engraphis 1.7__tar.gz → 1.7.3__tar.gz

{engraphis-1.7 → engraphis-1.7.3}/BENCHMARKS.md

@@ -11,12 +11,12 @@ For the locked operator sequence for a public canonical run, see
 Every exact public aggregate retained below comes from the checked-in, public-safe
 [`offline-fixtures-v1.json`](docs/benchmark-evidence/offline-fixtures-v1.json) artifact. Its
 SHA-256 is
-`8a74e9f48e25f33d625d4cc5c1b14fec3055891944adccf615c440e84e4b0255`, also recorded in the
+`4d5056d137182ae5cf116c5d59af18b38a7a0ed7731885e9597f63e549cb46b7`, also recorded in the
 adjacent `.sha256` file. The artifact contains no raw questions, answers, prompts, customer data,
 or per-record content fingerprints.
 
 The fixture-suite digest is
-`4d7e40607319cd4bf8caee3897f1e416dbe5b81998b37a7e4839409ee2923537`. The artifact defines
+`f5544b56f009b2fc16dbae992039971899daf2b0095ee8d15bad5914c7f399a9`. The artifact defines
 the digest algorithm and records the SHA-256 of every suite and dataset file. Each evidence ID
 also binds its exact command through `sha256(UTF-8 exact command)`:
 
@@ -72,8 +72,8 @@ frontier-model QA score.
   JSON-safe schema. Payload proxies are sampled once per question, independently of the number
   of timed iterations; they are not serialized MCP envelopes or transport responses. In the
   registered CodeMem run, 26 payload samples total **23,810** full-proxy
-  `engraphis.regex.v1` tokens versus **10,202** compact-proxy tokens, avoiding **13,608** proxy
-  tokens (**57.15% lower**), while 260 recalls are timed. Packed context across the same 26
+  `engraphis.regex.v1` tokens versus **10,982** compact-proxy tokens, avoiding **12,828** proxy
+  tokens (**53.88% lower**), while 260 recalls are timed. Packed context across the same 26
   samples averages **85.38** tokens and reaches **108** under a 1,500-token cap; Recall@5,
   hit@5, and answer-token recall remain 1.000. These aggregates are evidence ID
   `offline-performance` in the registry above. `--filler-memories`, `--candidate-k`, and

