Nicolò Boschi 2cd0561f14 feat(metrics): /metrics covers every worker (labelled api_worker), and event-loop lag as a histogram (#4319) ...

* feat(metrics): a metrics port per worker, and event-loop lag as a histogram

With --workers N every worker is its own process with its own metrics, but they
share one port, so a scrape of /metrics reaches one worker at random. Counters
jump between processes from one scrape to the next (a rate over them reads each
switch as a reset), and process_cpu_seconds_total describes a random worker, so
a worker whose event loop is saturated is invisible while the pod total still
looks like headroom.

- HINDSIGHT_API_METRICS_WORKER_BASE_PORT (default 0, off): each worker also
  serves its own registry on BASE + slot. The slot (0..N-1) is claimed with an
  exclusive flock on a per-slot lock file; the kernel drops it when the process
  exits, so a respawned worker takes over its predecessor's port. /metrics on
  the API port is unchanged.
- HINDSIGHT_API_LOOP_LAG_METRIC (default false): the existing loop-lag probe
  records every sample in a hindsight.event_loop.lag histogram (seconds, with
  sub-second buckets), independent of its log reports.

* feat(metrics): one /metrics covering every worker, labelled api_worker=<slot>

Replaces the per-worker ports from the previous commit. Labelling alone would
not fix the scrape: each scrape still reaches one worker, the others' series go
missing from about half the scrapes, and Prometheus marks them stale. So with
HINDSIGHT_API_METRICS_WORKER_LABEL on, each worker claims a slot (flock on a
per-slot lock file; the kernel frees it when the process exits) and publishes a
snapshot of its registry every 5 s to a directory the server's workers share.
/metrics, whichever worker answers, returns every live worker's series with
api_worker="<slot>": its own read live, the others from their latest snapshot,
never summed. A snapshot older than 15 s is a gone worker and is skipped; a
corrupt one is skipped without breaking the scrape.

One port and one scrape target, so no change to charts or scrape configs. The
event-loop lag histogram from the previous commit is unchanged.

2026-09-14 09:52:57 +02:00

..

data

test(temporal): re-snapshot the golden corpus for the "last weekend" fix (#4192)

2026-09-07 13:22:14 +02:00

fixtures

feat: introduce hindsight-api-slim and hindsight-all-slim packages (#560)

2026-03-13 13:50:03 +01:00

__init__.py

feat: introduce hindsight-api-slim and hindsight-all-slim packages (#560)

2026-03-13 13:50:03 +01:00

aiohttp_stub.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

chunking_reference.py

refactor(retain): format pgvector literals without orjson, and drop orjson entirely (#4040)

2026-09-02 17:58:22 +02:00

codex_stream_stub.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

conftest.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

consolidation_actions.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

e2e_oracle_smoke.py

fix: report total on the mental-model and directive list endpoints (#3548)

2026-08-17 16:02:09 +02:00

entity_resolution_cases.jsonl

fix(entities): don't merge a name onto an existing entity on history alone, + a resolution eval (#3752)

2026-08-24 13:45:01 +02:00

llm_judge.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

ollama_stub.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

pg_extension_fakes.py

fix(api): install every PostgreSQL extension in the public schema (#4118) (#4121)

2026-09-04 12:54:29 +02:00

query_analyzer_bench.py

perf(recall): make temporal extraction ~9x faster without changing behaviour (#3452)

2026-08-13 11:17:16 +02:00

query_analyzer_corpus.py

perf(recall): make temporal extraction ~9x faster without changing behaviour (#3452)

2026-08-13 11:17:16 +02:00

sub_batch_helpers.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_admin_backup_restore.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_admin_bank_transfer.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_agents_api.py

fix(api): paginate the bank list (#3586)

2026-08-18 17:27:27 +02:00

test_aiohttp_session.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_alembic_dag.py

fix(alembic): merge divergent heads for v0.5.3 (#1149)

2026-04-21 13:22:58 +02:00

test_ann_iterative_scan.py

perf(db): skip asyncpg's release-time reset (#4259)

2026-09-10 17:36:29 +02:00

test_anthropic_batch.py

feat(anthropic): carry the prompt-cache marker on batch system prompts (#2652)

2026-07-20 13:50:31 +02:00

test_anthropic_prompt_caching.py

fix(anthropic): group consecutive tool results into one user message (#3910)

2026-08-31 15:59:53 +02:00

test_anthropic_structured_tool_use.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_append_body_invariants.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_append_coalesces.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_async_batch_retain.py

fix(retain): scrub the whole retain item at the engine ingress (#3908) (#3921)

2026-09-07 11:04:44 +02:00

test_async_retain_operation_id.py

fix(retain): make async retries idempotent via caller-supplied operation_id (#2937) (#2947)

2026-07-24 13:43:57 +02:00

test_async_retain_tags.py

refactor(api): unify lazy bank-create into _ensure_bank_exists, couple to caller txn (#2004)

2026-06-05 12:47:23 +02:00

test_async_submit_bank_not_found.py

fix(async-op): return 404 when bank doesn't exist instead of raw FK 500 (#2352)

2026-06-23 10:56:44 +02:00

test_attachment_attribution_llm.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_attribution.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_aware_chunking.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_canonicalization.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_identity.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_lifecycle.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_attachment_read_surfaces.py

feat(api,control-plane): recall results carry the attachments behind each fact (#4277)

2026-09-09 18:37:52 +02:00

test_attachments_store_owned.py

feat(api): attachments on recall and read surfaces for store-owned banks (#4321)

2026-09-11 16:19:01 +02:00

test_audit_log.py

feat(audit): make audit_log_enabled overridable per bank (#2827)

2026-07-21 11:37:46 +02:00

test_audit_per_bank.py

fix(oracle): audit_log write qualification + llm_requests read gating (#3015)

2026-07-28 12:36:32 +02:00

test_backlog_metrics.py

fix(stats): drop permanently failed memories from pending_consolidation (#3362) (#3397)

2026-08-11 18:23:33 +02:00

test_bank_attribution_config.py

feat(reranker): failover chain via indexed HINDSIGHT_API_RERANKER_<n>_* members (#3176)

2026-08-04 19:00:27 +02:00

test_bank_attribution.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_bank_config_atomicity.py

fix(config): validate and persist bank config as one serialized unit (#3037) (#3639)

2026-08-19 23:55:15 +02:00

test_bank_config_value_types.py

fix(config): validate bank config value types and stop them wedging tasks (#3218) (#3319)

2026-08-10 13:04:45 +02:00

test_bank_health.py

feat(api): omit null fields from JSON responses where wire-safe (#2204)

2026-06-15 13:50:01 +02:00

test_bank_info_cache_invalidation.py

config: keep the hot path cached; freshness belongs to the caller that needs it (#3975)

2026-09-01 10:31:30 +02:00

test_bank_last_write_at.py

fix(api): paginate the bank list (#3586)

2026-08-18 17:27:27 +02:00

test_bank_lifecycle_deadlock_retry.py

fix(repair): stop the shared-DB deadlock flake in test-api (CONCURRENTLY test DDL + retry transient deadlocks) (#2943)

2026-07-24 14:04:13 +02:00

test_bank_scoped_read_404.py

fix(api): 404 bank-scoped reads for a bank that does not exist (#4175) (#4186)

2026-09-07 13:32:10 +02:00

test_bank_stats_cache_distributed.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_bank_stats_cache_invalidation.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_bank_stats_cache.py

fix(cache): preserve force refresh across event loops (#4072)

2026-09-07 11:06:06 +02:00

test_bank_stats.py

fix(api): 404 bank-scoped reads for a bank that does not exist (#4175) (#4186)

2026-09-07 13:32:10 +02:00

test_bank_template_configurable_fields.py

feat(mental-models): a minimum interval between automatic refreshes (#3480) (#3621)

2026-08-19 16:24:38 +02:00

test_bank_template_full_roundtrip.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_bank_templates.py

fix(api): expose the bank template import request body (#4247)

