Prepare 1.7.9 managed Jev client release

.claude-plugin/marketplace.json

@@ -9,7 +9,7 @@
       "name": "engraphis-memory",
       "source": "./",
       "description": "Discipline for giving agents durable, scoped, explainable memory across sessions and repos with the Engraphis MCP tools.",
-      "version": "1.7.8"
+      "version": "1.7.9"
     }
   ]
 }

.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "engraphis-memory",
-  "version": "1.7.8",
+  "version": "1.7.9",
   "description": "Give agents durable, scoped, explainable memory across sessions and repos via the Engraphis MCP tools. Use when you learn something worth keeping, need prior context before acting, or ask why/how a fact changed. Covers remember/recall, why/timeline, forget/pin/correct, sessions, and code search.",
   "author": {
     "name": "The Engraphis Authors",

.claude-plugin/skill-assets.sha256

@@ -1,5 +1,5 @@
-df65a383a1ff80572fb6ebd9a807dca6d1ffc0f99f373971262de035b8e3622d  .claude-plugin/marketplace.json
-a03b5c38d836651d2c5c52173d21a5adeacfbbc4b80aac991c2154b6e060e174  .claude-plugin/plugin.json
+c1d184247775c8b3bf4d3f185d54fcac983aca8c54638a05d85b21e20db3ae7c  .claude-plugin/marketplace.json
+c5d0c26f28c9ee14092f9deaf24c98dd8bef49d971fef2b7a537ffb1ab9f2887  .claude-plugin/plugin.json
 aeee7a94671ceb306fe2d24c5acc9f2d96ad8a8e7410536566799eea6265f080  skills/engraphis-memory/SKILL.md
 055655db84af07561d002f0c69744313d8413c39f3e873f941f0fa0b1e76dc66  skills/engraphis-memory/references/CONVENTIONS.md
 9d090a03f5b3f36a34d91f66b72c3844591f6915755ac3a6c6ba5f1b16977de5  skills/engraphis-memory/references/SCOPING.md

BENCHMARKS.md

@@ -94,14 +94,14 @@ interpretation and do not count as additional benchmark-quality gains.
 ### Public numeric evidence registry
 
 Every exact public aggregate retained below comes from the checked-in, public-safe
-[`offline-fixtures-v115.json`](docs/benchmark-evidence/offline-fixtures-v115.json) artifact. Its
+[`offline-fixtures-v117.json`](docs/benchmark-evidence/offline-fixtures-v117.json) artifact. Its
 SHA-256 is
-`4d2cd6be29324597f0b201f8c0bd753173be69d24a274dff570c95c0db6ae629`, also recorded in the
+`c64eadffbc7f87938e0e822b7aab33dcfece6fffa5175932ac208465d36df192`, 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
-`ddf3b70d426441a364ef02967f98454435ca045c3840e1c4b81ba05ab7f139bb`. The artifact defines
+`77b140068cd036330b421a0d7668847d196f1bbeaa36c965d68943457002e6eb`. 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)`:
 
@@ -123,10 +123,10 @@ Historical LoCoMo, graph, handoff, consolidation, and security figures remain pr
 source artifacts but are omitted from the current chart until each has a matching immutable,
 public-safe artifact. The chart labels coding outcomes, external datasets, and operational
 capacity as pending evaluation tracks rather than implying scores. Regenerate it with
-`python scripts/render_benchmark_report.py --report docs/benchmark-evidence/offline-fixtures-v115.json --output docs/images/context-efficiency.svg` after selecting the report to publish.
+`python scripts/render_benchmark_report.py --report docs/benchmark-evidence/offline-fixtures-v117.json --output docs/images/context-efficiency.svg` after selecting the report to publish.
 
