diff --git a/CHANGELOG.md b/CHANGELOG.md
index b8831a6..8760703 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,32 @@
 
 All notable changes to Process Engine are recorded here.
 
+## 1.9.6 — 2026-08-08
+
+**Prompts-only, Turnstone-native — bounded identity-preservation + repository hygiene** (engine v8.0):
+
+### Changed
+
+- **Identity-critical fact preservation:** intake now preserves material entity ↔ role ↔ identifier/alias relationships, explicitly distinct entities remain distinct, source-attributed identity conflicts remain unresolved until evidence disambiguates them, and genuine ambiguity is preserved.
+- **Pattern placement:** mutable operational/contextual identity facts default to package references/resources while stable behavior-defining identity may remain in personas.
+- **Review:** wrong-entity/ambiguity checking now produces REVISE when ambiguity could change the target.
+- **Trial:** identity-sensitive packages include contradiction/alias/near-match cases.
+- **Repository hygiene:** removed obsolete drafts→converter generated-source architecture; GitHub committed content is canonical.
+- **Structural validation:** retained small repository structural validation and made GitHub CI genuinely fail closed; structural PASS is explicitly separated from behavioral/release approval.
+
+### Validated
+
+- Canonical repository and live `process-engine-generator` runtime were reconciled and read back before product trials.
+- Two live late-invocation trials on the canonical Turnstone runtime passed across materially different domains:
+  - technical backup-appliance context;
+  - customer/order-record context.
+- Across the tested cases, Process Engine successfully reused relevant accumulated workstream context without repeated questions, preserved provenance/corrections/ambiguity, excluded irrelevant chatter, and maintained Summary Gate discipline.
+- These trials support the observed behavior in the tested scenarios; they do **not** claim a universal transcript-ingestion guarantee.
+
+### Evidence qualification
+
+The old GitHub workflow previously ran the structural validator with `|| true`, so a green check alone was not proof the validator passed; the tooling-hygiene change corrected this to fail-closed. Historical local structural PASS observations remain valid.
+
 ## 1.9.5 — 2026-08-04 (clean slate)
 
 **Prompts-only, Turnstone-native — the engine as a methodology factory** (engine v8.0):
diff --git a/README.md b/README.md
index 661395f..56f97dd 100644
--- a/README.md
+++ b/README.md
@@ -2,9 +2,9 @@
 
 **Turn your working knowledge, examples, and operating requirements into a Turnstone-native agent package — persona, project, skills, templates — through a gated pipeline with your sign-off at every step.**
 
-Process Engine is a persona-and-skills generator for Turnstone: tell it what you want, share what you have, and it produces a complete Turnstone package through a gated pipeline. The engine runs on prompts. Turnstone's native governance mechanisms (prompt policy, advisory judge) enforce the gates mechanically.
+Process Engine is a persona-and-skills generator for Turnstone: tell it what you want, share what you have, and it produces a complete Turnstone package through a gated pipeline. The engine runs on prompts. Turnstone provides the native governance surfaces Process Engine uses for persistent context and advisory evidence; operator approval remains the final authority at the defined gates.
 
-> Process Engine v1.9.5 is the **prompts-only, Turnstone-native** reference implementation. The prompt+code successor is [Method Factory](https://github.com/RedEyeNinja-BKK/Method-Factory).
+> Process Engine v1.9.6 is the **prompts-only, Turnstone-native** reference implementation. The prompt+code successor is [Method Factory](https://github.com/RedEyeNinja-BKK/Method-Factory).
 