2026-09-09 11:11:50 +02:00

test_bank_utils.py

fix(db): make min_rows=0 mean "off", and stop counting rows on every write (#3485) (#3638)

2026-08-19 23:51:37 +02:00

test_bank_vector_index_ddl_lock.py

fix(db): earn per-bank vector indexes by size instead of creating three per bank (#3485) (#3561)

2026-08-18 18:03:00 +02:00

test_banner.py

chore: update the tagline to "Agent Memory That Learns" (#4093)

2026-09-07 10:20:57 +02:00

test_base_path.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_batch_api_integration.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_batch_api_validation.py

fix(retain): stop deriving the extraction narrator from the bank's display name (#3962) (#3978)

2026-09-01 11:49:40 +02:00

test_batch_api.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_batch_chunking.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_bedrock_service_tier.py

Support service_tier selection for Amazon Bedrock (#2098)

2026-06-10 10:41:33 +02:00

test_bm25_min_score_pushdown.py

fix(recall): honour min_scores.keyword on every text-search backend (#3882) (#3938)

2026-08-31 17:54:20 +02:00

test_bm25_term_selection.py

fix(recall): keep the most selective terms for native BM25 long queries (#3498)

2026-08-14 16:39:09 +02:00

test_cache_affinity.py

fix(llm): identify OpenAI-compatible requests (#4078)

2026-09-07 11:05:20 +02:00

test_causal_link_taxonomy.py

fix(curation): keep causal links across edit and invalidate/restore (#2951)

2026-07-24 16:38:06 +02:00

test_causal_relation_offsets.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_causal_relations.py

fix(engine): clarify causal link compatibility (#2685)

2026-07-13 10:07:26 -04:00

test_causal_relationships.py

fix(engine): clarify causal link compatibility (#2685)

2026-07-13 10:07:26 -04:00

test_chunk_id_legacy_documents.py

fix(retain): stop chunk ids colliding across banks (#4257)

2026-09-09 14:54:55 +02:00

test_chunk_ids.py

fix(retain): stop chunk ids colliding across banks (#4257)

2026-09-09 14:54:55 +02:00

test_chunk_storage_delete_ordering.py

fix(retain): stop chunk ids colliding across banks (#4257)

2026-09-09 14:54:55 +02:00

test_chunk_storage_upsert.py

fix(retain): stop chunk ids colliding across banks (#4257)

2026-09-09 14:54:55 +02:00

test_chunking_matches_legacy.py

refactor(retain): format pgvector literals without orjson, and drop orjson entirely (#4040)

2026-09-02 17:58:22 +02:00

test_chunking_reference_matches_langchain.py

refactor(retain): format pgvector literals without orjson, and drop orjson entirely (#4040)

2026-09-02 17:58:22 +02:00

test_chunking_streams.py

refactor(retain): format pgvector literals without orjson, and drop orjson entirely (#4040)

2026-09-02 17:58:22 +02:00

test_chunking.py

docs(recall): align min_scores score field names (#2432)

2026-06-29 10:41:59 +02:00

test_claim_bank_fairness.py

fix(worker): rotate slots across banks so bulk ingest stops starving them (#3861) (#3980)

2026-09-01 12:08:09 +02:00

test_claim_bank_serialization.py

fix(worker): rotate slots across banks so bulk ingest stops starving them (#3861) (#3980)

2026-09-01 12:08:09 +02:00

test_claude_code_llm_error_result.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_claude_code_llm_isolation.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_claude_code_llm_tool_round.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_coarse_date_spans.py

fix(recall): score coarse dates from their period, not its first day (#3933)

2026-08-31 16:54:05 +02:00

test_codex_base_url.py

fix(api): ignore inherited v1 base URL for Codex (#1718)

2026-05-26 10:55:23 +02:00

test_codex_extra_body.py

fix(llm): give every provider a real per-request deadline (#3898) (#3946)

2026-09-01 09:15:06 +02:00

test_codex_home_env.py

feat(llm): per-provider Codex credentials directory (#3793) (#3983)

2026-09-01 12:43:58 +02:00

test_codex_oauth_refresh.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_codex_quota_reset_defer.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_codex_reasoning_effort.py

fix(llm): give every provider a real per-request deadline (#3898) (#3946)

2026-09-01 09:15:06 +02:00

test_codex_request_headers.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_codex_stream_deadline.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_codex_strict_schema.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_codex_tool_choice.py

fix(llm): give every provider a real per-request deadline (#3898) (#3946)

2026-09-01 09:15:06 +02:00

test_cohere_cross_encoder.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_cohere_embeddings_retry.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_combined_scoring.py

fix(search): clamp future dates before the exponential recency power (#4105)

2026-09-07 15:32:36 +02:00

test_config_is_the_only_env_reader.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_config_resolution_shape.py

perf(config): stop deep-copying the global config on every resolution (#4209) (#4211)

2026-09-08 11:36:52 +02:00

test_config_validation.py

feat(pg_search): allow the extension function schema to be configured (#3759)

2026-08-31 11:18:39 +02:00

test_consolidation_batch_atomicity.py

fix(consolidation): apply one LLM response's writes in one transaction (#3923)

2026-08-31 14:47:09 +02:00

test_consolidation_batch_failure_visibility.py

fix(consolidation): teach the prompt to name delete targets, and count discarded batch responses (#4151, #4152) (#4183)

2026-09-07 12:46:19 +02:00

test_consolidation_dedup.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_consolidation_delete_prompt_llm.py

fix(consolidation): teach the prompt to name delete targets, and count discarded batch responses (#4151, #4152) (#4183)

2026-09-07 12:46:19 +02:00

test_consolidation_delete_schema.py

fix(consolidation): teach the prompt to name delete targets, and count discarded batch responses (#4151, #4152) (#4183)

2026-09-07 12:46:19 +02:00

test_consolidation_embedding_validation.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_consolidation_failure_isolation.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_consolidation_failure_recovery.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_consolidation_multi_round_refresh_3411.py

feat(mental-models): a minimum interval between automatic refreshes (#3480) (#3621)

2026-08-19 16:24:38 +02:00

test_consolidation_output_language.py

fix(llm): forward consolidation temperature (#3511)

2026-08-19 17:49:56 +02:00

test_consolidation_reschedule_after_round.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_consolidation_retry_budget.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_consolidation_retry_dedup_by_bank.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_consolidation_round_limit.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_consolidation_scope_parallelism.py

fix(consolidation): resolve observation_scopes before tag-set batching (#3954)

2026-09-01 11:33:05 +02:00

test_consolidation_silent_requeue_failure.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_consolidation_submit_atomic_dedup.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_consolidation_temporal_merge.py

fix(consolidation): apply one LLM response's writes in one transaction (#3923)

2026-08-31 14:47:09 +02:00

test_consolidation_write_scopes.py

fix(consolidation): resolve observation_scopes before tag-set batching (#3954)

2026-09-01 11:33:05 +02:00

test_consolidation.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_container_detection.py

Warn for unstable standalone worker ids (#2383)

2026-06-24 11:20:35 +02:00

test_content_policy_refusal_permanent.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_count_unconsolidated.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_cross_loop_primitives.py

fix(concurrency): serve CrossLoopSemaphore waiters in arrival order (#4157)

2026-09-07 12:37:40 +02:00

test_curation_http.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_custom_embedding_dimension.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_daemon_host_resolution.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_daemon_idle_timeout_removed.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_daemonize.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_db_abstraction.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_db_parallel_workers_config.py

feat(config): optional cap on planner parallelism for pool connections (#2600)

2026-07-20 13:40:44 +02:00

test_db_pool_reset.py

perf(db): skip asyncpg's release-time reset (#4259)

2026-09-10 17:36:29 +02:00

test_db_session_setup_flag.py

perf(db): skip asyncpg's release-time reset (#4259)

2026-09-10 17:36:29 +02:00

test_db_url.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_db_utils.py

fix(db): log the exception class when a DB error has no message (#3762)

