v2.21.0

What's Changed

• bench(#475): multi-seed confirmation — 73.8% mean over 3 runs, worst seed +9.0 over Zep by @tcconnally in #576
• refactor(integrations): route langgraph/crewai/autogen adapters through the shared client by @tcconnally in #571
• bench(longmemeval): CPST results — Perseus 4.7× cheaper per correct answer, +10 pts accuracy by @tcconnally in #577
• docs(#554): enumerate all retrieval modes in one reference by @tcconnally in #573
• docs(#556): note direct SQL edits need maintain to reconcile FTS by @tcconnally in #574
• docs: value-first README lead + LongMemEval benchmark section by @tcconnally in #578
• Revise description for Perseus Computing LLC by @tcconnally in #581
• feat(#563,#562): capture --consume/--prune-source + enumeration semantics by @tcconnally in #582
• feat(#551): PraisonAI MemoryProtocol adapter by @tcconnally in #583
• bench(#553): reproducible three-axis temporal benchmark by @tcconnally in #584
• bench(#579,#580): official-CoT answer prompt + offline retrieval coverage diagnostic by @tcconnally in #585
• feat(#562): perseus_vault_scan — deterministic keyset enumeration + query-contract docs by @tcconnally in #586
• bench(#579): official-CoT LongMemEval distribution — 79.0% mean over 3 signed runs by @tcconnally in #587
• diag(#590): version-inversion rate for knowledge-update (free, no-LLM gate) by @tcconnally in #593
• fix(#591): scale_bench.py passes --embedding-model-name (+ --skip-seed) by @tcconnally in #594
• fix(#592): batch embed aborts + surfaces top-level error on persistent backend failure by @tcconnally in #595
• fix(#591): scale_bench --embedding-model-name (501 loop) + verified 100k recall result by @tcconnally in #596
• docs: add recall-accuracy-at-scale benchmark to README (100K hybrid 1.000@5 vs keyword 0.015) by @tcconnally in #597
• bench: commit portable Lambda orchestration kit + 100K recall headline by @tcconnally in #598
• bench: measured Zep + Letta same-box recall comparison by @tcconnally in #599
• bench: refresh 8×H100 fleet throughput at 100K + recall-hold check by @tcconnally in #600
• docs(bench): record 2026-07-12 fleet re-validation + recall-hold note by @tcconnally in #604
• docs: propagate measured Zep + Letta same-box results into READMEs/comparisons by @tcconnally in #606
• docs: standardize Vault MCP tool count as '55+' by @tcconnally in #611
• bench/lambda: harness fixes — llm-model arg, mean-of-N Zep runs, competitor stack bring-up (#607, #608, #609) by @tcconnally in #612
• bench/lambda: terminate verify+retry, incremental recall checkpointing (#602, #603) by @tcconnally in #613
• docs: measured A100 rows (8×40GB + 2×80GB) in agent-economics tables, with GPUs column (#605) by @tcconnally in #615
• perf: bounded concurrent worker pool for batch embedding (#601) by @tcconnally in #616
• perf: bounded-scan pre-rank for the BM25 keyword arm (#589) [needs Lambda measurement before merge] by @tcconnally in #617
• bench: 1M-entity recall scale point (fleet-embedded) by @tcconnally in #618
• #590: event-time supersession-aware recall tiebreak (off by default) by @tcconnally in #621
• #620: serve_fleet.sh — detach daemons so ssh returns; make nginx LB optional by @tcconnally in #622
• #590: correct SUPERSEDE_590 — recall-time reranking is ineffective on current main by @tcconnally in #623
• #588: multi-query expansion (flat N-arm RRF) + date-aware arm (off by default) by @tcconnally in #624
• bench(scale): 1M budget row + #589 scan-cap measurement (keep cap default off) by @tcconnally in #625
• #590: ingest-time analysis — engine already does latest-wins for same-key updates (demo + doc) by @tcconnally in #626
• #619 interim: MIMIR_DENSE_MAX_SCAN — explicit exact-vs-bounded dial for the dense arm by @tcconnally in #627
• #590: --ku-shared-key harness mode + measured close-out (inversion 47.4%→0.0%, KU QA 73.1%→78.2%) by @tcconnally in #628
• #619 step 1: resident sig cache — full-corpus Hamming prefilter for the dense arm (opt-in) by @tcconnally in #631
• perf(recall): cache parsed body_json to avoid re-parse in to_json_expanded by @tcconnally in #643
• perf(recall): add Vec capacity hints for hot-path allocations by @tcconnally in #644
• feat: mimir_recall_batch — server-side multi-query recall with RRF fusion by @tcconnally in #648
• fix(ci): repair grpc build break + model-cache immunity for all build jobs (#649) by @tcconnally in #650
• docs: client-compatibility note for canonical-only tool manifest (#633) by @tcconnally in #645
• bench(1m): skip the no-op serial embed crawl on fully-embedded corpora by @tcconnally in #647
• ci(recall-gate): settle the async auto-embed worker before measuring (#632) by @tcconnally in #646
• fix(embed): explicit --embedding-endpoint outranks the bundled model for query embedding by @tcconnally in #651
• bench(1m): #619 sig-cache A/B artifacts — dense@5 +0.20, hybrid@1 +0.27 at 1M by @tcconnally in #652
• perf(619): partial-select + corpus-scaled rerank pool for the sig cache by @tcconnally in #653
• bench(1m): #619 step-1.5 artifacts — hybrid@1 0.936 at baseline latency by @tcconnally in #654
• perf(619): sig cache defaults to AUTO — engages when the corpus outgrows max_scan by @tcconnally in #655
• audit: reconcile version/tool-count/LongMemEval drift + stale names across metadata & docs by @tcconnally in #656
• docs(readme): use signed 3-run LongMemEval mean (73.8%) in headline by @tcconnally in #659
• perf(619): step 2a — asymmetric sig scoring + flat cache buffer by @tcconnally in #660
• bench(1m): #619 step-2a artifacts — asym Pareto win, hybrid@1 0.948, step 2b GO by @tcconnally in #661
• feat(619): step 2b — int4 ADC refine tier between 1-bit coarse and exact rerank by @tcconnally in #662
• bench(1m): #619 step-2b artifacts — int4 refine flat; the wall is coarse recall by @tcconnally in #663
• perf(619): step 2c′ — resident int4 coarse + 2b refine default OFF by @tcconnally in #664
• bench(1m): #619 step-2c′ artifacts — resident int4 coarse also flat at 0.722 by @tcconnally in #665
• bench(1m): #619 exact-scan ceiling — the shipped default beats it; campaign closed by @tcconnally in #666
• fix: back-compat write flag aliases (#658) + explicit ok on MCP remember (#657) by @tcconnally in #667
• perf(#630): embedding-quantization ladder + pure-1-bit harness flag by @tcconnally in #668
• research(#629): NVFP4 e...