 ## What it does
 
@@ -29,7 +29,7 @@ flowchart TB
     M --> G
 ```
 
-Every gate requires operator sign-off. Turnstone's native prompt policy and advisory judge enforce this mechanically — the engine doesn't need to recite governance policy; Turnstone handles it.
+The meaningful operator gates are the Summary Gate (confirm material/intent/objective before generation), Review (operator accepts the reviewed package), Trial (operator accepts trial evidence/readiness), and Ship (operator authorizes deployment). Turnstone's prompt policy provides durable contextual guidance and the advisory judge provides review/trial evidence; neither silently replaces operator approval — the engine doesn't need to recite governance policy; Turnstone supplies the native mechanisms around it.
 
 > See a real run end to end: [case study: a shop package built, trialed, and shipped](case-study/case-study-first-run.md).
 
@@ -37,14 +37,14 @@ Every gate requires operator sign-off. Turnstone's native prompt policy and advi
 
 - **Intent before output** — collects what you already have, clarifies what you mean, asks what "good" looks like *before* it writes anything.
 - **Input-agnostic intake** — links, pasted text, files, store pages, existing skills, documents. Everything is assessed; incorporated or excluded with a recorded reason.
-- **Gates, not guesswork** — nothing ships without your review; nothing ships untried. Turnstone's native governance enforces these mechanically.
+- **Gates, not guesswork** — nothing ships without your review; nothing ships untried. You are the gate; the engine's workflow carries the discipline and Turnstone supplies the native surfaces around it.
 - **Adapts, never copies** — techniques are extracted and attributed, original instructions authored. Sources traced from intent through deployment.
 
 Process Engine is domain-neutral in package structure, artifact-specific, and runtime-aware. It's been exercised on Etsy store management, supplement product listings, incident response, employee onboarding, database backup operations, and financial planning advisories — not just software engineering.
 
 ## Why Turnstone
 
-Process Engine runs on Turnstone because Turnstone's governance surface is what the engine produces. Every generated package includes governance objects — prompt policy and advisory judge rules — that Turnstone enforces mechanically. The engine's "operator is the gate" philosophy maps directly to Turnstone's native approval surfaces.
+Process Engine runs on Turnstone because Turnstone's governance surface is what the engine produces. Every generated package includes governance objects — prompt policy and advisory judge rules — that provide persistent context and advisory evidence; the engine's "operator is the gate" philosophy maps directly to Turnstone's native approval surfaces.
 
 | Harness | Governance surface |
 |---|---|
@@ -53,7 +53,7 @@ Process Engine runs on Turnstone because Turnstone's governance surface is what
 | OpenClaw | Agent + channels (Discord/LINE) — emission-focused |
 | Claude | Commercial harness — no governance layer |
 
-The engine is built for Turnstone and deployed on Turnstone. Turnstone's native mechanisms — projects, personas, skills, prompt templates, the judge — are the engine's platform and enforcement layer. The model generates packages; Turnstone owns the guardrails.
+The engine is built for Turnstone and deployed on Turnstone. Turnstone's native mechanisms — projects, personas, skills, prompt templates, the judge — are the engine's platform and governance surfaces; the model generates packages and the operator is the final gate.
 
 ## The engine's components
 
@@ -105,7 +105,7 @@ The engine never copies what you give it — it extracts the techniques and inte
 Process Engine runs on Turnstone. Full governance included:
 
 1. Create the project: `POST /v1/api/projects`.
-2. Create the persona: `POST /v1/api/admin/personas` (base_prompt = `persona.md`).
+2. Create the persona: `POST /v1/api/admin/personas` (base_prompt = `persona.md`). `persona.md` defines the canonical **Process Engine generator persona**; deploy it under a distinct generator identity (e.g. `process-engine-generator`) so it does not collide with any development/maintenance persona used on the same Turnstone instance.
 3. Create the six skills and six templates via the skills API (prompt_templates store); attach the seven references as skill resources on `process-engine-core`.
 4. Create the engine's governance wiring: the prompt policy and advisory judge rules.
 5. Verify by reading every created object back (GET), then run trials as the verification gate.
diff --git a/docs/architecture.md b/docs/architecture.md
index 9e2aa70..ed8d171 100644
--- a/docs/architecture.md
+++ b/docs/architecture.md
@@ -2,20 +2,22 @@
 
 Process Engine is a Turnstone-native agent-skill framework: a persona, six skills,
 seven references, and six session templates that drive a gated authoring pipeline.
-Turnstone's native governance mechanisms (prompt policy, advisory judge) enforce
-the gates mechanically — the model generates packages; Turnstone owns the guardrails.
+Turnstone provides the native governance surfaces the engine uses for persistent
+context and advisory evidence; operator approval remains the final authority at
+the defined gates. The model generates packages; Turnstone supplies the native
+mechanisms around the workflow.
 
 For the platform-agnostic prompt+code successor, see
 [Method Factory](https://github.com/RedEyeNinja-BKK/Method-Factory).
 
 Turnstone is the engine's platform. The engine is built for Turnstone,
-deployed on Turnstone, and enforced by Turnstone's native governance.
+deployed on Turnstone, and uses Turnstone's native governance surfaces.
 
 ## Components
 