{engraphis-1.7 → engraphis-1.7.3}/CHANGELOG.md

@@ -1,7 +1,114 @@
 # Changelog
 
-All notable changes to Engraphis are documented here. Format loosely follows
-[Keep a Changelog](https://keepachangelog.com/); versions use SemVer.
+All notable changes to Engraphis are documented here. Format loosely follows
+[Keep a Changelog](https://keepachangelog.com/); versions use SemVer.
+
+## [Unreleased]
+
+- Added schema 18 content-free command receipts and cross-process source revalidation for
+  corrections, approvals, promotions and merges. Combined memory revisions have expected
+  versions, operation IDs, atomic metadata/history, and typed conflicts.
+- Sync publication uses current canonical state and generation-aware repair; delayed work
+  cannot restore erased vectors. Native-index failures roll back canonical changes.
+- Context retains distinct scoped evidence; synthesis falls back when complete source
+  units, titles, values or conditions are lost. Answer coverage defaults to unknown.
+- Added project-aware memory workflows and paginated record history.
+- Library cursors survive unrelated activity and work across processes. File-backed
+  browsing uses bounded live read snapshots; completed graph migrations are not
+  repeated at ordinary startup.
+- Ask separates answer/preview retries, cancellation and answer coverage. Home uses
+  actionable review state; Explore pauses hidden views through existing renderers.
+- Added content-free diagnostics and build/capability information, strict coding
+  acceptance validation and a file-backed independent-process capacity harness.
+  These provide measurement infrastructure, not verified 100k capacity claims.
+
+## [1.7.3] - 2026-09-07
+
+### Fixed
+
+- Preserved Galaxy carrier lane and kinematic orbit invariants through central-field slider
+  changes, including the global and core cached radii used by the next fixed slice.
+- Refreshed the retained local orbital speed budget when the effective local-gravity control
+  changes, preventing a stale phase cache from masking the slider.
+- Kept high-density Galaxy layouts inside the strict speed cap while maintaining authored
+  carrier and nested local orbit phase.
+- Bounded the zero central-gravity radius response so finite far-field envelopes cannot leave
+  oversized kinematic carrier caches behind, and counted fallback speed-cap activations.
+- Bumped the deterministic Galaxy scene algorithm identity to `galaxy-v13-responsive-compact-orbits`
+  so cached layouts cannot be confused with the revised placement contract.
+
+### Tests
+
+- Added deterministic regressions for central-field cache scaling and local-gravity phase
+  invalidation, alongside the existing 500-body and browser accessibility coverage.
+
+## [1.7.2] - 2026-09-05
+
+### Added
+
+- Added `idx_vector_index_repairs_queue` composite index on `(identity, generation, memory_id)`
+  in `engraphis/core/schema.py` to prevent table scans during external vector repair queue dequeue.
+- Added explicit operator opt-out verification with `403 Forbidden` (`processing_operator_disabled`)
+  for authenticated direct POST requests to `/managed-processing` in `engraphis/routes/v2_api.py`.
+- Added `_only_environment_title_order_changed` in `engraphis/core/resolve.py` ensuring unkeyed
+  facts with permuted environment titles resolve to `NOOP` rather than false conflicts.
+- Added comprehensive reliability regression coverage covering storage concurrency, vector index
+  repair indexing, and managed processing policy enforcement.
+
+### Fixed
+
+- Preserved `[all]` extras fallback for legacy editable installations in `scripts/update.py`
+  when no installation profile is recorded.
+- Fixed external vector index hydration on physical index recreation and rebuilds.
+- Fixed docstring dedenting and contract normalization across Python 3.9 through 3.14.
+
+### Changed
+
+- Bumped `tree-sitter-language-pack` to 1.16.1.
+- Updated `codeql-action`, `anchore/scan-action`, and `anchore/sbom-action` GitHub Actions dependencies.
+
+### Reliability and privacy
+
+- Preserve distinct context claims, qualified sentences and complete units under tight budgets;
+  measure false NOOP outcomes through real write sequences.
+- Preserve separate sources during packing and keep MCP gist responses within the canonical
+  context budget. Response caps retain or omit complete context and report accurate usage.
+- Canonical temporal browsing, server-side Library filtering/pagination, independent Ask states,
+  actionable setup diagnostics and retained installation capabilities.
+- Cross-process write resolution and schema 17 durable vector-index repair, with canonical
+  fallback and bounded NumPy scans. Public engine entrypoints remain compatible.
+- Commit native batch indexing with canonical memory state and roll back both on failure.
+  Retain the established 12,000-memory graph window pending quality evidence for a smaller one.
+- Explicit workspace managed-processing approval; missing legacy policy pauses readable uploads.
+  Requires the compatible cloud migration before rollout. Encrypted sync remains separate.
+- Generated Smart/Classic MCP contract and integration inputs; Pro three-day and Team ten-day
+  trial copy aligned with cloud authority. Real browser and Workers evidence remains distinct
+  from production verification. See `docs/RELIABILITY_PROGRAM.md`.
+- Isolate the manual graph diagnostic on an available local port with a private in-memory
+  server; fail before contacting an existing service when the requested port is occupied.
+
+## [1.7.1] - 2026-09-03
+
+### Fixed
+
+- Isolated stdio transport wire in `engraphis.mcp_server`: redirected `sys.stdout`
+  to `sys.stderr` while preserving the raw binary stream for JSON-RPC wire
+  communication, preventing external library stdout chatter (e.g. PyTorch,
+  Hugging Face, tqdm) from corrupting the wire and triggering `write EOF` stream
+  disconnection errors in Node.js harnesses (Command Code, Cursor, Claude Code, Cline).
+- Added thread-safe singleton initialization with `threading.Lock()` to
+  `engraphis.mcp_server.service()`.
+- Added non-blocking background daemon warmup (`_start_background_warmup()`) in
+  `engraphis.mcp_server` to pre-warm the database and embedder, eliminating
+  cold-start latency and timeout disconnects on the first MCP tool call. Can be
+  bypassed with `ENGRAPHIS_MCP_WARMUP=0`.
+- Added cache-first fast path (`local_files_only=True`) in
+  `SentenceTransformerEmbedder` (`engraphis/backends/embedder_st.py`), allowing
+  locally cached models to initialize in ~0.3s without network calls or remote
+  registry checks.
+- Added embedder forward-pass diagnostic check to `engraphis-init --check` and
+  added `engraphis-init --prefetch` command to download and cache model weights
+  during setup.
 