 The companion examples are also generated from that artifact with
-`python -m scripts.render_benchmark_examples --report docs/benchmark-evidence/offline-fixtures-v115.json --output docs/images/evidence-backed-agent-examples.svg`.
+`python -m scripts.render_benchmark_examples --report docs/benchmark-evidence/offline-fixtures-v117.json --output docs/images/evidence-backed-agent-examples.svg`.
 The historical-to-executable mapping is in
 [`docs/BENCHMARK_CHANGE_COVERAGE.md`](docs/BENCHMARK_CHANGE_COVERAGE.md).
 

CHANGELOG.md

@@ -3,7 +3,16 @@
 All notable changes to Engraphis are documented here. Format loosely follows
 [Keep a Changelog](https://keepachangelog.com/); versions use SemVer.
 
-## [Unreleased]
+## [Unreleased]
+
+## [1.7.9] - 2026-09-29
+
+- Added the saved-session managed Jev transport and Classic/Smart MCP decision route
+  for Pro and Team, with explicit consent and current hosted entitlement checks.
+  Cloud activation remains a separate rollout requirement; see the upgrade and
+  qualification instructions in `docs/RELEASE_1_7_9.md`.
+- Retire consumed refresh credentials when a successful response has an invalid
+  token subject, and keep empty graph-layer selections separate from all-layer cache entries.
 
 - Provide a valid offline decision example in Smart MCP action discovery.
 - Reject coerced numeric/string remote consent before Classic MCP can select a Jev backend.
@@ -36,7 +45,8 @@ All notable changes to Engraphis are documented here. Format loosely follows
 - Rebuild FastMCP settings eagerly to avoid a forward-reference warning on newer SDKs.
 - Preserve CSS gradient tiles, screen blending and reduced-motion opacity in graph PNGs,
   including fractional display scaling; synchronize graph cache metadata across threads.
-- Refreshed the public offline fixtures and source bindings in immutable v111 evidence.
+- Refreshed public offline fixtures and source bindings in immutable v117 evidence for 1.7.9.
+- Campaign adapter capabilities and metrics now report the loaded Engraphis runtime version.
 
 - Added saved project-to-workspace routing and connection instructions so agents can use the
   user's selected workspace. Routine MCP calls inherit an omitted workspace from an authorized

README.md

@@ -84,9 +84,9 @@ neither is an end-to-end question-answer score. Coding outcomes, external datase
 operational capacity remain separate pending evaluation tracks until their artifacts are selected.
 
 These values are evidence IDs `offline-chunking` and `offline-performance` in
-[`offline-fixtures-v115.json`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/benchmark-evidence/offline-fixtures-v115.json),
+[`offline-fixtures-v117.json`](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/benchmark-evidence/offline-fixtures-v117.json),
 SHA-256
-`4d2cd6be29324597f0b201f8c0bd753173be69d24a274dff570c95c0db6ae629`.
+`c64eadffbc7f87938e0e822b7aab33dcfece6fffa5175932ac208465d36df192`.
 [`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. The offline
 fixture registry intentionally excludes external, model-dependent, consolidation, productivity,

docs/HOSTED_PLANS.md

@@ -32,6 +32,12 @@ particular deployment has enabled Jev; the service checks current entitlement an
 Usage and availability are reported by the account portal. No latency, accuracy, or cost-saving
 guarantee is established by client configuration or a successful health check.
 
+The managed transport and MCP decision route require client **1.7.9 or newer**.
+The published 1.7.8 client has an experimental adapter but does not provide this route.
+After 1.7.9 is published, upgrade the Python environment that launches your MCP host
+and restart that host. See the [1.7.9 upgrade and release checklist](RELEASE_1_7_9.md).
+Installing a new client does not enable a deployment whose managed service is disabled.
+
 Set `ENGRAPHIS_DECISION_BACKEND=managed` and connect the installation through the ordinary
 Cloud account flow. The client refreshes its saved session and sends only to that session's
 bound control origin. `auto` chooses this managed route when configured; it never silently