2026-08-24 14:35:10 +02:00

test_deepseek_tool_call_compat.py

fix: propagate configured temperature through reflect calls (#3828)

2026-08-31 12:10:24 +02:00

test_delete_memory_units_validation.py

feat(engine): add MemoryEngine.delete_memory_units bulk primitive (#2659)

2026-07-24 14:38:04 +02:00

test_delta_editorial_fusion.py

fix(ci): stop one stalled Gemini call from outliving the caller, and isolate the third claim-test file (#3982)

2026-09-01 13:59:39 +02:00

test_delta_operation_parse.py

fix(reflect): accept id-bearing blocks in delta add_section ops (#3901) (#3919)

2026-08-31 15:36:55 +02:00

test_delta_oversized_replacement.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_delta_prompt_schema_parity.py

fix(mental-models): store delta documents as verbatim markdown blocks (#3361, #3273) (#3622)

2026-08-20 11:35:32 +02:00

test_delta_removed_chunks.py

fix(retain): delete the removed chunks when a re-retain only removes content (#3676)

2026-08-20 16:38:12 +02:00

test_delta_retain_duplicates.py

Memory backend: store-owned delta retain, bank-list times, and generic comments (#3660)

2026-08-20 11:31:59 +02:00

test_delta_retain_orphan_observations.py

fix(retain): sweep observations when delta retain deletes chunks (#3384)

2026-08-11 13:26:41 +02:00

test_delta_retain.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_disposition_config.py

fix(config): validate disposition_* range on bank-config write (#2348) (#2349)

2026-06-23 10:50:40 +02:00

test_document_body_accumulation.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_document_body_hash_once.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_document_chunks_and_reprocess.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_document_tracking.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_document_transfer.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_dotenv_loading.py

fix(server): load .env before importing MemoryEngine to honor concurrency limits (#3939)

2026-09-07 11:22:23 +02:00

test_dotenv_wins_over_early_config_build.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_dsn_application_name.py

fix(db): re-assert the DSN's application_name on every pool acquire (#3491)

2026-08-14 15:35:09 +02:00

test_embedding_coalescer.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_embedding_retry.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_embeddings_asymmetric_prefixes.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_embeddings_max_input_tokens.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_embeddings_openai_batch_size.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_embeddings_request_concurrency.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_enable_text_search_flag.py

fix(knowledge): switch native knowledge BM25 from websearch AND to disjunctive OR (#3945)

2026-09-01 11:01:21 +02:00

test_enqueue_graph_maintenance_ordered.py

fix(graph-maintenance): sort unit_ids in enqueue to eliminate insert deadlock (#2353)

2026-06-23 11:26:27 +02:00

test_ensure_bank_exists_optimization.py

refactor(bank): eliminate redundant existence queries in _ensure_bank_exists (#3672)

2026-08-31 12:35:26 +02:00

test_ensure_vector_no_global_index.py

test(vector-index): resolve the repair migration's parent instead of branch@-1 (#3238)

2026-08-07 12:42:07 +02:00

test_entity_intrabatch_clustering.py

perf(entity-resolver): optimize in-batch dedup via prefix filtering (#3991)

2026-09-08 09:51:16 +02:00

test_entity_intrabatch_dedup.py

fix(entities): dedup same-batch entity variants via pg_trgm (#3107) (#3197)

2026-08-06 18:02:51 +02:00

test_entity_labels_openapi_schema.py

fix(api): publish LabelGroup schema for entity_labels in OpenAPI (#3107) (#3190)

2026-08-05 17:54:32 +02:00

test_entity_labels.py

feat(entity-labels): add an open-vocabulary multi-valued label type (multi-text) (#4027)

2026-09-02 12:51:29 +02:00

test_entity_mention_count.py

fix(entities): give mention_count back when the mentions go (#4291) (#4296)

2026-09-10 12:35:58 +02:00

test_entity_merge_floor.py

fix(entities): don't merge a name onto an existing entity on history alone, + a resolution eval (#3752)

2026-08-24 13:45:01 +02:00

test_entity_name_hygiene.py

fix(memory): add resolve_entities flag to update_memory and retain (#3576)

2026-08-18 16:29:33 +02:00

test_entity_resolution_eval.py

fix(entities): don't merge a name onto an existing entity on history alone, + a resolution eval (#3752)

2026-08-24 13:45:01 +02:00

test_entity_resolver_candidate_cap.py

fix(entities): don't merge a name onto an existing entity on history alone, + a resolution eval (#3752)

2026-08-24 13:45:01 +02:00

test_entity_resolver_pg_trgm.py

perf(entity-resolution): exclude label entities from the trigram fuzzy-match index (#3208) (#3214)

2026-08-06 19:41:58 +02:00

test_entity_resolver.py

fix(memory): add resolve_entities flag to update_memory and retain (#3576)

2026-08-18 16:29:33 +02:00

test_entity_trgm_threshold_config.py

feat(entity-resolution): make pg_trgm similarity threshold configurable, set at connection setup (#3188)

2026-08-05 11:44:58 +02:00

test_experience_fact_type_passthrough.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_extension_bank_tables.py

feat(extensions): declare + provision extension-owned bank-scoped tables (#2903)

2026-07-24 13:32:59 +02:00

test_extension_passthrough_headers.py

feat(api): forward allowlisted request headers to extensions (#3428)

2026-08-13 14:07:48 +02:00

test_extension_unbundling.py

feat(extensions): add hindsight-extensions registry and unbundle Supabase (#3988)

2026-09-01 14:47:37 +02:00

test_extensions.py

feat(extensions): add hindsight-extensions registry and unbundle Supabase (#3988)

2026-09-01 14:47:37 +02:00

test_externally_owned_routines.py

feat(alembic): let a deployment own a maintenance routine (#3842)

2026-08-27 22:54:04 +02:00

test_extract_dry_run_http.py

feat(api,control-plane): a prompt tester for retain, and prompt preview for every operation (#4140)

2026-09-07 17:27:51 +02:00

test_fact_budget_selection.py

fix(recall): skip over-budget facts instead of stopping, and never answer a match with nothing (#3688) (#3704)

2026-08-24 13:09:04 +02:00

test_fact_extraction_agent_experience.py

Split test suite into deterministic mock and real LLM buckets (#1469)

2026-05-27 10:08:52 +02:00

test_fact_extraction_analysis.py

test: fix CI regressions (Vertex/litellmrouter construction, dedup config, trace recorder leak) (#2491)

2026-07-01 13:35:22 +02:00

test_fact_extraction_entities_llm.py

fix(retain): make entities a plain list of strings (#2749) (#2830)

2026-07-20 18:07:12 +02:00

test_fact_extraction_entities_schema.py

fix(retain): make entities a plain list of strings (#2749) (#2830)

2026-07-20 18:07:12 +02:00

test_fact_extraction_fact_type_prompt.py

fix(retain): clarify fact type boundary for user rules (#2440)

2026-06-30 10:08:55 +02:00

test_fact_extraction_metadata.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_fact_extraction_occurred_schema.py

fix(retain): constrain occurred_start/occurred_end behind a backend capability flag (#3845)

2026-08-31 12:33:13 +02:00

test_fact_extraction_output_ratio.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_fact_extraction_quality.py

fix(ci): stop one stalled Gemini call from outliving the caller, and isolate the third claim-test file (#3982)

2026-09-01 13:59:39 +02:00

test_fact_extraction_retry.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_fact_ordering.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_file_retain.py

fix: two test regressions from #4022 and #4031 (tokenizer cache leak, markitdown OCR fail-fast) (#4036)

2026-09-02 16:08:23 +02:00

test_file_storage_s3.py

fix(tests): wait for testcontainers port mappings (#2283)

2026-06-18 10:35:37 +02:00

test_fireworks_batch_integration.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_fireworks_batch.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_flush_stats_outside_connection.py

fix(oracle): unblock Oracle CI — free runner disk space + fix the retain deadlock (#2948)