v2.20.2

What's Changed

• feat(#552): Pydantic AI integration — Perseus Vault memory capability by @tcconnally in #567
• bench(longmemeval): CPST harness — stateless arm, API-billed usage, stratified subset + synthesis by @tcconnally in #572
• release: v2.20.2 — resilient attest + drop EOL macos-13 leg by @tcconnally in #575

Full Changelog: v2.20.1...v2.20.2

v2.20.1

What's Changed

• fix(#547): deterministic parent-death detection via PR_SET_PDEATHSIG + ppid poll by @tcconnally in #550
• security: internal vulnerability-response policy (OSTIF step 3) by @tcconnally in #558
• bench(#475): LongMemEval 73.6% — beats Zep (63.8%), official metric, signed + independently verified by @tcconnally in #557
• fix(#547): orphan-reap guard false-fires when vault is born under PID 1 (container entrypoint) by @tcconnally in #559
• ci: add CodeQL SAST for the Rust source by @tcconnally in #560
• docs: security knowledgebase index + escalation milestones (OSTIF steps 6-7) by @tcconnally in #564
• docs: confirm access-register roster + branch protection (follow-up to #564) by @tcconnally in #565
• feat(#561): official Python client for the MCP stdio transport by @tcconnally in #566
• docs: record Mark as repo admin in access register by @tcconnally in #568
• ci: add SLSA build provenance (signed attestations) to releases by @tcconnally in #569
• release: v2.20.1 (validate SLSA provenance) by @tcconnally in #570