 ```
 Process Engine (project)
-├── persona: process-engine        # identity, standards, working style
+├── persona: process-engine-generator   # canonical generator persona (repo persona.md; deploy under a distinct generator identity so it does not collide with any development/maintenance persona)
 ├── skills/ (6)
 │   ├── process-engine-core        # entry + routing
 │   ├── process-engine-pattern-author
@@ -55,8 +57,7 @@ flowchart TB
     M --> G
 ```
 
-**Pipeline gates** (v1.9.5): every stage has an operator gate. Turnstone's native
-prompt policy and advisory judge enforce these mechanically.
+**Pipeline gates** (v1.9.6): the meaningful operator gates are the Summary Gate (confirm material/intent/objective before generation), Review (operator accepts the reviewed package), Trial (operator accepts trial evidence/readiness), and Ship (operator authorizes deployment). Not every conversational stage is an approval gate — the pipeline stays conversational between the defined gates. Turnstone's native prompt policy provides durable contextual guidance and the advisory judge provides review/trial evidence; neither silently replaces operator approval.
 
 - **Orient** — declare scope and ask what to build.
 - **Collect** — the engine invites material (links, text, files, docs) and
diff --git a/docs/governance-usage.md b/docs/governance-usage.md
index 1379de2..7eec80c 100644
--- a/docs/governance-usage.md
+++ b/docs/governance-usage.md
@@ -1,20 +1,19 @@
 # Governance Usage — Process Engine × Turnstone's Governance Surface
 
-> **Status:** DEPLOYED (2026-08-01). Live: prompt policy `process-engine-context`
-> (content-only, priority 1) + heuristic rule `process-engine-authoring`
-> (advisory, review, low risk). Operator is the only gate — every governance
-> artifact is advisory, never blocking, always reversible.
+> **Status:** Durable product contract (architecture) + dated deployment
+> snapshot below. Operator is the only gate — every governance artifact is
+> advisory, never blocking, always reversible.
 >
 > **Context:** Process Engine runs on Turnstone, using Turnstone's native
-> governance surface to enforce the engine's core philosophy mechanically:
-> collect before creating, operator-gated shipping, nothing ships untried,
+> governance surface for persistent context and advisory evidence: collect
+> before creating, operator-gated shipping, nothing ships untried,
 > heads-up not police.
 
 ---
 
 ## 1. Why Turnstone
 
-Turnstone's Governance surface is the engine's enforcement layer — a
+Turnstone's Governance surface is the engine's native governance layer — a
 first-class, self-hosted mechanism over how agents operate:
 
 | Harness | Governance surface |
@@ -25,9 +24,9 @@ first-class, self-hosted mechanism over how agents operate:
 | Claude | Commercial harness — no governance layer |
 
 Process Engine's "operator is the gate" philosophy maps directly to Turnstone's
-approval surfaces. The governance layer is not an add-on — it's the engine's
-native enforcement mechanism. The model generates packages; Turnstone owns
-the guardrails.
+approval surfaces. The governance layer is a native mechanism, not an add-on —
+the model generates packages; Turnstone supplies the native governance
+surfaces; the operator is the final authority at the defined gates.
 
 ---
 
@@ -46,13 +45,13 @@ Tool policies remain off the table. The operator is the only gate.
 
 ---
 
-## 3. Deployed governance artifacts
+## 3. Durable product contract
 
 ### Prompt policy: `process-engine-context`
 
-A content-only, priority-1 prompt policy stating the engine's durable
-operating stance. Every session on the Process Engine project starts
-aligned even before the core skill loads:
+A content-only prompt policy stating the engine's durable operating stance.
+Every session on the Process Engine project starts aligned even before the
+core skill loads:
 