2026-07-24 17:31:38 +02:00

test_format_facts_for_prompt.py

feat: include occurred_end and mentioned_at in think-prompt fact serialization (#929)

2026-04-08 10:33:14 +02:00

test_fusion_cap.py

fix(recall): gate VectorChord BM25 + add per-source candidate cap (#1707) (#1947)

2026-06-03 15:05:54 +02:00

test_gemini_400_fail_fast.py

fix(gemini): fail fast on deterministic 400 INVALID_ARGUMENT (#3256) (#3347)

2026-08-10 16:58:48 +02:00

test_gemini_batch_integration.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_gemini_batch.py

fix(gemini): grammar-enforce batch structured output regardless of strict flag (#2719)

2026-07-15 10:33:39 -04:00

test_gemini_cache.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_gemini_deadline_retry.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_gemini_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_gemini_implicit_cache_ratio.py

feat(reflect): step-by-step context caching for the Gemini tool loop (#2540)

2026-07-20 10:45:00 +02:00

test_gemini_safety_settings.py

fix(reflect): emit canonical OpenAI tool result messages (#2844)

2026-07-21 14:42:17 +02:00

test_gemini_service_tier.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_get_entity_uuid_validation.py

fix(api): validate UUID on get_entity and get_observation_history (#3260)

2026-08-10 14:34:07 +02:00

test_github_copilot_llm.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_github_copilot_registry_parity.py

feat(llm): add GitHub Copilot subscription provider (#3597)

2026-08-19 15:48:00 +02:00

test_google_cross_encoder.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_graph_entity_fanout_cap.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_graph_filtering.py

fix(api): paginate GET /observations/scopes and GET /webhooks (#4075)

2026-09-04 10:34:54 +02:00

test_graph_maintenance_deadlock.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_graph_maintenance_queue_race.py

fix(graph): serialize graph-maintenance queue enqueue against worker drain (#3034) (#3045)

2026-07-29 14:56:35 +02:00

test_graph_maintenance_submit_dedup.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_graph_maintenance.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_health_probes.py

fix(api): add a DB-free liveness probe so a slow database stops restarting pods (#3337)

2026-08-10 14:49:11 +02:00

test_hermes_templates_import.py

feat(templates): Hermes-branded bank templates, grounded in real Hermes user stories (#2996)

2026-08-06 10:50:26 -04:00

test_hierarchical_config.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_hnsw_indexes.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_horse_observations.py

feat(history): move mental-model & observation history into dedicated tables (#2007)

2026-06-05 15:07:30 +02:00

test_http_api_integration.py

fix(reflect): say when the structured-output extraction failed (#4230) (#4248)

2026-09-09 12:59:00 +02:00

test_http_internal_error.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_http_observability_middleware.py

perf(api): replace both BaseHTTPMiddleware with pure ASGI (3.2x on cheap routes) (#4235)

2026-09-09 10:06:44 +02:00

test_http_probe.py

fix(docker): drop curl from the API runtime images (#4202)

2026-09-08 11:11:04 +02:00

test_input_robustness.py

fix(retain): scrub the whole retain item at the engine ingress (#3908) (#3921)

2026-09-07 11:04:44 +02:00

test_integrity_violation_not_retried.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_interleave_fusion.py

fix(consolidation): eliminate duplicate observations via interleave dedup recall (#1907)

2026-06-03 17:37:57 +02:00

test_iris_parser_stub.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_iris_parser.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_jina_mlx_import_error.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_knowledge_base.py

fix(knowledge): switch native knowledge BM25 from websearch AND to disjunctive OR (#3945)

2026-09-01 11:01:21 +02:00

test_knowledge_bm25_dispatch.py

fix(knowledge): switch native knowledge BM25 from websearch AND to disjunctive OR (#3945)

2026-09-01 11:01:21 +02:00

test_knowledge_search_text_search_disabled.py

fix(api): 404 bank-scoped reads for a bank that does not exist (#4175) (#4186)

2026-09-07 13:32:10 +02:00

test_large_document_replacement.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_link_expansion_config.py

fix(recall): apply the created_after/created_before window to graph expansion (#3153)

2026-08-03 15:31:19 +02:00

test_link_expansion_projection.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_link_expansion_retrieval.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_link_expansion_scoring.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_link_utils.py

fix(recall): make the recall budget reach the vector index (#3541)

2026-08-19 10:46:04 +02:00

test_list_banks_config_overlay.py

fix(api): paginate the bank list (#3586)

2026-08-18 17:27:27 +02:00

test_list_banks_non_sql_store.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_list_banks_pagination.py

fix(api): paginate the bank list (#3586)

2026-08-18 17:27:27 +02:00

test_list_documents.py

fix(api): report the real last-write time on banks and documents (#3109)

2026-07-31 18:11:24 +02:00

test_list_endpoint_pagination_validation.py

fix(http): reject negative limit/offset on list endpoints with 422 instead of raw Postgres 500 (#2357)

2026-06-23 11:31:27 +02:00

test_list_memory_units_filters.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_list_memory_units_read_model.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_list_pagination.py

fix: report total on the mental-model and directive list endpoints (#3548)

2026-08-17 16:02:09 +02:00

test_list_row_models.py

fix(api): type the list and graph rows instead of returning bare dicts (#4218) (#4233)

2026-09-08 18:48:52 +02:00

test_litellm_embeddings_retry.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_litellm_forced_tool_structured_output.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_litellm_json_repair.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_litellm_sdk_cross_encoder.py

feat(api): attribute remote reranker calls by bank (#2740)

2026-07-20 14:32:08 +02:00

test_litellm_sdk_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_litellm_timeout.py

fix(litellm): hard-cap completions with asyncio.wait_for so a hung call can't block forever (#2224)

2026-06-17 12:08:49 +02:00

test_litellm_tool_choice.py

test(litellm): cover Responses named tool choice (#2953) (#2957)

2026-07-27 12:08:47 +02:00

test_llama_parse_parser.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_llamacpp_attempt_context.py

fix(llm): release concurrency permits during retry backoff (#3145)

2026-08-05 16:07:07 +02:00

test_llamacpp_cache_type.py

security: pin the llama.cpp prompt cache to RAM (#4102)

2026-09-03 18:25:30 -04:00

test_llamacpp_server_lifecycle.py

fix(llamacpp): report a missing llama.cpp instead of endless connection errors (#3758)

2026-08-24 13:00:50 +02:00

test_llm_4xx_dump.py

fix(gemini): fail fast on deterministic 400 INVALID_ARGUMENT (#3256) (#3347)

2026-08-10 16:58:48 +02:00

test_llm_extra_body.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_judge_prompt.py

fix(tests): stop the LLM judge mistaking context for the response (#3546)

2026-08-17 13:27:11 +02:00

test_llm_max_items.py

fix(consolidation): stop emitting unsupported maxItems that breaks all Bedrock consolidation (#2500) (#2502)

2026-07-27 12:10:19 +02:00

test_llm_per_op_concurrency.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_llm_provider.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_reasoning_effort_env.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_llm_router_provider.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_stage_breadcrumbs.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_strict_schema.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_temperature_env.py

fix(llm): make per-operation temperature configurable (#2459) (#2469)

2026-06-30 16:25:14 +02:00

test_llm_temperature_pipeline.py

fix(llm): make per-operation temperature configurable (#2459) (#2469)

2026-06-30 16:25:14 +02:00

test_llm_timeout_propagation.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_token_metrics.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_tools.py

fix(retain, reflect): drop the source-language rule when an output language is set (#3777)

2026-08-31 11:32:30 +02:00

test_llm_trace.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_llm_transport_diagnostics.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_llm_wrapper.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_lmstudio_tool_choice.py

refactor(llm): type provider tool choices (#2843)

2026-07-21 16:26:31 +02:00

test_load_large_batch.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_local_cross_encoder.py

fix(reranker): batch FlashRank passages instead of one unbounded forward pass (#3355) (#3441)

