erinepshovel-code · Oct 14, 2026 · Oct 14, 2026
diff --git a/.agents/skills/README.md b/.agents/skills/README.md
@@ -5,7 +5,7 @@ This directory contains repo-local consumed copies of canonical skills from
 
 Installed subset:
 
-- `stack-update` — source `The-Interdependency/skill-lib@a7b95f891d2e88c62e61524dc52a2e1577c9983d`, path `stack-update/SKILL.md`, source blob `a1e914893fa047e28d039050395937a1cf6e0138`.
+- `stack-update` — source `The-Interdependency/skill-lib@321e721fb660e0fea5bb8c37d943611963039a57`, path `stack-update/SKILL.md`, source blob `9ec21b07d01a099021778a0c8774b810c2619d63`.
 
 Canonical doctrine remains in `skill-lib`; vendoring does not transfer authority.
 Structural stack changes must follow `.agents/skills/stack-update/SKILL.md` and pass

diff --git a/.agents/skills/stack-update/PROVENANCE.json b/.agents/skills/stack-update/PROVENANCE.json
@@ -3,9 +3,9 @@
   "version": "1.0.0",
   "skill": "stack-update",
   "source_repository": "The-Interdependency/skill-lib",
-  "source_commit": "a7b95f891d2e88c62e61524dc52a2e1577c9983d",
+  "source_commit": "321e721fb660e0fea5bb8c37d943611963039a57",
   "source_path": "stack-update/SKILL.md",
-  "source_blob_sha": "a1e914893fa047e28d039050395937a1cf6e0138",
+  "source_blob_sha": "9ec21b07d01a099021778a0c8774b810c2619d63",
   "authority_transfer": false,
   "relation": "repo-local consumed copy; canonical doctrine remains in source repository"
 }
diff --git a/.agents/skills/stack-update/SKILL.md b/.agents/skills/stack-update/SKILL.md
@@ -5,9 +5,8 @@ description: Fail-closed update protocol for The-Interdependency/stack. Load thi
 
 # stack-update — change the stack as one coherent transaction
 
-Use this procedural skill for structural changes to `The-Interdependency/stack`.
-It specializes `interdependent-work-graph`; it does not replace that skill or
-`project-incubation-graduation`.
+Load the companion skills specified below; this protocol specializes their
+contracts for structural Stack changes.
 
 ## Core contract
 

diff --git a/README.md b/README.md
@@ -257,7 +257,7 @@ Pinned and vendored components keep their own licenses:
 The whole `skill-lib/` snapshot, including
 [`skill-lib/LICENSES/Apache-2.0.txt`](skill-lib/LICENSES/Apache-2.0.txt) and
 `skill-lib/ATTRIBUTION.md`, is an exact copy of The-Interdependency/skill-lib
-`023ae12`. The doctrine and MSDMD generator pin is that same commit (see
+`321e721`. The doctrine and MSDMD generator pin is that same commit (see
 `stack-manifest.json` and `backend/fresh-making-provenance.json`). The root
 [`LICENSES/Apache-2.0.txt`](LICENSES/Apache-2.0.txt) is the same text.
 

diff --git a/STACK_MANIFEST.md b/STACK_MANIFEST.md
@@ -17,7 +17,7 @@ Provenance and authority-boundary record for `The-Interdependency/stack`.
 - English four-view complete-corpus audit UTC: `2026-09-22`, exact view source `1f9a35eb355296fc88d09784c7a3e2e95511ca31`; all 164,864 words, native and independent agreement.
 - Stack-manifest schema: `the-interdependency.stack-manifest` version `1.1.0`
 - Work-graph digest (SHA-256 over canonical `repositories` + `research_participants` + `boundaries` JSON):
-  `b7cd4a9b8f1908166024b4ebf8e3fa040f97d4f4cbbff190702a9553a6665e45`
+  `f4acf6c40732f486388ed43df896d6e21590f1bf567310a0f45814c56d506268`
 - Weave workspace creation UTC: `2026-09-28`; new specification-first research starts from the intended multi-arity interleaving mechanism without URPCS inheritance
 - Pi radix workspace creation UTC: `2026-10-10`; bounded cross-base computation starts from Stack `ccf707f933a9611d7b7fe13a0417378d8b453a14`
 - Machine-readable copy: [`stack-manifest.json`](stack-manifest.json)
@@ -45,7 +45,7 @@ meaning used by that repository.
 