docs/RELEASE_1_7_9.md

@@ -0,0 +1,71 @@
+# Engraphis 1.7.9 release preparation
+
+This candidate packages the reviewed managed Jev client and graph fixes from
+Core PRs #238 through #241. Its base is
+`f10dbc40de2d8bb2b0b7ab9bddb4870f6115a64c`. Preparing a version and validating its
+source does not publish it or enable the Cloud service. The existing 1.7.8
+release remains immutable.
+
+## Customer upgrade after publication
+
+Managed Jev requires both this client and an enabled, qualified Cloud deployment.
+The published 1.7.8 distribution contains an experimental injected adapter but
+does not ship the managed transport or the registered MCP decision tool.
+
+1. Upgrade the Python environment that actually launches MCP. For an MCP
+   installation, run `python -m pip install --upgrade "engraphis[mcp]==1.7.9"`
+   after this version is published. Preserve any other extras your installation
+   needs. Check the installation with `python -m pip show engraphis`.
+2. Keep an existing valid saved Cloud session. If the installation is not
+   connected, use the account portal's generated `engraphis connect` command in
+   that same environment. Never copy a provider key into a customer installation.
+3. Set `ENGRAPHIS_DECISION_BACKEND=managed` in the process environment or the
+   owner-private `~/.engraphis/config.env`. A searched working-directory `.env`
+   is not a trusted configuration source. Restart the MCP host.
+4. Classic MCP exposes `engraphis_decide`. In Smart MCP, discover the decision
+   action and execute it using the returned schema and `engraphis_execute_action`.
+   Every remote request requires literal `allow_remote=true` and a
+   `data_classification` of `public` or `internal`. Review the text before sending
+   it. Requests classified as secret and recognized secret patterns are rejected;
+   `offline_mode=true` prevents remote execution. Pattern filtering cannot detect
+   every secret in arbitrary prose.
+
+The service enforces current membership, entitlement, monthly allowance and a
+separate fleet cap. A local success, fallback, health response or configured
+backend does not establish a successful managed provider request. The account
+portal reports availability and usage. Jev advice does not authorize actions or
+replace deterministic checks.
+
+## Qualification and publication sequence
+
+1. Freeze the final reviewed source and give all package, plugin and commercial
+   version surfaces the same unused version, 1.7.9. Preserve historical evidence;
+   generate new offline fixtures and skill checksums for these bytes.
+2. Pass the required CI checks on the final commit. The release workflow checks
+   that the package version matches `v1.7.9` and that its commit is on protected
+   `main`. Merge and tag creation remain release-owner actions.
+3. Build and retain the exact wheel and source archive. Verify their manifests,
+   dependency/security results, installed-client journeys and reproducibility.
+   Local candidate builds are preparation; the final publication artifacts must
+   match the signed qualification after the final source is selected.
+4. Complete the private full-product ledger, including the selected Cloud/site
+   identities, staging acceptance, recovery/rollback evidence and real Pro/Team
+   saved-session Jev journeys. Run `check_release_readiness.py --require-release`
+   with that ledger and its actual evidence. An unresolved gate stays unresolved.
+5. The existing owner-controlled release authority must review the completed
+   ledger and issue a fresh, time-bounded Ed25519 qualification for the exact
+   commit, distribution hashes, candidate ID and ledger digest. Keep private-key
+   custody outside this repository, workflow artifacts and runners. Preparing
+   this branch does not create a signing key, select an approver or grant approval.
+6. Supply the four protected `release-qualification` environment secrets required
+   by [the qualification contract](RELEASE_QUALIFICATION.md), and obtain its
+   configured authorized review. Verify the envelope with
+   `python -m scripts.verify_release_qualification --dist DIST --commit COMMIT --tag v1.7.9`
+   against those independently selected identities before publication.
+7. Publish through the existing release workflow. Check PyPI and GitHub artifact
+   hashes against the qualified bytes, then verify a fresh customer upgrade and
+   update public availability only from the accepted deployment and client.
+
+The old v1.7.6 and v1.7.8 waivers apply only to their named retained candidates.
+They cannot qualify or authorize this release. See [release readiness](RELEASE_READINESS.md)
+for the mandatory evidence categories and recovery requirements.