2026-08-13 09:34:18 +02:00

test_local_device.py

perf(local): skip full GC for CPU reranker providers (#3858)

2026-08-31 11:10:50 +02:00

test_local_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_local_model_alignment.py

fix(local-models): copy misaligned weights out of mapped storage, and fail fast (#4020)

2026-09-02 10:53:16 +02:00

test_local_provider_base_url.py

fix(llm): normalize bare LM Studio / Ollama base URL to /v1 (#2941)

2026-07-24 11:55:01 +02:00

test_lock_document_for_write.py

fix(retain): make document lock/upsert dialect-aware for Oracle (#1944) (#1952)

2026-06-04 10:37:22 +02:00

test_loop_lag.py

feat(metrics): /metrics covers every worker (labelled api_worker), and event-loop lag as a histogram (#4319)

2026-09-14 09:52:57 +02:00

test_loop_watchdog.py

fix(loop-watchdog): don't report the stall that stop() itself causes (#4021)

2026-09-02 11:07:43 +02:00

test_main_module.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_maintenance_loop.py

perf(maintenance): make the cross-tenant sweeps proportionate to tenant count (#3552)

2026-08-17 15:42:09 +02:00

test_maintenance_multitenant.py

fix(tests): eliminate test-api shard cross-test contamination (vchord cache, tenant schemas, maintenance routine TOCTOU) (#2310)

2026-06-19 15:32:03 +02:00

test_maintenance_routines.py

feat(alembic): let a deployment own a maintenance routine (#3842)

2026-08-27 22:54:04 +02:00

test_markitdown_parser.py

fix(parsers): coerce non-bytes buffers before the UTF-8 charset probe in markitdown (#2586)

2026-07-07 11:04:22 -04:00

test_mcp_endpoint_routing.py

feat(extensions): add hindsight-extensions registry and unbundle Supabase (#3988)

2026-09-01 14:47:37 +02:00

test_mcp_error_payloads.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_mcp_extension.py

fix(deps): address critical and high severity security vulnerabilities (#827)

2026-04-01 09:20:34 +02:00

test_mcp_routing.py

feat(mcp): expose knowledge-base CRUD as native MCP tools (#3486) (#3611)

2026-08-19 11:58:18 +02:00

test_mcp_server_version.py

fix(mcp): report Hindsight's version in serverInfo, not FastMCP's (#1281)

2026-04-27 16:41:39 +02:00

test_mcp_string_coercion.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_mcp_tool_filtering.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_mcp_tool_pair_parity.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_mcp_tools.py

feat(mental-models): default list to metadata; MCP list returns metadata only (#4225)

2026-09-08 18:47:54 +02:00

test_memories_extension.py

fix: defer retain completion outbox until store commit (#4203)

2026-09-08 17:44:58 +02:00

test_memory_curation.py

fix(curation): clear the legacy event_date when occurred_start is cleared (#4278)

2026-09-10 11:00:57 +02:00

test_memory_defense.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_memory_links_deferred_fk.py

fix(retain): defer memory_links → memory_units FKs to break cascade deadlock (#1398)

2026-05-04 12:25:11 +02:00

test_memory_links_to_unit_id_concurrent_delete.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_memory_units_updated_at.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_mental_model_consolidation_refresh_scope.py

feat(mental-models): a minimum interval between automatic refreshes (#3480) (#3621)

2026-08-19 16:24:38 +02:00

test_mental_model_delta.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_mental_model_document_budget.py

fix(mental-models): store delta documents as verbatim markdown blocks (#3361, #3273) (#3622)

2026-08-20 11:35:32 +02:00

test_mental_model_dry_run_refresh.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_mental_model_duplicate_id_409.py

fix(mental-models): return 409 on a duplicate mental-model id (#4024)

2026-09-02 14:10:08 +02:00

test_mental_model_hooks.py

feat: introduce hindsight-api-slim and hindsight-all-slim packages (#560)

2026-03-13 13:50:03 +01:00

test_mental_model_min_refresh_interval_3480.py

Memory backend: store-owned delta retain, bank-list times, and generic comments (#3660)

2026-08-20 11:31:59 +02:00

test_mental_model_no_answer_preserved.py

fix(reflect): fail the run when a retrieval tool fails, and record refused refreshes (#4003)

2026-09-01 17:07:15 +02:00

test_mental_model_refresh_pending_dedupe_3487.py

fix(mental-models): keep at most one queued refresh per model (#3487) (#3550)

2026-08-18 10:46:54 +02:00

test_mental_model_retractions.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_mental_model_scheduled_refresh.py

feat(mental-models): a minimum interval between automatic refreshes (#3480) (#3621)

2026-08-19 16:24:38 +02:00

test_mental_model_structured_output.py

fix(reflect): say when the structured-output extraction failed (#4230) (#4248)

2026-09-09 12:59:00 +02:00

test_mental_model_tool_failure_preserved.py

fix(reflect): fail the run when a retrieval tool fails, and record refused refreshes (#4003)

2026-09-01 17:07:15 +02:00

test_mental_model_trigger_flags.py

fix(mental-models): skip the reflect loop when the scope holds nothing to read (#3875) (#3943)

2026-09-01 11:51:57 +02:00

test_mental_model_trigger_patch_3687.py

fix(knowledge-pages): move a page's scope onto its mental model, and stop trigger patches resetting it (#3687) (#3755)

2026-08-24 14:06:12 +02:00

test_mental_model_trigger_tag_groups.py

feat(recall): fuzzy tag matching on tag_groups leaves (#4026) (#4028)

2026-09-02 16:50:53 +02:00

test_mental_models.py

feat(mental-models): default list to metadata; MCP list returns metadata only (#4225)

2026-09-08 18:47:54 +02:00

test_meta_provider.py

feat(llm): add Meta Model API as a first-class provider (#4081)

2026-09-04 12:08:55 +02:00

test_meta_tool_choice.py

feat(llm): add Meta Model API as a first-class provider (#4081)

2026-09-04 12:08:55 +02:00

test_metadata_llm_routing.py

feat(llm): route retain items to a chain member by their metadata (#4188)

2026-09-07 12:11:59 +02:00

test_metrics_collector_lifecycle.py

test: restore the global metrics collector instead of leaking it (#3780) (#3816)

2026-08-31 10:37:44 +02:00

test_metrics_multiworker.py

feat(metrics): /metrics covers every worker (labelled api_worker), and event-loop lag as a histogram (#4319)

2026-09-14 09:52:57 +02:00

test_metrics.py

perf(recall): three CPU cuts on the recall path (audit serialization, phase sampling, on-loop query embedding) (#4314)

2026-09-11 11:06:50 +02:00

test_migration_backsweep.py

fix(tests): isolate backsweep migration pg0 state (#2282)

2026-06-18 10:33:26 +02:00

test_migration_drop_access_count.py

chore(api): drop the never-written memory_units.access_count column (#3157)

2026-08-03 18:07:45 +02:00

test_migration_drop_v1_structured_content.py

fix(mental-models): store delta documents as verbatim markdown blocks (#3361, #3273) (#3622)

2026-08-20 11:35:32 +02:00

test_migration_entity_kind.py

perf(entity-resolution): exclude label entities from the trigram fuzzy-match index (#3208) (#3214)

2026-08-06 19:41:58 +02:00

test_migration_entity_maintenance_queue.py

fix(graph-maintenance): drive the entity prune off a queue, not a bank-wide sweep (#3222) (#3409)

2026-08-12 16:14:19 +02:00

test_migration_history_long_bank_id.py

fix(api): widen history bank_id to TEXT on PostgreSQL (#2106) (#2110)

2026-06-10 18:11:42 +02:00

test_migration_isolation.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_migration_observation_search_vector_backfill.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_migration_remaining_bank_id_text.py

test: pin migration-remaining-bankid tests to one xdist worker (#2225)

