFlowsE2E->>MemoryStorage: install backend
  StorageFlowsE2E->>CronDocuments: create, list, record run, remove
  StorageFlowsE2E->>FlowCatalogDocuments: create and list flow
  StorageFlowsE2E->>FlowState: open state and round-trip KV value
  StorageFlowsE2E->>MemoryStorage: clear backend

Merge Risk: 🔵 Low · up to 7df8d

The storage integration test covers cron, catalog, and flow-state behavior, but not delegation checkpoints; a regression in that persistence path could go undetected, so merge with bounded follow-up to add coverage.

Architecture Summary

Architecture risk: 🔵 Low · up to 7df8d

The change affects 2 systems.

Changed systems: crates, tests

Architecture concerns
No architecture-level concerns identified.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 75.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 84 functions across 11 files. (3 skipped:…	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Description Check		Check skipped - CodeRabbit’s high-level summary is enabled.
Title check		The title clearly and concisely describes the main change: routing cron and flow persistence through storage ports.

---

Full details: Docstring Coverage

Explanation

Docstring coverage is 75.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 84 functions across 11 files. (3 skipped: 3 unsupported.)

---

---

• Fix all pre-merge checks with AI

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{A rabbit checks the flow of bytes,
Through document stores and scheduled nights.
Cron jobs hop from scope to scope,
State finds room to save its hope.
SQLite waits when storage’s away,
And tests keep watch along the way.}

_{Comment @coderabbitai help to get the list of available commands.}

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Tiny Sweeper review

Tiny Sweeper completed its review; deterministic results follow.

State: Changes requested
Priority: critical
Reviewed head: 7df8d512b795
Updated: 1791551041 (Unix time)

Review snapshot

Completeness: Complete
Test assessment: No supported feature-to-test mapping was available; this does not mean tests are absent or passed.

What changed

This pull request routes the cron store, the flow catalog, flow drafts, per-flow engine state, and both checkpointer call sites (delegation graphs and flow runs) onto the configured storage backend when one is present, falling back to the existing SQLite files otherwise. A new FlowState abstraction in the flows tinyflows module picks Sqlite, document-backed, or an explicitly Unavailable store per call, and a new block_on_anyhow helper bridges the synchronous host APIs to the async document stores. A dedicated end-to-end test binary (storage_flows_e2e) drives cron jobs, the flow catalog, flow state and the KV layer against an in-memory backend and asserts the workspace SQLite files are not written. The cron and flows READMEs document the storage-backed behavior, and the storage README lists the newly ported stores plus the block_on_anyhow helper.