 ## [1.7] - 2026-09-03
 
@@ -13,9 +120,8 @@ All notable changes to Engraphis are documented here. Format loosely follows
   (savings_ratio 0.0 -> 0.4975) with no caller-side arguments. The packer is the
   existing 1.6 contract; the change just makes it the default fast path.
 - Smart MCP `engraphis_remember` now accepts and forwards `subject_key` and
-  `claim_kind` to the classic tool. Without this, every keyed write silently stored
-  empty keys because the served gateway surface dropped the parameters; the
-  documented safe-supersession mechanism is now reachable through MCP.
+  `claim_kind` to the classic tool, so the documented safe-supersession
+  mechanism is reachable through MCP.
 - A new integration at `integrations/commandcode/session_start_hook.py` (with
   `scripts/install_cc_hook.py` for idempotent user-scope install/uninstall) wires
   durable-memory recall into Command Code's SessionStart lifecycle: each new
@@ -27,7 +133,11 @@ All notable changes to Engraphis are documented here. Format loosely follows
   / `ENGRAPHIS_RERANK_MODEL`). Evaluated offline on the bundled retrieval gates
   (sample.jsonl, codemem.jsonl, k=5): hit@5 stays at 1.0 with zero per-question
   regressions, MRR@5 lifts 0.889 -> 0.944 (sample) and 0.962 -> 0.981 (codemem),
-  with ~15 ms per query added. Not the default; flip with a one-line config.
+  with ~15 ms per query added. Not the default; set the value in the trusted
+  config file (`~/.engraphis/config.env` on the operator account, or as a
+  process environment variable); Engraphis deliberately does not read the
+  CWD `.env`, so editing `./.env` and restarting leaves the identity
+  reranker active. Restart the MCP server and dashboard after the change.
 
 ### Changed
 
@@ -55,10 +165,8 @@ All notable changes to Engraphis are documented here. Format loosely follows
 
 ### Fixed
 
-- The Smart MCP gateway `engraphis_remember` binding was silently dropping
-  `subject_key` and `claim_kind`; this is the underlying cause of the
-  benchmark correction-miss pattern that the reworded-correction detector
-  then had to compensate for.
+- The Smart MCP gateway `engraphis_remember` now forwards `subject_key` and
+  `claim_kind` end to end, matching the **Added** entry above.
 
 ### Operational
 

{engraphis-1.7 → engraphis-1.7.3}/MANIFEST.in

@@ -10,6 +10,8 @@ recursive-include engraphis/dashboard_assets/vendor *
 include engraphis/commercial_manifest.json
 include LICENSE NOTICE README.md CHANGELOG.md BENCHMARKS.md
 include docs/RECALL_RECOVERY.md
+include docs/REWORK_EXECUTION.md docs/ENGINE_CAPACITY_PROTOCOL.md docs/CODING_ACCEPTANCE_CORPUS.md
+include docs/PAID_EVALUATION_PROPOSAL.md
 include docs/DOCUMENT_IMPORT.md docs/OBSIDIAN_IMPORT.md
 include docs/images/context-efficiency.svg
 include pyproject.toml

{engraphis-1.7 → engraphis-1.7.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: engraphis
-Version: 1.7
+Version: 1.7.3
 Summary: Local-first AI memory engine for agents — Ebbinghaus decay, interaction-aware recall, bi-temporal facts, hybrid retrieval, and an MCP server. You bring the LLM.
 Author: The Engraphis Authors
 License-Expression: Apache-2.0
@@ -52,7 +52,7 @@ Requires-Dist: sentence-transformers>=2.7; python_version >= "3.10" and extra ==
 Provides-Extra: code
 Requires-Dist: tree-sitter>=0.23; python_version >= "3.10" and extra == "code"
 Requires-Dist: tree-sitter-language-pack==0.9.0; python_version < "3.10" and extra == "code"
-Requires-Dist: tree-sitter-language-pack==1.15.8; python_version >= "3.10" and extra == "code"
+Requires-Dist: tree-sitter-language-pack==1.16.1; python_version >= "3.10" and extra == "code"
 Provides-Extra: documents
 Requires-Dist: pypdf>=6.15.0; extra == "documents"
 Requires-Dist: Pillow>=12.3.0; python_version >= "3.10" and extra == "documents"
@@ -78,7 +78,7 @@ Requires-Dist: pydantic-settings>=2.14.2; python_version >= "3.10" and extra ==
 Requires-Dist: cryptography>=50.0.0; python_version >= "3.10" and extra == "all"
 Requires-Dist: tree-sitter>=0.23; python_version >= "3.10" and extra == "all"
 Requires-Dist: tree-sitter-language-pack==0.9.0; python_version < "3.10" and extra == "all"
-Requires-Dist: tree-sitter-language-pack==1.15.8; python_version >= "3.10" and extra == "all"
+Requires-Dist: tree-sitter-language-pack==1.16.1; python_version >= "3.10" and extra == "all"
 Requires-Dist: pypdf>=6.15.0; extra == "all"
 Requires-Dist: Pillow>=12.3.0; python_version >= "3.10" and extra == "all"
 Requires-Dist: pytesseract>=0.3.10; python_version >= "3.10" and extra == "all"
@@ -107,7 +107,7 @@ Requires-Dist: pydantic-settings>=2.14.2; python_version >= "3.10" and extra ==
 Requires-Dist: cryptography>=50.0.0; python_version >= "3.10" and extra == "test"
 Requires-Dist: tree-sitter>=0.23; python_version >= "3.10" and extra == "test"
 Requires-Dist: tree-sitter-language-pack==0.9.0; python_version < "3.10" and extra == "test"
-Requires-Dist: tree-sitter-language-pack==1.15.8; python_version >= "3.10" and extra == "test"
+Requires-Dist: tree-sitter-language-pack==1.16.1; python_version >= "3.10" and extra == "test"
 Requires-Dist: pypdf>=6.15.0; extra == "test"
 Requires-Dist: Pillow>=12.3.0; python_version >= "3.10" and extra == "test"
 Requires-Dist: pytesseract>=0.3.10; python_version >= "3.10" and extra == "test"
@@ -162,7 +162,7 @@ by default, or accept an explicit workspace plus optional `from_ts`, `to_ts`, an
 `release_version` filters.
 