2026-06-16 13:42:14 +02:00

test_migration_shape.py

fix(db): unblock PostgreSQL upgrade to v0.7.x (sqlalchemy<2.1 + autocommit_block migrations) (#1904)

2026-06-01 14:42:59 +02:00

test_migrations_newer_revision.py

fix(migrations): preserve percent-encoded database URLs (#2733)

2026-07-20 14:28:29 +02:00

test_migrations_parallel_schemas.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_migrations_thread_safety.py

feat: introduce hindsight-api-slim and hindsight-all-slim packages (#560)

2026-03-13 13:50:03 +01:00

test_minimax_thinking_body.py

fix(minimax): disable thinking by default (#2558)

2026-07-07 15:27:50 -04:00

test_model_init_timeout.py

feat(reranker): failover chain via indexed HINDSIGHT_API_RERANKER_<n>_* members (#3176)

2026-08-04 19:00:27 +02:00

test_model_load_default_dtype.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_multi_llm_batch.py

fix(retain): stop deriving the extraction narrator from the bank's display name (#3962) (#3978)

2026-09-01 11:49:40 +02:00

test_multi_llm_config.py

feat(llm): per-provider Codex credentials directory (#3793) (#3983)

2026-09-01 12:43:58 +02:00

test_multi_llm_provider.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_multilingual_bm25.py

fix(knowledge): switch native knowledge BM25 from websearch AND to disjunctive OR (#3945)

2026-09-01 11:01:21 +02:00

test_multilingual.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_multimodal_extraction_llm.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_multimodal_prompt_assembly.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_named_result_stubs.py

fix(retain): scrub the whole retain item at the engine ingress (#3908) (#3921)

2026-09-07 11:04:44 +02:00

test_narrator_context_override.py

fix(retain): stop bank_id routing key polluting fact attribution (#1680) (#1948)

2026-06-03 16:22:51 +02:00

test_narrator_resolution.py

fix(retain): stop deriving the extraction narrator from the bank's display name (#3962) (#3978)

2026-09-01 11:49:40 +02:00

test_none_llm_provider.py

Split test suite into deterministic mock and real LLM buckets (#1469)

2026-05-27 10:08:52 +02:00

test_nous_auth_refresh.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_nous_provider.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_observation_expansion_scoring.py

fix(recall): keep the observation graph arm out of a nested-loop plan (#3510) (#3588)

2026-08-18 17:38:24 +02:00

test_observation_expansion_single_fetch.py

Fuse observation graph expansion into one database fetch (#3859)

2026-09-01 09:15:17 +02:00

test_observation_history_fk_race.py

fix: handle FK violation in observation_history during parallel consolidation (#2620)

2026-07-10 11:12:04 -04:00

test_observation_history.py

Memory backend: store-owned delta retain, bank-list times, and generic comments (#3660)

2026-08-20 11:31:59 +02:00

test_observation_invalidation.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_observation_scope_limit_resolution.py

feat(consolidation): per-scope observation limits (#2140)

2026-06-11 18:27:43 +02:00

test_observations.py

Memory backend: store-owned delta retain, bank-list times, and generic comments (#3660)

2026-08-20 11:31:59 +02:00

test_ollama_native_think.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_ollama_native_timeout.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_ollama_num_ctx_transport.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_onnx_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_op_cancellation.py

fix(operations): allow cancelling in-flight operations (#4131)

2026-09-07 11:21:06 +02:00

test_openai_compatible_response_hardening.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_openai_compatible_truncation.py

fix(llm): raise OutputTooLongError when a non-streaming completion is truncated (#3827)

2026-08-31 13:03:33 +02:00

test_openai_max_tokens_param.py

fix(llm): send max_completion_tokens for reasoning models and Azure OpenAI (#979)

2026-04-13 08:56:16 +02:00

test_openai_responses_provider.py

fix(llm): identify OpenAI-compatible requests (#4078)

2026-09-07 11:05:20 +02:00

test_openapi_binary_responses.py

fix(api): declare binary download bodies so generated clients return bytes (#4315)

2026-09-11 11:26:52 +02:00

test_opencode_go_provider.py

feat: add opencode-go LLM provider (#1652)

2026-05-19 16:34:11 -04:00

test_opencode_go_session_header.py

fix(api-slim): read the opencode-go retry assertion off LLMCallResult.content (#4181)

2026-09-07 11:33:15 +02:00

test_openrouter_null_content.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_operation_completion.py

fix(engine): isolate operation completion from best-effort side-effects (#2823)

2026-07-20 14:25:45 +02:00

test_operation_progress.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_operation_retention_migration.py

revert(migrations): drop advisory-lock maintenance-routines install (#2690) (#2817)

2026-07-20 12:15:41 +02:00

test_operation_status.py

fix(operations): allow cancelling in-flight operations (#4131)

2026-09-07 11:21:06 +02:00

test_oracle_backend_integration.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_oracle_http_integration.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_oracle_integration.py

fix(consolidation): apply one LLM response's writes in one transaction (#3923)

2026-08-31 14:47:09 +02:00

test_oversized_append_truncates_document.py

fix(retain): give a split append the whole document as its body, not just the tail (#3989) (#4229)

2026-09-08 17:29:24 +02:00

test_packed_embeddings.py

refactor(retain): format pgvector literals without orjson, and drop orjson entirely (#4040)

2026-09-02 17:58:22 +02:00

test_page_markdown.py

feat(knowledge-base): client-managed knowledge pages, control-plane UI + hindsight fs CLI (#2455)

2026-07-30 16:02:41 +02:00

test_per_operation_llm_config.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_pg0_url.py

fix(pg0): honor URL credentials during MemoryEngine startup (#2836)

2026-07-21 11:50:36 +02:00

test_pg_extensions.py

fix(api): install every PostgreSQL extension in the public schema (#4118) (#4121)

2026-09-04 12:54:29 +02:00

test_pool_acquire_timeout.py

fix(worker): stop wedged retains from holding worker slots forever (#3020)

2026-07-28 14:40:42 +02:00

test_pool_instrumentation.py

feat(observability): diagnose blocked-loop vs pool-exhaustion on stalled /health (#2942)

2026-07-24 12:11:07 +02:00

test_profiling.py

perf(recall): account for a recall's time per phase, and fix the three things that showed up (#4299)

2026-09-10 17:48:15 +02:00

test_prompt_brace_escape.py

fix(retain): drop silent getattr config defaults so a wrong config fails loudly (#3610)

2026-08-19 15:28:53 +02:00

test_prompt_length_retries.py

fix(api): mental model delta refresh (prompt size, JSON, consolidation) (#2170)

2026-06-15 13:40:18 +02:00

test_prompt_preview_http.py

feat(api,control-plane): a prompt tester for retain, and prompt preview for every operation (#4140)

2026-09-07 17:27:51 +02:00

test_provider_capability_parity.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_provider_default_models.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_provider_quota_reset_defer.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_quality_integration.py

test(ci): de-flake TEI parallelism timing + disposition judge reruns (#2045)

2026-06-08 12:35:49 +02:00

test_query_analyzer.py

fix(temporal): "last weekend" asked on a weekend means the previous one (#4076)

2026-09-07 11:04:21 +02:00

test_read_backend.py

feat(engine): optional read-only backend for recall queries (#1460)

2026-05-06 14:51:25 +02:00

test_reasoning_effort_explicit_config.py

fix(llm): send an explicitly configured reasoning_effort (#3449) (#3459)

2026-08-13 17:15:14 +02:00

test_reasoning_effort_provider_coverage.py

fix(llm): send an explicitly configured reasoning_effort (#3449) (#3459)

2026-08-13 17:15:14 +02:00

test_recall_attachment_provenance.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_recall_boost.py

fix(recall): boost the prioritised arm in rank space, not score space (#3956) (#3987)

2026-09-01 14:09:42 +02:00

test_recall_budget_config.py

feat(recall): make budget mapping configurable per bank (#1106)