Full Changelog: v2.20.0...v2.20.1

v2.20.0

What's Changed

• bench: scale harness — 10K/100K committed baselines + latency-budget CI gate by @tcconnally in #506
• chore: remove stale Mimir-era quickstart example by @tcconnally in #508
• perf: signature-driven near-duplicate scan — 100K bulk load 7/s → 39/s (~4h → 43min) by @tcconnally in #509
• perf: covering index for dense phase-0 — dense recall p50 360ms → 25ms at 100K (14.5×) by @tcconnally in #510
• site: benchmark-page generator — perseus.observer/benchmarks from signed reports by @tcconnally in #512
• bench(scale): post-#476/#507 signed report + tightened budgets by @tcconnally in #513
• site: benchmark page scale table gains the dense p50/p99 column by @tcconnally in #514
• Add funding.json (v1.1.0) + wellKnown provenance for FLOSS/fund by @tcconnally in #515
• bench(longmemeval): crash-safe progress journal + --resume by @tcconnally in #519
• docs(readme): bump version badge to 2.19.1, drop GitHub-stars comparison row by @tcconnally in #524
• fix(embed): add --embedding-model-name for remote embedding endpoints (#525) by @tcconnally in #527
• bench: Lambda dynamic-range benchmark suite (recall@scale, 8-GPU fleet throughput, competitive) by @tcconnally in #533
• fix(#528): configurable LLM timeout via MIMIR_LLM_TIMEOUT_SECS by @tcconnally in #536
• fix(#531): make near-duplicate merging visible and controllable on remember by @tcconnally in #535
• fix(#517): self-heal orphaned-FTS-rowid drift on write; ok:false JSON on write failure (#516) by @tcconnally in #534
• docs(#532): AMD MI300X co-residency deployment reference by @tcconnally in #537
• bench(#530): GPU-contention + agent-economics suite, measured MI300X/H100 rows in PERF.md by @tcconnally in #538
• docs(#523): session lifecycle hook contract — wire the recall/capture/consolidate loop in any MCP client by @tcconnally in #540
• fix(#526): default bundled-embeddings build links on glibc 2.35 by @tcconnally in #539
• feat(#521): mimir_check_failure_pattern — failure-pattern / deja-vu guard by @tcconnally in #541
• feat(#522): install-client — one-command multi-client recall/capture loop wiring by @tcconnally in #542
• feat(#520): opt-in in-session memory capture — perseus-vault capture + mimir_capture by @tcconnally in #543
• perf(#511): hybrid recall at 100K — 295.5 → 80.9 ms p50 (3.7×): two-phase BM25 sparse arm + concurrent arms + stage-timing attribution by @tcconnally in #544
• docs: note the mcpservers.org (Awesome MCP Servers) listing by @tcconnally in #545
• listing: new phrasing in registry + directory source metadata by @tcconnally in #546
• fix(mcp): advertise canonical perseus_vault_* tools only by default (stop tripling the manifest) by @tcconnally in #548
• release: v2.20.0 — canonical-only tool manifest (#548) + capture / install-client / deja-vu + perf by @tcconnally in #549

Full Changelog: v2.19.1...v2.20.0

v2.19.1

What's Changed

• feat: cross-scope promotion — promote recurring facts up the scope ladder by @tcconnally in #501
• bench: LongMemEval end-to-end QA harness — pinned answerer+judge, vs-Zep comparison scaffold by @tcconnally in #502
• fix: bump SCHEMA_VERSION 15 → 16 — deliver #487 usefulness columns to upgraded stores by @tcconnally in #504
• release: v2.19.1 — migration fix for upgraded stores + cross-scope promotion by @tcconnally in #505

Full Changelog: v2.19.0...v2.19.1

v2.19.0

What's Changed

• feat: auto-reinforce memories via derived_from on remember by @tcconnally in #488
• feat: scope as a ranking multiplier in recall by @tcconnally in #489
• feat: one-shot maintain CLI verb — shell entry point for the full hygiene pass by @tcconnally in #495
• feat: maintenance dry-run reports would-do estimates for vacuum and reindex by @tcconnally in #496
• feat: optional in-server maintenance loop (serve --maintain-every <hours>) by @tcconnally in #497
• fix(stats): additive active-only entity counts (gather_stats counted archived rows) by @tcconnally in #498
• feat(web): 'About you' dashboard tab — plain-language, trust-annotated, read-only by @tcconnally in #499
• release: v2.19.0 — hands-off hygiene, active-only stats, About-you tab by @tcconnally in #500

Full Changelog: v2.18.2...v2.19.0

v2.18.2

What's Changed

• release: v2.18.2 — harden release.yml upload race (Closes #483) by @tcconnally in #484

Full Changelog: v2.18.1...v2.18.2

v2.18.1

v2.18.1 — Temporal RAG completion (#472) + temporal-arg client compat…

v2.18.0

v2.18.0 — Temporal RAG: point-in-time semantic recall (#472/#478)

v2.17.5

v2.17.5 — recall category/metadata filter fix (#467/#468); crossbeam-…