 <p align="center">
-  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/main/docs/images/context-efficiency.svg" alt="Dark chart of local measurements and deterministic fixtures, including a local LoCoMo diagnostic marked with an asterisk. Cross-session handoff satisfaction rises from 3 of 15 queries with the last memories to 15 of 15 with proactive ranking or a consolidated summary. Intent-layered graph routing rises from 0 of 3 to 3 of 3 correct top-1 targets, and two-hop graph recall rises from 0 of 3 with one-hop expansion to 3 of 3 with Personalized PageRank. Consolidation-aware ranking selects the expected digest in 2 of 2 summary cases instead of 0 of 2 for the baseline. Structure-aware chunks reduce context from 740.3 to 214.3 tokens and the smallest evidence-holding memory from 162.2 to 42.4 tokens. A compact JSON-shape proxy uses 10,202 rather than 23,810 tokens. Grounded recall makes 10 of 10 correct decisions and packed context averages 85.38 tokens under a 1,500-token cap." width="100%">
+  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/main/docs/images/context-efficiency.svg" alt="Dark chart of local measurements and deterministic fixtures, including a local LoCoMo diagnostic marked with an asterisk. Cross-session handoff satisfaction rises from 3 of 15 queries with the last memories to 15 of 15 with proactive ranking or a consolidated summary. Intent-layered graph routing rises from 0 of 3 to 3 of 3 correct top-1 targets, and two-hop graph recall rises from 0 of 3 with one-hop expansion to 3 of 3 with Personalized PageRank. Consolidation-aware ranking selects the expected digest in 2 of 2 summary cases instead of 0 of 2 for the baseline. Structure-aware chunks reduce context from 740.3 to 214.3 tokens and the smallest evidence-holding memory from 162.2 to 42.4 tokens. A compact JSON-shape proxy uses 10,982 rather than 23,810 tokens. Grounded recall makes 11 of 11 correct decisions and packed context averages 85.38 tokens under a 1,500-token cap." width="100%">
   <br>
   <sup>Less repeated history means more room for the task, tools, and useful evidence.</sup>
 </p>
@@ -191,13 +191,13 @@ its counting boundary explicit.
 |---|---|---|---|
 | Retrieved top-5 memory content, averaged per question | Whole documents: **740.3** tokens → structure-aware chunks: **214.3** tokens | **526.0 fewer tokens per question** (**71.1% lower**, about **3.5× smaller**) | Recall@5 **1.000** in both modes across 6 documents and 18 questions |
 | Smallest returned memory that contains the reference evidence | Whole documents: **162.2** tokens → chunks: **42.4** tokens | **119.8 fewer tokens to evidence** (**73.9% lower**, about **3.8× smaller**) | The same 18 questions had a returned evidence-holding memory in both modes |
-| Full versus compact recall payload proxy across one 26-question pass within a 260-timed-recall CodeMem run | Full proxy: **23,810** `engraphis.regex.v1` tokens → compact proxy: **10,202** tokens | **13,608 proxy tokens avoided** (**57.15% lower**) | 26 payload samples; 260 timed recalls; Recall@5, hit@5, and answer-token recall all **1.000** |
+| Full versus compact recall payload proxy across one 26-question pass within a 260-timed-recall CodeMem run | Full proxy: **23,810** `engraphis.regex.v1` tokens → compact proxy: **10,982** tokens | **12,828 proxy tokens avoided** (**53.88% lower**) | 26 payload samples; 260 timed recalls; Recall@5, hit@5, and answer-token recall all **1.000** |
 | Packed prompt-context usage in the same 26-question CodeMem sample pass | Hard budget: **1,500** tokens; observed mean: **85.38**; observed maximum: **108** | A hard cap prevents a recall from exceeding its configured context budget | This is usage accounting, not a before/after savings comparison |
 