 | Repository | Exact source commit | Source branch | Authority | Stack relation |
 |---|---|---|---|---|
-| `The-Interdependency/skill-lib` | `023ae12bb19ced1a4f2494ed6507fc31285a5b8e` | main | organization-wide build and evidence doctrine | operational snapshot at `skill-lib/` |
+| `The-Interdependency/skill-lib` | `321e721fb660e0fea5bb8c37d943611963039a57` | main | organization-wide build and evidence doctrine | operational snapshot at `skill-lib/` |
 | `The-Interdependency/metapat` | `34d954aa1e2092e615b03a180500f6b6977f501e` | main | semantic authority (Meta Energy Theory) | canon view `libs/metapat/`; research `research/metapat/` |
 | `The-Interdependency/ucns` | `828c0b8bbcfc267efb5701da714191c1f73a81ff` | main | geometry and mathematical representation | canon view `libs/ucns/`; research `research/ucns/` |
 | `The-Interdependency/edcm` | `7951ca32ba0f2494dc68ff9b7f6a80151918a56d` | main | measurement and evaluation of text-domain outputs | canon view `libs/edcm/`; measurement research `research/edcm/`; English Gonol construction is separate at `research/english-gonol/` |

diff --git a/ahbg/integration/work-graph.json b/ahbg/integration/work-graph.json
@@ -60,12 +60,12 @@
     {
       "repository": "The-Interdependency/skill-lib",
       "reviewed_commit": "516933d98f9de376f4e498f44059043dc0d96470",
-      "consumed_commit": "023ae12bb19ced1a4f2494ed6507fc31285a5b8e",
+      "consumed_commit": "321e721fb660e0fea5bb8c37d943611963039a57",
       "relevant_advance_commit": "284ebefad76a6589788ce8d16d47ad75bbfaa333",
       "authority": "build, evidence, metadata, and cross-repository coordination doctrine",
       "relation": "evidence-doctrine-peer",
       "integration": "drift-visible",
-      "decision": "Native-first MSDMD readers are reviewed but are not copied into AHBG. Stack's root skill-lib/ snapshot now consumes 023ae12, which carries them; whether AHBG adopts them is a separate AHBG integration decision (hmmm) that this snapshot refresh does not make."
+      "decision": "Native-first MSDMD readers are reviewed but are not copied into AHBG. Stack's root skill-lib/ snapshot now consumes 321e721, which carries them; whether AHBG adopts them is a separate AHBG integration decision (hmmm) that this snapshot refresh does not make."
     },
     {
       "repository": "The-Interdependency/epac",

diff --git a/backend/README.md b/backend/README.md
@@ -218,8 +218,11 @@ Skipped PostgreSQL tests are deployment gates, not passes.
 
 `fresh-making-provenance.json` binds this runtime to exact `skill-lib` fresh-making
 doctrine. The root `skill-lib/` snapshot and the doctrine pin are the same skill-lib
-commit; the comparison records that `fresh-making/SKILL.md` is unchanged from the
-previous doctrine pin.
+commit. The `023ae12..321e721` operational refresh changes
+`fresh-making/SKILL.md` while preserving the exact `msdmd/` generator sources.
+The cumulative comparison retains earlier generator-change evidence.
+`deploy/VM_CONTACT.md` keeps its separately pinned `023ae12` bootstrap and acceptance
+evidence; a source refresh does not deploy or requalify that VM runtime.
 
 ## hmmm
 

diff --git a/backend/deploy/VM_SETUP.md b/backend/deploy/VM_SETUP.md
@@ -120,7 +120,7 @@ operator action rather than writing an artifact.
 