 ```
 Process Engine context (applies to sessions on the Process Engine project):
@@ -67,8 +66,6 @@ Process Engine context (applies to sessions on the Process Engine project):
 - Heads-up, not police: surface useful considerations, never obstruct.
 ```
 
-**Policy ID:** `f0991b6598ad4623a410b45cc7aabf28`
-
 ### Advisory judge rule: `process-engine-authoring`
 
 A heuristic rule on the skills tool family (tool=skills, arg=process-engine,
@@ -76,21 +73,6 @@ risk=low, recommendation=review, advisory). Produces per-call intent
 verdicts as machine-consistent evidence that each gate action matched
 the operator's intent. Always advisory — the operator decides.
 
-**Rule ID:** `64b985f99743463a9ddf1a78ded01466`
-
----
-
-## 4. Judge surface — verified
-
-The judge admin API is live and verified:
-
-- **14** settings (enabled, model, smart_approvals, confidence_threshold, output_guard)
-- **36** heuristic rules
-- **19** output-guard patterns
-
-Judge model: `deepseek-deepseek-v4-flash`. Smart approvals at confidence
-threshold 0.95. Output guard active.
-
 ### How Judge serves Process Engine
 
 1. **Trial grading** — verdicts provide machine-consistent evidence per case.
@@ -101,6 +83,23 @@ threshold 0.95. Output guard active.
 
 All verdicts are advisory. The operator remains the only gate.
 
+## 4. Deployment snapshot — 2026-08-01 (dated, mutable)
+
+The following are **mutable operational facts observed on one Turnstone
+deployment on 2026-08-01**. They are not timeless product architecture;
+re-verify live before relying on them. No local object IDs are product
+authority.
+
+- Live prompt policy `process-engine-context` (content-only, priority 1) +
+  heuristic rule `process-engine-authoring` (advisory, review, low risk).
+- Policy ID: `f0991b6598ad4623a410b45cc7aabf28`; rule ID:
+  `64b985f99743463a9ddf1a78ded01466`.
+- Judge admin API live and verified: **14** settings
+  (enabled, model, smart_approvals, confidence_threshold, output_guard),
+  **36** heuristic rules, **19** output-guard patterns.
+- Judge model: `deepseek-deepseek-v4-flash`. Smart approvals at confidence
+  threshold 0.95. Output guard active.
+
 ---
 
 ## 5. Design principles (current)
diff --git a/docs/standards.md b/docs/standards.md
index e971b94..55607c3 100644
--- a/docs/standards.md
+++ b/docs/standards.md
@@ -18,8 +18,10 @@ The actual operator gates are:
 Nothing ships without operator approval, and trials with evidence precede ship.
 A REVISE result returns through diagnose, rewrite, audit, and re-review.
 
-Turnstone's native prompt policy and advisory judge enforce these gates
-mechanically. The model generates packages; Turnstone owns the guardrails.
+Turnstone's native prompt policy provides durable contextual guidance and the
+advisory judge provides review/trial evidence; neither silently replaces
+operator approval. The model generates packages; the operator is the final
+gate at summary, review, trial, and ship.
 
 ## Evidence
 
@@ -30,9 +32,9 @@ to scope limits, review, trial evidence, and operator approval.
 ## Turnstone-native
 
 Process Engine is **Turnstone only**. It is built for Turnstone, deployed on
-Turnstone, and enforced by Turnstone's native governance. Nothing in the
+Turnstone, and uses Turnstone's native governance surfaces. Nothing in the
 engine — and nothing the engine generates — targets any other harness.
 Every generated package includes Turnstone governance objects — prompt
-policy and advisory judge rules — that enforce the engine's operating stance.
-The governance layer is the enforcement mechanism, not an add-on. See
-[Governance usage](governance-usage.md).
+policy and advisory judge rules — that provide persistent context and