 These values are evidence IDs `offline-chunking` and `offline-performance` in
 [`offline-fixtures-v1.json`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/benchmark-evidence/offline-fixtures-v1.json),
 SHA-256
-`8a74e9f48e25f33d625d4cc5c1b14fec3055891944adccf615c440e84e4b0255`.
+`4d5056d137182ae5cf116c5d59af18b38a7a0ed7731885e9597f63e549cb46b7`.
 [`BENCHMARKS.md`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/BENCHMARKS.md#public-numeric-evidence-registry)
 records the matching suite digest, exact commands, and per-command config digests. External,
 model-dependent, consolidation, productivity, and latency results remain unpublished until the
@@ -922,7 +922,7 @@ file. It never searches the working directory for `.env`, and explicit process v
 | `ENGRAPHIS_CLOUD_REFRESH_CREDENTIAL` | Not set | Bootstrap-only rotating hosted credential; after first use the owner-only cloud session replacement takes precedence |
 | `ENGRAPHIS_CLOUD_TOKEN_SUBJECT` | `member` | Subject fixed during hosted bootstrap (`device` or `member`); set explicitly with an environment-only refresh credential |
 | `ENGRAPHIS_CLOUD_ACCESS_TOKEN` | Not set | Optional short-lived access token for ephemeral jobs |
-| `ENGRAPHIS_MANAGED_COMPUTE_CONSENT` | *(auto)* | Operator override only; default follows whether a cloud session is configured (connected = allowed, local-only = never). `0` opts a connected installation out; `1` permits local snapshot preparation but does not create a cloud credential or authorize an upload |
+| `ENGRAPHIS_MANAGED_COMPUTE_CONSENT` | *(unset)* | Deny-only operator override: `0` pauses readable managed processing. A truthy value cannot grant approval. Each workspace requires explicit confirmation in Manage → Settings; encrypted sync is separate |
 
 The optional cross-encoder reranker is model- and hardware-dependent. Treat its quality and
 latency as deployment-specific until a versioned model identity, exact configuration, and
@@ -987,3 +987,19 @@ under Apache-2.0 keeps that grant; later releases cannot retroactively withdraw
 official hosted control plane, its production credentials and records, managed operations,
 support, and future separately delivered commercial modules are outside the public source
 grant. See [`docs/LICENSING.md`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/LICENSING.md) for the complete boundary.
+
+### Reliability implementation candidate
+
+The current source uses schema 18 for durable, content-free vector-index repair and
+atomic memory-command receipts. Upgrades use the existing verified-backup migration path.
+The [rework execution register](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/REWORK_EXECUTION.md) records the current findings,
+compatibility decisions, acceptance evidence, remaining work and recovery procedure.
+See [the reliability program](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/RELIABILITY_PROGRAM.md) for exact implementation,
+validation, migration and release boundaries. Managed processing now requires explicit
+workspace approval in Settings. Existing installations start with readable
+uploads paused until confirmed; connecting an account does not grant approval.
+
+For setup diagnostics use `engraphis-init --check --json`. New configurations get an
+owner-private local API token. Existing configs are preserved. Record selected install
+capabilities with `engraphis-init --extras server,mcp` or `--extras none`; future updates
+preserve that choice. `ENGRAPHIS_UPDATE_EXTRAS` remains an explicit override.

{engraphis-1.7 → engraphis-1.7.3}/README.md

@@ -44,7 +44,7 @@ by default, or accept an explicit workspace plus optional `from_ts`, `to_ts`, an
 `release_version` filters.
 