 `stack-orchestrator-worker.service` keeps `MemoryDenyWriteExecute=true`. V8's
 JIT cannot run under it: Node dies with a V8 fatal error (SIGTRAP). The pinned
-skill-lib snapshot (`023ae12`; `--jitless` since The-Interdependency/skill-lib#120)
+skill-lib snapshot (`321e721`; `--jitless` since The-Interdependency/skill-lib#120)
 runs Node with `--jitless`, so the sandboxed probe and TypeScript extraction
 work under MDWE. `node` must also be on the unit's default PATH.
 

diff --git a/backend/fresh-making-provenance.json b/backend/fresh-making-provenance.json
@@ -3,19 +3,19 @@
   "version": "1.0.0",
   "doctrine": {
     "repository": "The-Interdependency/skill-lib",
-    "commit": "023ae12bb19ced1a4f2494ed6507fc31285a5b8e",
+    "commit": "321e721fb660e0fea5bb8c37d943611963039a57",
     "path": "fresh-making/SKILL.md",
     "relation": "exact governing fresh-making doctrine for the stack backend"
   },
   "local_generator_snapshot": {
     "repository": "The-Interdependency/skill-lib",
-    "commit": "023ae12bb19ced1a4f2494ed6507fc31285a5b8e",
+    "commit": "321e721fb660e0fea5bb8c37d943611963039a57",
     "path": "skill-lib/",
     "relation": "stack-local pinned snapshot used by the default MSDMD generator root"
   },
   "evidence": {
-    "skill_lib_compare": "1b1a9473fcbaab16965af9907e4aafd12853a704..023ae12bb19ced1a4f2494ed6507fc31285a5b8e",
-    "fresh_making_doctrine_changed_in_range": false,
+    "skill_lib_compare": "1b1a9473fcbaab16965af9907e4aafd12853a704..321e721fb660e0fea5bb8c37d943611963039a57",
+    "fresh_making_doctrine_changed_in_range": true,
     "msdmd_generator_changed_in_range": true,
     "doctrine_and_snapshot_same_commit": true,
     "msdmd_generator_changes": [
@@ -27,6 +27,13 @@
       "skill-lib#124: structured-document and standards facts redact URL userinfo (with #redacted-N ordinals, never a hash of the secret), keep typed YAML mapping association while redacting, withhold Kubernetes Secret data and encoded DSSE payloads, tag out-of-range YAML/TOML integers as json-number, and make invalid_statement_subject an error (affects exit only under --strict)",
       "skill-lib#127 (023ae12): vm-mcp only (install hardening and the server's fd 0 owner/mode check); msdmd/ is unchanged from 24029ca, so the collector sources and generator identity are the same as at 24029ca",
       "collector sources changed in 1b1a947..24029ca (#121, #124), so the generator identity and every MSDMD derivation re-key relative to 1b1a947"
+    ],
+    "snapshot_refresh_compare": "023ae12bb19ced1a4f2494ed6507fc31285a5b8e..321e721fb660e0fea5bb8c37d943611963039a57",
+    "snapshot_refresh_msdmd_generator_changed": false,
+    "snapshot_refresh_notes": [
+      "skill-lib#128: refresh the exact operational snapshot and narrative doctrine; fresh-making/SKILL.md changed, while msdmd/ source bytes are unchanged from 023ae12.",
+      "The cumulative comparison and prior generator-change evidence above are retained; this refresh itself does not change the MSDMD generator source identity.",
+      "backend/deploy/VM_CONTACT.md retains its separately pinned 023ae12 vm-mcp bootstrap and acceptance evidence. This source refresh does not deploy, requalify, or change the VM runtime identity."
     ]
   },
   "boundaries": {

diff --git a/skill-lib/ATTRIBUTION.md b/skill-lib/ATTRIBUTION.md
@@ -25,8 +25,9 @@ Third-party skills vendored into this library, with provenance and license.
 Local modifications to imported files: frontmatter trigger phrasing
 normalized to skill-lib convention (`Use this when`), Cowork-only
 frontmatter keys removed, and a Workflow/Anti-patterns/Provenance/hmmm
-bookend appended per `skill-build` compliance. Upstream bodies are
-otherwise unmodified. `data-visualization/SKILL.md` also has a clearly marked
+bookend appended per `skill-build` compliance. The narrative compression
+migration removes redundant opening summaries and consolidates the explore-data
+and validate-data workflows, retaining their operative obligations. `data-visualization/SKILL.md` also has a clearly marked
 information-design extension appended locally (skill-lib `868de86`). These
 local additions to `data-visualization/SKILL.md` are contributed under
 Apache-2.0, so the whole file stays under a single license. Other local files in

diff --git a/skill-lib/ORG_DISTRIBUTION.md b/skill-lib/ORG_DISTRIBUTION.md
@@ -17,7 +17,8 @@ Propagation PRs should cite this repository and the source commit SHA.
 