Features

• Added — Cron store served by the storage backend: With a storage backend configured, every cron store function is served by tinyflows_drivers::schedule::CronDocuments instead of cron/jobs.db, in the acting agent's storage scope, with the same run-history cap and due-job batch size; reschedule_after_run advances a job only when its stored schedule still matches, so two processes sharing one database do not double-advance it. Without a backend the SQLite path is unchanged. (crates/openhuman-core/src/cron/store.rs#fn opts(config: &Config) -> CronStoreOptions {, crates/openhuman-core/src/cron/store.rs#pub fn add_agent_job(, crates/openhuman-core/src/cron/store.rs#pub fn add_agent_job_with_definition(, crates/openhuman-core/src/cron/store.rs#pub fn add_flow_schedule_job(, crates/openhuman-core/src/cron/store.rs#pub fn add_shell_job(, crates/openhuman-core/src/cron/store.rs#pub fn record_last_run(, crates/openhuman-core/src/cron/store.rs#pub fn record_run(, crates/openhuman-core/src/cron/store.rs#pub fn record_run_with_delivery(, crates/openhuman-core/src/cron/store.rs#pub fn reschedule_after_run(, crates/openhuman-core/src/cron/README.md#missing row, reschedules a drifted one, and removes the retired)
• Added — Flow catalog and drafts served by the storage backend: With a storage backend configured, every flow catalog and draft store function is served by tinyflows_drivers::catalog::FlowCatalogDocuments on the document port in the acting agent's scope instead of flows/flows.db and flows/drafts/; the SQLite file paths remain the fallback when no backend is configured. (crates/openhuman-core/src/flows/store.rs#pub fn create_flow(, crates/openhuman-core/src/flows/store.rs#pub fn update_flow_graph(, crates/openhuman-core/src/flows/store.rs#pub fn insert_flow_run(, crates/openhuman-core/src/flows/store.rs#pub fn finish_flow_run(, crates/openhuman-core/src/flows/store.rs#pub fn expire_parked_runs(, crates/openhuman-core/src/flows/store.rs#pub fn kv_set(, crates/openhuman-core/src/flows/store.rs#pub fn list_running_run_ids(, crates/openhuman-core/src/flows/store.rs#pub fn list_suggestions(, crates/openhuman-core/src/flows/store.rs#pub fn revision_by_id(, crates/openhuman-core/src/flows/store.rs#pub fn force_corrupt_graph_json_for_test(, crates/openhuman-core/src/flows/store.rs#pub fn force_run_status_for_test(, crates/openhuman-core/src/flows/store.rs#pub fn dir(config: &Config) -> PathBuf {, crates/openhuman-core/src/flows/store.rs#use tinyflows_catalog::{, crates/openhuman-core/src/flows/draft_store.rs#pub fn create_draft(, crates/openhuman-core/src/flows/draft_store.rs#pub fn update_draft()
• Added — Per-flow engine state abstraction (FlowState): A new FlowState type picks the store for a flow's namespaced state: SqliteStateStore over flows/flows.db without a backend, FlowStateDocuments in the acting agent's scope with one, or an Unavailable variant that fails every call rather than falling back to the local file when the scope cannot be resolved (SaaS mode with no acting agent). It implements both the engine's StateStore and the dedup settlement's DedupKv, so a run's writes and the post-run dedup settlement read the same records. (crates/openhuman-core/src/flows/tinyflows/state.rs, crates/openhuman-core/src/flows/tinyflows/mod.rs#pub mod caps;, crates/openhuman-core/src/flows/bus/dedup_commit.rs#impl DedupCommitSubscriber {)
• Modified — Dedup settlement shares FlowState with the engine: The dedup commit subscriber settles dedup nodes through FlowState instead of SqliteStateStore, so on a storage backend the settlement reads the same document-backed records the engine wrote during the run. (crates/openhuman-core/src/flows/bus/dedup_commit.rs#impl DedupCommitSubscriber {)
• Modified — Delegation graph checkpoints on the storage backend: Subagent delegation graph checkpoints move from workspace graph_checkpoints.db to tinyagents-graph's DriverCheckpointer over the storage backend's document port (prefixed delegation_graph) in the acting agent's scope, in the acting agent's scope, when a backend is configured; the SQLite checkpointer remains the no-backend path. Reviewers flagged that this new branch currently has no test exercising it. (crates/openhuman-core/src/agent/orchestration/delegation.rs#pub(crate) async fn run_subagent_delegation_with_parent_context()
• Modified — Flow-run checkpointer on the storage backend: The durable cross-process checkpointer a flows_run uses opens as tinyflows_drivers::DriverCheckpointer over the configured backend in the acting agent's scope instead of flows/checkpoints.db; the SQLite checkpointer under the workspace remains the fallback. (crates/openhuman-core/src/flows/tinyflows/caps/ops.rs#pub fn build_capabilities(config: Arc<Config>, state_namespace: impl Into<String, crates/openhuman-core/src/flows/tinyflows/caps/ops.rs#impl ToolInvoker for OpenHumanTools {)
• Added — block_on_anyhow bridge for the new document stores: Adds a storage::block_on_anyhow helper so the synchronous host APIs can drive the async document stores, whose errors are anyhow::Error; typed errors such as FlowUpdateError pass through unchanged for downcasting by callers. (crates/openhuman-core/src/storage/mod.rs#where, crates/openhuman-core/src/storage/README.md#on `storage-mongodb`.)
• Added — Dedicated storage-flows end-to-end test binary: A new storage_flows_e2e test target runs in its own binary because it installs a backend into the process-wide storage slot, which would reroute other suites' stores in a shared process. (crates/openhuman-cli/Cargo.toml#path = "../../tests/storage_approvals_e2e.rs")

Tests

No supported feature-to-test mapping was produced. Test execution is not inferred.

Findings

• critical · critique · Add the referenced storage-flows test source — The new explicit test target points to `../../tests/storage_flows_e2e.rs`, but repository search found no file at that path. Cargo will fail to load or build this package because t (crates/openhuman\-cli/Cargo\.toml:98)
• medium · tests · Assert the SQLite fallback without the conditional skip — This test gates its only assertion on `installed().is_none()`. If some other test in the same binary (or the same process, since the storage slot is process-wide) installs a backen (crates/openhuman\-core/src/flows/tinyflows/state\_tests\.rs:68)
• medium · description · Test the delegation checkpointer on a storage backend — Every other store rerouted by this PR is covered — cron, the flow catalog and flow state in `tests/storage_flows_e2e.rs`, and the `FlowState` variants in `state_tests.rs` — but the (\(pull request description\))
• medium · e2e · Drive the storage-backed delegation checkpointer end to end — The delegation checkpointer now routes to tinyagents-graph's `DriverCheckpointer` when a storage backend is configured, but tests/storage_flows_e2e.rs — the only e2e harness that i (crates/openhuman\-core/src/agent/orchestration/delegation\.rs:91)