+advisory evidence. The governance layer is a native mechanism, not an
+add-on. See [Governance usage](governance-usage.md).
diff --git a/process-engine.toml b/process-engine.toml
index f8e59fd..5f84fe1 100644
--- a/process-engine.toml
+++ b/process-engine.toml
@@ -1,12 +1,12 @@
 # Process Engine — repository release metadata
-# v1.9.5 prompt-only — stripped back to the engine's original intent.
+# v1.9.6 prompt-only — stripped back to the engine's original intent.
 # Turnstone-native. All logic in prompts; Turnstone governance enforces gates.
 # GitHub is canonical; this file is lightweight release/version/count metadata
 # for the structural validator, not a product manifest or runtime contract.
 # For the prompt+code successor, see github.com/RedEyeNinja-BKK/Method-Factory.
 
-version = "1.9.5"
-engine = "process-engine 1.9.5"
+version = "1.9.6"
+engine = "process-engine 1.9.6"
 compatibility = "Turnstone 1.8.x"
 author = "RedEyeNinja-BKK"
 
diff --git a/skills/process-engine-core/SKILL.md b/skills/process-engine-core/SKILL.md
index aae586a..2cf4ded 100644
--- a/skills/process-engine-core/SKILL.md
+++ b/skills/process-engine-core/SKILL.md
@@ -4,8 +4,8 @@ description: Engine identity, pipeline, routing, and standards checklist. Load F
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 This is the engine's entry point. It declares what the Process Engine is (a
diff --git a/skills/process-engine-pattern-author/SKILL.md b/skills/process-engine-pattern-author/SKILL.md
index 4e8ab8a..8143832 100644
--- a/skills/process-engine-pattern-author/SKILL.md
+++ b/skills/process-engine-pattern-author/SKILL.md
@@ -4,8 +4,8 @@ description: Generate the project/persona/skills package for an intent — to th
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 Turns an operator intent into the engine's final product shape — a
diff --git a/skills/process-engine-review/SKILL.md b/skills/process-engine-review/SKILL.md
index a3541ca..3eff5e2 100644
--- a/skills/process-engine-review/SKILL.md
+++ b/skills/process-engine-review/SKILL.md
@@ -4,8 +4,8 @@ description: The engine's review gate — spec compliance, standards, scope, evi
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 Reviews a draft against the engine's standards. Produces a verdict (PASS /
diff --git a/skills/process-engine-ship/SKILL.md b/skills/process-engine-ship/SKILL.md
index 9172ca1..7779c0a 100644
--- a/skills/process-engine-ship/SKILL.md
+++ b/skills/process-engine-ship/SKILL.md
@@ -4,8 +4,8 @@ description: Deploy an approved, trialed package via Turnstone's native mechanis
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 Ships only what has PASSED review AND trial — the generated package (project /
diff --git a/skills/process-engine-triage/SKILL.md b/skills/process-engine-triage/SKILL.md
index 05a80de..4637cf5 100644
--- a/skills/process-engine-triage/SKILL.md
+++ b/skills/process-engine-triage/SKILL.md
@@ -4,8 +4,8 @@ description: Feedback sensor — collect public feedback (GitHub issues/discussi
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 The engine improves its own process. Feedback is the sensor: issues and
diff --git a/skills/process-engine-trial/SKILL.md b/skills/process-engine-trial/SKILL.md
index 4f4d5f7..6084dc6 100644
--- a/skills/process-engine-trial/SKILL.md
+++ b/skills/process-engine-trial/SKILL.md
@@ -4,8 +4,8 @@ description: Trial harness — scripted cases and trigger sets that prove an art
 compatibility: Turnstone 1.8.x
 metadata:
   author: RedEyeNinja-BKK
-  version: "1.9.5"
-  engine: process-engine 1.9.5
+  version: "1.9.6"
+  engine: process-engine 1.9.6
 ---
 ## Overview
 Trials are the engine's proof. Every package is trialed before shipping.