 Selected propagation includes the transitive `depends_on` prerequisites from
 `skills.json`. For example, `python tools/propagate_skills.py ../a0 --skills
-agent-instantiation --apply` installs both the requested skill and `wdll`.
+agent-instantiation --apply` installs the requested skill, `wdll`, and the
+`a0p-instancing` owner reached by its architecture-routing link.
 Verify the result with `python tools/check_consumer_drift.py ../a0
 --require-vendored`: the checker requires that same dependency closure, including
 its canonical files and referenced doctrine. Missing prerequisites and invalid

diff --git a/skill-lib/README.md b/skill-lib/README.md
@@ -26,6 +26,8 @@ Agents consuming this lib should start at
 [`skills.json`](skills.json). Root LLM-facing instructions are generated
 into [`llms.txt`](llms.txt) from self-declared `LLMS` blocks.
 
+Narrative maintenance: see the [reviewed compression inventory and preservation checks](docs/narrative-migration/README.md).
+
 ## What's inside
 
 | Skill | Purpose |

diff --git a/skill-lib/a0p-instancing/SKILL.md b/skill-lib/a0p-instancing/SKILL.md
@@ -1,33 +1,10 @@
 ---
 name: a0p-instancing
-description: Methodology for instancing agents in a0-betatest (the a0p research instrument), whose model diverges from canonical a0. Load this when adding or changing an AgentInstance / CharacterSheet CRUD path, a per-instance native ZFAE weight bank or its training/distillation loop, a ZFAE inference mode, a sentinel evaluation or pending-override gate, a per-agent safetensors checkpoint, or volatile sub-context memory — anywhere under a0-betatest `backend/`. Use it before writing code that creates, addresses, trains, runs, governs, or persists an a0p agent, so the code follows a0p's per-user CRUD + native-ZFAE + sentinel model instead of a0's spawn/fork/merge model. For canonical a0 and its mirror a0ucns, use `agent-instantiation` instead — a0p does NOT have `sub_agent_spawn`, a spawn executor, or `InstanceMerge`.
+description: Agent instancing in a0-betatest backend (a0p). Load this when creating, addressing, training, running, governing, persisting, or changing AgentInstance/CharacterSheet CRUD, native ZFAE weight banks, teacher distillation, inference modes, sentinel evaluation, pending overrides, per-agent safetensors checkpoints, Mongo metadata, FIQ events, or volatile sub-context memory. Do not load for ordinary LLM-prompt work that creates no instance. For a0, a0ucns, or a0-betatest/_legacy_a0 use agent-instantiation/SKILL.md.
 ---
 
 # a0p-instancing — How a0-betatest brings agents into being
 
-a0-betatest (the **a0p** research instrument) replaced a0's single-persistent
-+ fork model with a **multi-agent, per-user model**: an agent is a persistent
-CRUD entity (`AgentInstance`) bound to an editable `CharacterSheet`, and each
-instance **owns its own trained native ZFAE weight bank**. The LLM is optional
-energy; once an instance's native core is trained, it can run with no LLM at
-all. This skill captures that model so a coding agent extends it instead of
-reaching for a0's spawn/fork/merge machinery, which **does not exist here**.
-
-This skill is the **peer** of `agent-instantiation` (which documents a0 /
-a0ucns). They describe genuinely different architectures; do not mix them.
-
-## Load this when
-
-- Adding/editing an `AgentInstance` or `CharacterSheet` **CRUD** path.
-- Touching the **native ZFAE weight bank**, its **teacher-distillation
-  training**, or the native **inference** engine / modes.
-- Adding/changing a **sentinel** evaluation or a **pending-override** gate.
-- Persisting per-agent state (**safetensors checkpoints**, Mongo metadata,
-  **FIQ** audit events) or volatile **sub-context** memory.
-
-Do not load this for canonical a0 / a0ucns work (use `agent-instantiation`),
-or for ordinary LLM-prompt work that creates no instance.
-
 ## Scope and source boundary
 
 The **canonical source is `a0-betatest/backend/`** (the a0p package per its
@@ -54,7 +31,8 @@ reproduce them from memory. Where a mechanism is unclear or unwired, write
   **fresh per agent** at creation (`backend/agents/store.py`).
 