2026-04-16 18:35:36 +02:00

test_recall_cancellation.py

fix(api): make recall/reflect disconnect cancellation actually work (#2131)

2026-06-11 12:52:38 +02:00

test_recall_chunks_independence.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_recall_config.py

fix(mental-models): skip the reflect loop when the scope holds nothing to read (#3875) (#3943)

2026-09-01 11:51:57 +02:00

test_recall_error_propagation.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_recall_invalid_fact_types.py

fix: return 422 for invalid recall fact types (#3062)

2026-08-03 17:51:02 +02:00

test_recall_min_score.py

fix(recall): honour min_scores.keyword on every text-search backend (#3882) (#3938)

2026-08-31 17:54:20 +02:00

test_recall_observation_entities.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_recall_pg_enrichment_combo.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_recall_phase_instrumentation.py

metrics: time the recall phases, so the request's largest share stops being a residual (#3999)

2026-09-01 16:49:56 +02:00

test_recall_pipeline_toggles.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_recall_prefer_observations.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_recall_query_token_cap.py

perf(tokenizer): move token counting from quicktok to toktok-rs (#4022)

2026-09-02 12:52:08 +02:00

test_recall_scoring_time.py

feat(recall): configurable recency decay function (linear/exponential/none) (#2318)

2026-06-23 10:16:29 +02:00

test_recall_temporal_window.py

feat(recall): let callers supply the temporal window instead of parsing it (#3678)

2026-08-21 13:02:46 +02:00

test_recall_time_range_graph.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_recall_time_range.py

engine: serve the addressed document/chunk/entity reads from the store too (#3560)

2026-08-19 18:09:12 +02:00

test_recall_token_budget.py

fix(recall): skip over-budget facts instead of stopping, and never answer a match with nothing (#3688) (#3704)

2026-08-24 13:09:04 +02:00

test_recall_tracer_payload_gating.py

perf(recall): stop building trace payloads when no trace was asked for (#4231)

2026-09-08 18:36:23 +02:00

test_reflect_agent.py

fix(reflect): uniquify tool_call ids so strict APIs accept the turn (#4250)

2026-09-10 10:55:06 +02:00

test_reflect_anthropic_protocol.py

fix(anthropic): group consecutive tool results into one user message (#3910)

2026-08-31 15:59:53 +02:00

test_reflect_document_answer.py

fix(reflect): name the sections array the document schema requires (#3984)

2026-09-01 14:00:01 +02:00

test_reflect_empty_based_on.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_reflect_expand_store_reads.py

Memory backend: store-owned delta retain, bank-list times, and generic comments (#3660)

2026-08-20 11:31:59 +02:00

test_reflect_grounding_behaviour.py

fix(reflect): grounding defects in reflect and knowledge-page delta refresh, plus system-evals (#4304)

2026-09-11 11:41:34 +02:00

test_reflect_grounding_boundary.py

fix(reflect): grounding defects in reflect and knowledge-page delta refresh, plus system-evals (#4304)

2026-09-11 11:41:34 +02:00

test_reflect_internal_billing.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_reflect_page_length_decoupling.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_reflect_prompt_builder.py

fix(reflect): grounding defects in reflect and knowledge-page delta refresh, plus system-evals (#4304)

2026-09-11 11:41:34 +02:00

test_reflect_recall_entities.py

feat(reflect): resolve entity names on reflect sub-recalls (#3334)

2026-08-10 13:59:02 +02:00

test_reflect_source_facts_config.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_reflect_split_synthesis.py

fix(reflect): run split synthesis's map calls at temperature 0 (#4054) (#4115)

2026-09-04 15:17:38 +02:00

test_reflect_tokenizer_lazy_load.py

fix: two test regressions from #4022 and #4031 (tokenizer cache leak, markitdown OCR fail-fast) (#4036)

2026-09-02 16:08:23 +02:00

test_reflect_tool_result_fields.py

perf(reflect): drop retrieval plumbing from reflect tool results (#3310)

2026-08-10 13:41:47 +02:00

test_reflect_tools.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_reflect_tracing.py

fix(tests): unbreak main — post-#3649 batch stubs and a leaked tracing global (#3668)

2026-08-20 14:33:01 +02:00

test_reflections.py

fix(search): embed the whole mental-model document on every write (#3926) (#3928)

2026-08-31 15:50:44 +02:00

test_refresh_outcome_metadata.py

fix(reflect): say when the structured-output extraction failed (#4230) (#4248)

2026-09-09 12:59:00 +02:00

test_repair_bank_vector_indexes.py

fix(db): make min_rows=0 mean "off", and stop counting rows on every write (#3485) (#3638)

2026-08-19 23:51:37 +02:00

test_reprocess_force_reextract.py

fix(retain): make reprocess re-extract instead of no-opping (#3899) (#3949)

2026-09-01 09:40:11 +02:00

test_reranker_error_handling.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_reranker_failover_chain.py

feat(reranker): failover chain via indexed HINDSIGHT_API_RERANKER_<n>_* members (#3176)

2026-08-04 19:00:27 +02:00

test_reranker_max_candidates_budget.py

feat(recall): per-budget reranker candidate cap via env config (#3107) (#3191)

2026-08-05 16:50:58 +02:00

test_reranker_retry.py

fix(reranker): give every remote reranker bounded retry (#4139)

2026-09-07 10:06:50 +02:00

test_reranker_score_normalization.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_reranker_timeouts.py

feat(reranker): failover chain via indexed HINDSIGHT_API_RERANKER_<n>_* members (#3176)

2026-08-04 19:00:27 +02:00

test_reranking_proof_count.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_response_exclude_none.py

fix(api): type the list and graph rows instead of returning bare dicts (#4218) (#4233)

2026-09-08 18:48:52 +02:00

test_response_models.py

fix(engine): drop null metadata values at retain and recall (#3209) (#3531)

2026-08-18 15:19:11 +02:00

test_response_schema_validation.py

feat(reflect,mental-models): surface structured output in the control plane (#3113)

2026-08-03 16:51:38 +02:00

test_retain_append_mode.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_retain_attachment_ingress.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_retain_chunking_config_access.py

fix(retain): drop silent getattr config defaults so a wrong config fails loudly (#3610)

2026-08-19 15:28:53 +02:00

test_retain_chunks_embedding_batching.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_retain_document_serialization.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_retain_entity_prune_race.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_retain_label_tag_projection.py

fix(retain): keep per-memory entity-label tags when a re-retain rewrites document tags (#4068) (#4120)

2026-09-04 13:06:14 +02:00

test_retain_memory_budget.py

fix(retain): bound retain's memory by a budget instead of by the document (#3756) (#3763)

2026-08-24 15:03:07 +02:00

test_retain_metadata_normalization.py

fix(engine): drop null metadata values at retain and recall (#3209) (#3531)

2026-08-18 15:19:11 +02:00

test_retain_multimodal_extraction.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_retain_orchestrator_mapping.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_retain_outbox_session.py

fix: defer retain completion outbox until store commit (#4203)

2026-09-08 17:44:58 +02:00

test_retain_params_roundtrip.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_retain_pipeline_cancellation.py

test: make two deselection markers actually apply (#4126)

2026-09-04 17:01:49 +02:00

test_retain_reflect_output_language.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_retain_result_shape.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_retain_same_document_concurrency.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_retain_store_owned_no_connection.py

fix(retain): a store-owned retain must not fail on its own log line (#4236)

2026-09-08 18:58:48 +02:00

test_retain_tag_change_observations.py

fix(retain): invalidate observations when a re-retain changes a document's scoping (#4019)

2026-09-02 10:20:29 +02:00

test_retain_transient_extraction_failure.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_retain_validator_attachments.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_retain_write_gate_selection.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_retain.py

test: make two deselection markers actually apply (#4126)

2026-09-04 17:01:49 +02:00

test_schema_isolation.py

fix(api): 404 bank-scoped reads for a bank that does not exist (#4175) (#4186)