 <p align="center">
-  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/main/docs/images/context-efficiency.svg" alt="Dark chart of local measurements and deterministic fixtures, including a local LoCoMo diagnostic marked with an asterisk. Cross-session handoff satisfaction rises from 3 of 15 queries with the last memories to 15 of 15 with proactive ranking or a consolidated summary. Intent-layered graph routing rises from 0 of 3 to 3 of 3 correct top-1 targets, and two-hop graph recall rises from 0 of 3 with one-hop expansion to 3 of 3 with Personalized PageRank. Consolidation-aware ranking selects the expected digest in 2 of 2 summary cases instead of 0 of 2 for the baseline. Structure-aware chunks reduce context from 740.3 to 214.3 tokens and the smallest evidence-holding memory from 162.2 to 42.4 tokens. A compact JSON-shape proxy uses 10,202 rather than 23,810 tokens. Grounded recall makes 10 of 10 correct decisions and packed context averages 85.38 tokens under a 1,500-token cap." width="100%">
+  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/main/docs/images/context-efficiency.svg" alt="Dark chart of local measurements and deterministic fixtures, including a local LoCoMo diagnostic marked with an asterisk. Cross-session handoff satisfaction rises from 3 of 15 queries with the last memories to 15 of 15 with proactive ranking or a consolidated summary. Intent-layered graph routing rises from 0 of 3 to 3 of 3 correct top-1 targets, and two-hop graph recall rises from 0 of 3 with one-hop expansion to 3 of 3 with Personalized PageRank. Consolidation-aware ranking selects the expected digest in 2 of 2 summary cases instead of 0 of 2 for the baseline. Structure-aware chunks reduce context from 740.3 to 214.3 tokens and the smallest evidence-holding memory from 162.2 to 42.4 tokens. A compact JSON-shape proxy uses 10,982 rather than 23,810 tokens. Grounded recall makes 11 of 11 correct decisions and packed context averages 85.38 tokens under a 1,500-token cap." width="100%">
   <br>
   <sup>Less repeated history means more room for the task, tools, and useful evidence.</sup>
 </p>
@@ -73,13 +73,13 @@ its counting boundary explicit.
 |---|---|---|---|
 | Retrieved top-5 memory content, averaged per question | Whole documents: **740.3** tokens → structure-aware chunks: **214.3** tokens | **526.0 fewer tokens per question** (**71.1% lower**, about **3.5× smaller**) | Recall@5 **1.000** in both modes across 6 documents and 18 questions |
 | Smallest returned memory that contains the reference evidence | Whole documents: **162.2** tokens → chunks: **42.4** tokens | **119.8 fewer tokens to evidence** (**73.9% lower**, about **3.8× smaller**) | The same 18 questions had a returned evidence-holding memory in both modes |
-| Full versus compact recall payload proxy across one 26-question pass within a 260-timed-recall CodeMem run | Full proxy: **23,810** `engraphis.regex.v1` tokens → compact proxy: **10,202** tokens | **13,608 proxy tokens avoided** (**57.15% lower**) | 26 payload samples; 260 timed recalls; Recall@5, hit@5, and answer-token recall all **1.000** |
+| Full versus compact recall payload proxy across one 26-question pass within a 260-timed-recall CodeMem run | Full proxy: **23,810** `engraphis.regex.v1` tokens → compact proxy: **10,982** tokens | **12,828 proxy tokens avoided** (**53.88% lower**) | 26 payload samples; 260 timed recalls; Recall@5, hit@5, and answer-token recall all **1.000** |
 | Packed prompt-context usage in the same 26-question CodeMem sample pass | Hard budget: **1,500** tokens; observed mean: **85.38**; observed maximum: **108** | A hard cap prevents a recall from exceeding its configured context budget | This is usage accounting, not a before/after savings comparison |
 
 These values are evidence IDs `offline-chunking` and `offline-performance` in
 [`offline-fixtures-v1.json`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/benchmark-evidence/offline-fixtures-v1.json),
 SHA-256
-`8a74e9f48e25f33d625d4cc5c1b14fec3055891944adccf615c440e84e4b0255`.
+`4d5056d137182ae5cf116c5d59af18b38a7a0ed7731885e9597f63e549cb46b7`.
 [`BENCHMARKS.md`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/BENCHMARKS.md#public-numeric-evidence-registry)
 records the matching suite digest, exact commands, and per-command config digests. External,
 model-dependent, consolidation, productivity, and latency results remain unpublished until the
@@ -804,7 +804,7 @@ file. It never searches the working directory for `.env`, and explicit process v
 | `ENGRAPHIS_CLOUD_REFRESH_CREDENTIAL` | Not set | Bootstrap-only rotating hosted credential; after first use the owner-only cloud session replacement takes precedence |
 | `ENGRAPHIS_CLOUD_TOKEN_SUBJECT` | `member` | Subject fixed during hosted bootstrap (`device` or `member`); set explicitly with an environment-only refresh credential |
 | `ENGRAPHIS_CLOUD_ACCESS_TOKEN` | Not set | Optional short-lived access token for ephemeral jobs |
-| `ENGRAPHIS_MANAGED_COMPUTE_CONSENT` | *(auto)* | Operator override only; default follows whether a cloud session is configured (connected = allowed, local-only = never). `0` opts a connected installation out; `1` permits local snapshot preparation but does not create a cloud credential or authorize an upload |
+| `ENGRAPHIS_MANAGED_COMPUTE_CONSENT` | *(unset)* | Deny-only operator override: `0` pauses readable managed processing. A truthy value cannot grant approval. Each workspace requires explicit confirmation in Manage → Settings; encrypted sync is separate |
 