Previously reported and still active

• Exercise the storage-backed delegation checkpointer
• Exercise the delegation checkpointer on the document store

Resolved this pass

• Register the storage flows E2E test target
• Own the agent job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the job name before moving it into the future
• Add the storage flows test source before declaring its target
• Test the delegation checkpointer on a storage backend
• Exercise the delegation checkpointer on the storage backend
• Exercise delegation checkpointing on the document store
• Own the shell job name before moving it into the async call
• Exercise cron delegation on the storage backend
• Register the storage flows E2E test target
• Exercise delegation checkpointing in the storage-flow test
• Drive the delegation checkpointer on a storage backend
• Drive the fallback path after clearing the storage backend
• Test the storage-backed delegation checkpointer
• Assert the SQLite fallback without the conditional skip
• Register the storage flows E2E test target
• Own the agent job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the job name before moving it into the future
• Add the storage flows test source before declaring its target
• Test the delegation checkpointer on a storage backend
• Exercise the delegation checkpointer on a storage backend
• Exercise delegation checkpointing on the document store
• Own the shell job name before moving it into the async call
• Exercise cron delegation on the storage backend
• Register the storage flows E2E test target
• Exercise delegation checkpointing in the storage-flow test
• Drive the fallback path after clearing the storage backend
• Assert the SQLite fallback without the conditional skip
• Own the agent job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the job name before moving it into the future
• Own the agent job name before moving it into the future
• Own the shell job name before moving it into the async call
• Own the agent job name before moving it asynchronously
• Own the shell job name before moving it asynchronously
• Add the storage flows test source before declaring its target
• Register the storage flows E2E test target
• Drive the fallback path after clearing the storage backend
• Own the agent job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the job name before moving it into the future
• Own the agent job name before moving it into the future
• Own the shell job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the shell job name before moving it asynchronously
• Own the agent job name before moving it asynchronously
• Add the storage flows test source before declaring its target
• Register the storage flows E2E test target
• Drive the fallback path after clearing the storage backend
• Assert the SQLite fallback without the conditional skip
• Drive the delegation checkpointer on a storage backend end to end
• Own the agent job name before moving it into the async call
• Own the defined agent job name before moving it asynchronously
• Own the job name before moving it into the future
• Own the shell job name before moving it into the async call
• Own the agent job name before moving it into the future
• Own the shell job name before moving it asynchronously
• Own the agent job name before moving it asynchronously
• Add the storage flows test source before declaring its target
• Register the storage flows E2E test target
• Exercise cron delegation on the storage backend
• Drive the fallback path after clearing the storage backend

Pending checks: Rust E2E (mock backend), Build Playwright E2E Artifact, E2E (Playwright / web lane), Desktop E2E (full suite, 3 OS)

Before merge

• Address carried finding Exercise the storage-backed delegation checkpointer.
• Address carried finding Exercise the delegation checkpointer on the document store.
• Address Add the referenced storage-flows test source (crates/openhuman\-cli/Cargo\.toml).
• Wait for Rust E2E (mock backend), Build Playwright E2E Artifact, E2E (Playwright / web lane), Desktop E2E (full suite, 3 OS).

How this fits together

flowchart LR
  n0["flows_create"]:::impacted
  n1["..._expires_stale_runs_but_spares_fresh_ones"]:::impacted
  n2["flows_run"]:::impacted
  n3["flows_resume"]:::impacted
  n4["...t_with_an_incompatible_saved_child_failed"]:::impacted
  n5["structurally_valid_graph"]:::impacted
  n1 -->|calls| n0
  n1 -->|tests| n0
  n1 -->|calls| n2
  n1 -->|tests| n2
  n1 -->|calls| n3
  n1 -->|tests| n3
  n4 -->|calls| n0
  n4 -->|tests| n0
  n4 -->|calls| n2
  n4 -->|tests| n2
  n4 -->|calls| n3
  n4 -->|tests| n3
  n4 -->|calls| n5
  n4 -->|tests| n5
  classDef changed fill:#0d4429,stroke:#238636,color:#e6edf3
  classDef impacted fill:#161b22,stroke:#6e7681,color:#c9d1d9
  classDef flagged fill:#5a1e02,stroke:#d93f0b,color:#ffffff
  classDef blocking fill:#67060c,stroke:#f85149,color:#ffffff