 There is **one weight bank per agent**, not a shared singleton engine. Never
-share or fork a bank between instances.
+share or fork a bank between instances. A trained native core can run without
+an LLM; the inference mode governs optional LLM energy.
 
 ## Instancing sequence (dependency order — follow it top to bottom)
 
@@ -82,7 +60,7 @@ share or fork a bank between instances.
    or both produce/judge the reply. `zfae_native` refuses LLM fallback.
 6. **Evaluate sentinels and honor the override halt** (`sentinel_eval.py`,
    `sentinels.py`, `overrides.py`). Resolve per-agent sentinel modes/weights,
-   compute the `Verdict13`; on a blocking flag, create a `PendingOverride`
+   compute the `Verdict13` from 13 sentinels (S1–S13); on a blocking flag, create a `PendingOverride`
    and **halt** until approved. Do not bypass the halt.
 7. **Checkpoint to safetensors, not a DB blob.** Persist the bank to
    `storage/agents/{id}/zfae_core.safetensors` (+ `…meta.json`: digest,

diff --git a/skill-lib/action-calibration/SKILL.md b/skill-lib/action-calibration/SKILL.md
@@ -5,10 +5,6 @@ description: Action sizing and escalation doctrine. Load this when choosing betw
 
 # action-calibration — spend only enough to change the decision
 
-`action-calibration` selects the right-sized action before implementation, research,
-formalization, or publication consumes the available time, attention, money, compute,
-or coordination capacity.
-
 The governing distinctions are:
 
 ```text

diff --git a/skill-lib/adaptive-elicitation/SKILL.md b/skill-lib/adaptive-elicitation/SKILL.md
@@ -1,33 +1,9 @@
 ---
 name: adaptive-elicitation
-description: Adaptive, context-first interviewing for recovering missing information without making the user answer a questionnaire. Load this when a task requires eliciting facts, preferences, requirements, incident details, form fields, debugging context, or other missing inputs across turns; when prior conversation, files, connected sources, or authoritative records may already answer part of the question set; or when the user asks to be interviewed one question at a time. Do not load for a one-shot clarification that can be asked and resolved in a single ordinary prompt.
+description: Adaptive, context-first interviewing for recovering missing information without making the user answer a questionnaire. Load this when a task requires eliciting facts, preferences, requirements, incident details, form fields, debugging context, or other missing inputs across turns; when prior conversation, files, connected sources, or authoritative records may already answer part of the question set; or when the user asks to be interviewed one question at a time. Do not load for a one-shot clarification that can be asked and resolved in a single ordinary prompt. This includes missing facts in forms, filings, reports, applications, audits, or debugging; questionnaire replacement; debriefing; and sensitive, stressful, or non-linear recall.
 ---
 
-# adaptive-elicitation — ask only the next useful question
 
-A procedural skill for semi-structured, context-aware interviewing. Its job is
-to recover the missing information needed for the user's actual task with the
-least avoidable repetition, suggestion, ambiguity, and response burden.
-
-The interview state is a living ledger. Every answer can satisfy more than the
-question that elicited it. Consume volunteered facts immediately and recompute
-what is still missing before asking again.
-
-## When to load
-
-Load this skill when:
-
-- a form, filing, report, specification, application, incident record, audit,
-  debugging task, or other deliverable has missing user-supplied facts;
-- the agent would otherwise present a multi-question questionnaire;
-- prior conversation, memory, files, connected apps, repository state, or
-  authoritative sources may already contain answers;
-- the user asks to be interviewed, questioned, debriefed, or walked through
-  missing information one question at a time;
-- the subject is sensitive, stressful, cognitively expensive, or likely to
-  produce non-linear recall.
-
-Do not load merely because an ordinary request needs one obvious clarification.
 
 ## Source and evidence boundary