 The optional cross-encoder reranker is model- and hardware-dependent. Treat its quality and
 latency as deployment-specific until a versioned model identity, exact configuration, and
@@ -869,3 +869,19 @@ under Apache-2.0 keeps that grant; later releases cannot retroactively withdraw
 official hosted control plane, its production credentials and records, managed operations,
 support, and future separately delivered commercial modules are outside the public source
 grant. See [`docs/LICENSING.md`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/LICENSING.md) for the complete boundary.
+
+### Reliability implementation candidate
+
+The current source uses schema 18 for durable, content-free vector-index repair and
+atomic memory-command receipts. Upgrades use the existing verified-backup migration path.
+The [rework execution register](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/REWORK_EXECUTION.md) records the current findings,
+compatibility decisions, acceptance evidence, remaining work and recovery procedure.
+See [the reliability program](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/RELIABILITY_PROGRAM.md) for exact implementation,
+validation, migration and release boundaries. Managed processing now requires explicit
+workspace approval in Settings. Existing installations start with readable
+uploads paused until confirmed; connecting an account does not grant approval.
+
+For setup diagnostics use `engraphis-init --check --json`. New configurations get an
+owner-private local API token. Existing configs are preserved. Record selected install
+capabilities with `engraphis-init --extras server,mcp` or `--extras none`; future updates
+preserve that choice. `ENGRAPHIS_UPDATE_EXTRAS` remains an explicit override.