2026-09-07 13:32:10 +02:00

test_schema_mode_extensions.py

fix(api): install every PostgreSQL extension in the public schema (#4118) (#4121)

2026-09-04 12:54:29 +02:00

test_search_trace.py

fix(recall): report the caller's query_timestamp in the search trace (#4227)

2026-09-08 17:25:46 +02:00

test_server_module.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_source_facts_selection.py

fix(recall): fill source_facts budget in rank order and flag truncation (#3221) (#3419)

2026-08-12 10:08:02 +02:00

test_source_facts_tokens.py

perf(tokenizer): move token counting from quicktok to toktok-rs (#4022)

2026-09-02 12:52:08 +02:00

test_sql_schema_safety.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_stage.py

feat(worker): diagnostic logging for stuck/slow async tasks (#1017)

2026-04-13 11:37:52 +02:00

test_store_backpressure_defers.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_store_document_text.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_strict_json_schema.py

fix(structured-output): strip $ref sibling keywords from strict schemas (#3944)

2026-09-01 11:48:52 +02:00

test_strict_schema_entity_labels.py

fix(structured-output): strip $ref sibling keywords from strict schemas (#3944)

2026-09-01 11:48:52 +02:00

test_strip_code_fences.py

feat(llm): recover outer JSON span when fence stripping yields non-JSON (#2610)

2026-07-07 15:44:02 -04:00

test_strip_reasoning_tags.py

fix(llm): preserve inline <think> literals in _strip_reasoning_tags (#3426)

2026-08-12 17:17:40 +02:00

test_structured_chunk_size_retain.py

fix(retain): make chunk_text idempotent so raised structured chunk size doesn't fail retains (#2301) (#2308)

2026-06-19 11:21:33 +02:00

test_structured_delta_prompt_budget.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_structured_doc.py

fix(mental-models): store delta documents as verbatim markdown blocks (#3361, #3273) (#3622)

2026-08-20 11:35:32 +02:00

test_structured_output_unions.py

feat(reflect): send the operation schema on mental model delta refresh (#3937)

2026-09-01 09:38:18 +02:00

test_sub_batch_rejoin_fragmentation.py

memories: raise the store seam above persistence, and collapse its ownership flags

2026-08-31 17:16:39 +02:00

test_sub_batch_span_reconstruction.py

feat(retain): inline images and files as first-class content (#4077)

2026-09-04 12:48:08 +02:00

test_subbatch_chunk_coverage.py

fix(retain): offset chunk_index across sub-batches of an oversized document (#1888) (#1896)

2026-06-01 12:08:37 +02:00

test_subbatch_multichunk_coverage.py

test(retain): serialize multichunk sub-batch coverage test on worker_tests xdist group (#2272)

2026-06-18 12:06:12 +02:00

test_tag_resolution.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_tags_visibility.py

fix(api): 404 bank-scoped reads for a bank that does not exist (#4175) (#4186)

2026-09-07 13:32:10 +02:00

test_tei_client_threads.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_tei_cross_encoder.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_tei_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_tei_retry.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_temporal_extraction.py

perf(recall): make temporal extraction ~9x faster without changing behaviour (#3452)

2026-08-13 11:17:16 +02:00

test_temporal_language_detection_threading.py

revert: drop free-threaded CPython 3.14 support (#4037, #4067) (#4234)

2026-09-09 10:30:09 +02:00

test_temporal_ranges.py

test: assert memory state via the engine read API, not raw SQL (#3591)

2026-08-19 00:23:44 +02:00

test_temporal_recall_selection.py

refactor(api-slim): write each duplicated policy once (#4083)

2026-09-03 17:43:41 +02:00

test_text_search_reconciliation.py

fix(search): use PGroonga for Knowledge Pages (#3335)

2026-08-10 16:37:27 +02:00

test_think.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_thread_limits.py

fix(api): bound native ML thread pools to available CPUs (#1901)

2026-06-01 16:05:11 +02:00

test_token_counting_memory.py

perf(tokenizer): move token counting from quicktok to toktok-rs (#4022)

2026-09-02 12:52:08 +02:00

test_token_encoding.py

perf(tokenizer): move token counting from quicktok to toktok-rs (#4022)

2026-09-02 12:52:08 +02:00

test_token_usage_cached_thoughts.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_tool_choice_required_downgrade.py

refactor(llm): type provider tool choices (#2843)

2026-07-21 16:26:31 +02:00

test_tool_path_reasoning_effort.py

fix(llm): send reasoning_effort on the tool path, matching call() (#2983)

2026-07-28 14:31:29 +02:00

test_tracing_context_propagation.py

fix(tracing): name API traces after the operation and link queued retain to its worker (#4006)

2026-09-01 19:29:23 +02:00

test_tracing_integration.py

feat: introduce hindsight-api-slim and hindsight-all-slim packages (#560)

2026-03-13 13:50:03 +01:00

test_tracing_spans_verification.py

chore: format test files with ruff (enable formatter on tests/) (#2074)

2026-06-09 13:23:11 +02:00

test_tracing_task_handoff.py

fix(tracing): name API traces after the operation and link queued retain to its worker (#4006)

2026-09-01 19:29:23 +02:00

test_tracing.py

refactor(config): make HindsightConfig the only parser of HINDSIGHT_API_* env vars (#4260)

2026-09-09 18:40:13 +02:00

test_transfer_observation_source_liveness.py

fix(api-slim): never import an observation citing a missing source unit (#4179)

2026-09-07 12:00:08 +02:00

test_unknown_params.py

perf(api): replace both BaseHTTPMiddleware with pure ASGI (3.2x on cheap routes) (#4235)

2026-09-09 10:06:44 +02:00

test_validation_result_enrichment.py

test: add unit tests for pg_trgm auto-detection and ValidationResult.accept_with() enrichment (#650)

2026-03-23 10:33:09 +01:00

test_validator_phase_instrumentation.py

metrics: time the operation-validator hooks, which run outside the operation's timer (#3976)

2026-09-01 11:21:07 +02:00

test_vector_index.py

fix(api): install every PostgreSQL extension in the public schema (#4118) (#4121)

2026-09-04 12:54:29 +02:00

test_verify_connection_soft_fail.py

fix(startup): downgrade LLM verify_connection failure to warning instead of crash (#1166)

2026-04-20 10:25:47 +02:00

test_vertexai_provider.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_vlm_slot.py

refactor(api-slim): return a named result from LLMInterface.call (#4122)

2026-09-04 15:14:17 +02:00

test_webhook_url_guard.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_webhooks.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_within_batch_link_bounds.py

perf(retain): accelerate within-batch semantic link calculation (#3977)

2026-09-08 11:38:19 +02:00

test_worker_claim_connection_reuse.py

fix(worker): rotate slots across banks so bulk ingest stops starving them (#3861) (#3980)

2026-09-01 12:08:09 +02:00

test_worker_main.py

fix(observability): trace the worker process and join the caller's trace (#3625)

2026-08-19 15:42:16 +02:00

test_worker_proc_stats.py

fix(reranker): batch FlashRank passages instead of one unbounded forward pass (#3355) (#3441)

2026-08-13 09:34:18 +02:00

test_worker_retry_knobs.py

fix(worker): wire HINDSIGHT_API_WORKER_MAX_RETRIES into task retry decision

2026-05-29 13:57:43 +02:00

test_worker_stuck_watchdog.py

fix(worker): detect stuck tasks by stage progress (#3720)

2026-08-31 10:34:04 +02:00

test_worker_wall_timeout.py

fix(worker): bound stalled consolidation tasks with an idle wall-clock ceiling (#3746)

2026-08-31 11:22:00 +02:00

test_worker.py

fix(operations): allow cancelling in-flight operations (#4131)

2026-09-07 11:21:06 +02:00

test_xai_oauth_llm.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_zeroentropy_embeddings.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00

test_zeroentropy_live.py

refactor: replace httpx and sync HTTP with aiohttp across production code (#4318)

2026-09-11 11:36:28 +02:00