engraphis-1.7.3/docs/CODING_ACCEPTANCE_CORPUS.md

@@ -0,0 +1,159 @@
+# Independent coding-memory acceptance corpus
+
+`eval.coding_acceptance` defines a versioned manifest and validates structural
+and provenance declarations. It does not contain an independently authored corpus,
+execute model calls, prove authorship, or authorize evaluation spending.
+Existing deterministic fixtures remain useful regression tests and keep their
+implementation-authored or synthetic labels.
+
+## Frozen task structure
+
+Require **400 scenarios in 40 repository families**. Each family supplies exactly
+one scenario in each of these ten categories:
+
+| Category | Required behavior to label |
+| --- | --- |
+| Corrections | Real remember/correct/recall sequence; false NOOP and false merge |
+| Temporal history | Future/current/expired facts, known-at and valid-at history |
+| Scope boundaries | Workspace/repository/session ownership and equal-content facts |
+| Paraphrases | Retrieval with a pinned semantic embedder; no supplied similarities |
+| Code relationships | Symbols, files/calls and explicit evidence bridges |
+| Unsupported questions | Grounded abstention and unsupported-answer rate |
+| Poisoning | Untrusted ingested instructions cannot change review/authorization |
+| Conditions and values | Complete subjects, numbers, units, negation and conditions |
+| Long documents | Evidence beyond early chunks; budget omissions remain explicit |
+| Multilingual | Whole conditions and titles retained across languages |
+
+The split is by family, never by question: **80 development, 80 validation and
+240 held-out** scenarios. Sort families by SHA-256 of `split_seed:family_id`; assign
+the first eight families to development, the next eight to validation and the
+remaining 24 to held-out. Freeze the family identities and seed before tuning.
+Human reviewers must also check related forks, shared templates and paraphrased
+tasks across families; an identifier validator cannot detect semantic leakage.
+
+Each scenario binds an ID, family, category, split, declared origin, author IDs,
+source SHA-256, oracle SHA-256 and required evidence IDs. The manifest binds the
+implementation authors, two distinct reviewers, a timezone-aware freeze time and
+an authorship-attestation SHA-256. Independent task authors, implementation
+authors and reviewers must not overlap. Licensing, consent, authorship method,
+family lineage and review/adjudication should be recorded in the attestation.
+
+Source and oracle hashes reference separately frozen task artifacts. Those
+artifacts must contain the initial history/repository, ordered operations, query,
+scope/time anchors, expected memory transitions, evidence units, unsupported
+claims, deterministic task oracle and safe disposable setup/cleanup instructions.
+This manifest validator checks the hash format, not the referenced bytes or the
+behavior of those tasks. A future corpus loader must verify the bytes and execute
+the sequence before a report can claim coverage.
+
+## Validator use and evidence boundaries
+
+```console
+python -m eval.coding_acceptance
+python -m eval.coding_acceptance --corpus corpus.json --attestation authorship.txt
+python -m eval.coding_acceptance --corpus generated-fixture.json --fixture
+```
+
+With no corpus, the command emits portable JSON Schema. The Python validator also
+enforces cross-row counts, categories, exact family splits, unique IDs, author
+separation and the matched arms/budgets. Candidate independent validation requires
+the actual attestation file to match its digest. `--fixture` permits a structural
+check while retaining the declared origin.
+
+Synthetic and implementation-authored manifests are rejected by the independent
+candidate mode. Changing only a manifest's origin cannot relabel its scenarios.
+No automated metadata check can prove that every declaration is truthful: even a
+matching attestation returns `independently_authored_verified=false`,
+`publication_ready=false` and at most `authorship_status=attested_unverified`.
+The generated 400-row test data is explicitly a test of this validator, never
+400 completed acceptance tasks or independent evidence.
+
+## Matched comparison bindings
+
+Use **five arms** (no memory, full history, lexical, dense, hybrid) at **three
+retrieval budgets** (512, 1,500, 4,096). A binding file contains exactly 15 cells.
+All cells share corpus/prompt/source-visibility hashes, reader and tokenizer
+identities/revisions, local semantic embedding identity/revision, random seed,
+and hard input/output ceilings. Hashing cannot be declared a semantic dense
+baseline. Validate with:
+
+```console
+python -m eval.coding_acceptance --bindings bindings.json
+```
+
+Each cell contains `arm`, `token_budget`, `corpus_sha256`, `prompt_sha256`,
+`reader_id`, `reader_revision`, `tokenizer_id`, `tokenizer_revision`,
+`max_input_tokens`, `max_output_tokens`, `embedding_id`, `embedding_revision`,
+`embedding_semantic`, `source_visibility_sha256`, `history_overflow_policy` and
+`seed`. All fields except arm/budget must be identical. The overflow policy is
+`fail_preflight`: full history receives all eligible history within the common
+hard input ceiling, or that task fails preflight for every matched arm. Do not
+silently truncate full history. No-memory/full-history are repeated in each
+budget stratum and must not be counted as new independent tasks.
+
+Bindings are a preflight contract, not an execution adapter. Validation reports
+zero model calls and `paid_run_authorized=false`. The cost and execution-approval
+boundary remains [the separate paid evaluation proposal](PAID_EVALUATION_PROPOSAL.md).
+
+## Scoring and release gates
+
+Keep candidate discovery, write resolution, ranking, packing, grounding and final
+task success separate. Citation validity asks whether a cited source supports a
+claim; answer completeness asks whether all required evidence was covered. Count
+false NOOPs, false merges, unsupported assertions, missed conditions, scope/time
+violations and abstention independently. Record context/model tokens, latency,
+memory, disk and costs without using token reduction as evidence of correctness.
+
+Run development tasks while tuning; freeze the candidate and all bindings before
+validation; reserve held-out families for the final comparison. Require zero
+critical integrity/authorization violations. Predeclare one percentage point as
+the task-success non-inferiority margin and require paired family-cluster 95%
+intervals, with category-level failures reported. If uncertainty cannot exclude
+the declared degradation, retain the current default. Missing/error outcomes
+stay in the denominator; do not silently retry or exclude difficult tasks.
+
+The remaining work is actual independent authorship and adjudication, frozen
+artifact/license verification, the executable five-arm adapter, task-oracle and
+budget-interruption tests, and approved paired validation/held-out runs. Passing
+this validator establishes none of those operational results.
+
+## Matched outcome aggregation
+
+`eval.task_pairs` analyzes one matched arm/budget comparison using the full corpus
+manifest and its 15 run bindings. It does not execute tasks or call models.
+
+```console
+python -m eval.task_pairs --baseline baseline.json --candidate candidate.json --corpus corpus.json --bindings bindings.json --output paired-results.json
+```
+
+The default comparison is full history versus hybrid, budget 1,500, 240 held-out
+tasks and three repetitions. Outcomes bind `scenario_id`, source/oracle hashes,
+the canonical run-binding hash, `repetition` (zero-based), `status`, boolean
+`task_success`, `evidence_retained_ids`, `critical_violations`, implementation
+source hash and a unique `run_id`. Only arm identity and the explicitly reported
+implementation source identity may differ between paired conditions. The common
+reader, model, tokenizer, prompts, resource limits and source visibility are bound
+through the validated matrix. Duplicate attempts and incompatible bindings fail.
+
+Missing attempts remain zero-success/zero-retention failures in the predeclared
+denominator and disable non-inferiority. Tasks with no required evidence report
+retention as unavailable, rather than receiving a free perfect score. Task success
+and evidence retention have separate means and candidate-minus-baseline intervals;
+required-evidence coverage does not establish citation validity or answer completeness.
+
+The 95% percentile bootstrap first averages paired effects within each repository
+family, then resamples families as blocks. Repeated answers never become new
+independent families. Report category intervals as exploratory; they have no
+multiple-comparison adjustment and cannot select a winning default independently.
+At least 24 held-out families, 240 tasks and three repetitions are required before
+the statistical non-inferiority check at margin 0.01 is eligible. The lower bound
+must strictly exceed -0.01; incomplete/error pairs, critical violations, synthetic
+authorship and degenerate intervals disable that check. Constant effects produce
+no defensible resampling uncertainty and cannot manufacture a pass, even when
+both conditions report perfect scores. These minima are not a power guarantee.
+
+Even statistical support retains `default_action=keep_current_defaults` and
+`independently_authored_verified=false`: matching declarations and attestations
+cannot prove real execution or independent authorship. Completing one pair is
+also not execution of the full five-arm corpus. Local tests and benchmarking are
+already authorized; paid calls remain governed by their separate proposal.