Repair gonol authority and canonicalize ai.sh

.github/workflows/ci.yml

@@ -21,13 +21,8 @@ jobs:
           node-version: "24.15.0"
       - name: Unit tests
         run: python -m unittest discover -s tests
-      - name: Exact EDCM cross-source construction witness
-        run: |
-          EDCM_CHECKOUT=$(mktemp -d)
-          git -C "$EDCM_CHECKOUT" init
-          git -C "$EDCM_CHECKOUT" fetch --depth 1 https://github.com/The-Interdependency/edcm.git ddc89a97ebbcf0a5863dad6e633b01b520e9bccf
-          git -C "$EDCM_CHECKOUT" checkout --detach FETCH_HEAD
-          python tools/check_edcm_boundary.py "$EDCM_CHECKOUT"
+      - name: Gonol authority gate
+        run: bash tools/check_gonol_authority.sh
       - name: Skill library drift
         run: python tools/check_skill_lib_drift.py --warnings-fail
       - name: Skill compliance
@@ -41,4 +36,4 @@ jobs:
       - name: RepoLOTO audit
         run: python tests/test_repo_loto.py --audit
       - name: RepoLOTO checks
-        run: python tests/test_repo_loto.py
+        run: python tests/test_repo_loto.py

AGENTS.md

@@ -51,6 +51,8 @@ Read [`RESOURCE_RUN_INVARIANT.md`](RESOURCE_RUN_INVARIANT.md) before any compute
 <skill-name>/SKILL.md          # required: the skill itself
 <skill-name>/<helpers>...      # optional: parsers, executors, examples
 llms/                          # stdlib module for python -m llms.build
+tools/ai.sh                    # canonical Termux -> a0 SSH/tmux coding-agent launcher
+tools/install_ai.sh            # installs ai.sh into caller PATH (Termux $PREFIX/bin first)
 ```
 
 Every skill is a directory at the repo root containing at least a
@@ -143,11 +145,13 @@ propagate from here.
 - If you are translating raw, recursive, fragmentary, coined, or private-language
   thought for strangers or a specific audience, load `thought-lens/SKILL.md`;
   freeze the claim kernel before changing vocabulary and back-check the result.
-- If you are constructing, reviewing, replaying, or continuing UCNS gonols,
-  including lexical floors, morphology, definitions, punctuation functions,
-  closure, atomic promotion, or recursive relations, load `gonol-build/SKILL.md`.
-  Resolve current UCNS geometry and EDCM text/scale authority first; never impose a universal adjacent-scale ladder or restore historical
-  `gonal-morphology` doctrine as current canon.
+- If you are constructing, reviewing, replaying, or continuing language-gonol
+  research, including lexical floors, morphology, definitions, punctuation
+  functions, closure, atomic promotion, or recursive relations, load
+  `gonol-build/SKILL.md`. Resolve current UCNS gonol-object/constructor/geometry
+  authority and the exact owning Stack research workspace first. EDCM is
+  measurement/evaluation only; never impose a universal adjacent-scale ladder
+  or restore historical `gonal-morphology` doctrine as current canon.
 - If you are comparing UCNS options, deciding whether evidence authorizes a
   winner, or issuing a scoped selection receipt, load
   `ucns-option-selection/SKILL.md`. Hard eligibility and evidence gates cannot
@@ -189,3 +193,5 @@ propagate from here.
 - `llms.txt` — generated LLM-facing root instructions.
 - Each `<skill>/SKILL.md` — the authoritative skill spec.
 - `llms/build.py` — reference runner for `llms-build`.
+- `tools/ai.sh` — canonical Termux-side launcher for the remote `a0` tmux coding-agent session.
+- `tools/install_ai.sh` — installs `ai.sh` into caller PATH, preferring Termux `$PREFIX/bin`.

CLAUDE.md

@@ -11,6 +11,7 @@ AI-assistant guidance for `The-Interdependency/skill-lib`.
 - Entry points: `README.md`, `AGENTS.md`, `skills.json`, `ORG_DISTRIBUTION.md`, `llms.txt`, each `<skill>/SKILL.md`.
 - CI workflows: `.github/workflows/hygiene.yml` guards against tracked Python bytecode, `.github/workflows/ci.yml` runs the editorial/helper verification stack, and `.github/workflows/consumer-drift.yml` is a scheduled/dispatch detector that runs `tools/check_consumer_drift.py` against each consumer repo (the consumer repos are public, so it uses the default `GITHUB_TOKEN`).
 - Validation here is editorial plus pure-stdlib helper scripts in `tools/`, `ratios/`, `llms/`, and the RepoLOTO check module.
+- `tools/ai.sh` is the canonical Termux-side SSH/tmux launcher for the `a0` VM; `tools/install_ai.sh` installs `ai.sh` into the caller PATH (Termux `$PREFIX/bin`, otherwise `~/.local/bin`).
 - The `llms/` package exists only to expose the stdlib `python -m llms.build` runner for `llms-build`.
 
 ## Layout
@@ -53,7 +54,7 @@ llms/                  # python -m llms.build reference runner
 | `agent-instantiation/` | procedural | — | a0/a0ucns agent lifecycle methodology. Spawn sub-agents via the `sub_agent_spawn` tool → spawn executor; fork/merge `PCNAEngine` instances via `InstanceMerge` (fork/absorb/converge); compose identities per the canonical `username(a0(energy)auditor)` grammar; honor spawn caps + write-route gating. Canonical source is `a0`; `a0-betatest` diverged (per-user native-ZFAE) and is out of scope. Repo-specific runtime doctrine (no theorem transfer). |
 | `a0p-instancing/` | procedural | — | Peer for a0-betatest (a0p): agents are per-user CRUD `AgentInstance` + `CharacterSheet`, each owning a trained native ZFAE weight bank (three 157-seed cores); no `sub_agent_spawn`/executor/`InstanceMerge` — only volatile `MemoryCore.spawn_sub/merge_sub`. Sequence: create→distill-train→readiness gate→mode inference→sentinel/pending-override→safetensors checkpoint. Canonical source is `a0-betatest`. |
 | `plain-lens/` | procedural | — | Plain-language, multi-lens companion views of dense canonical text. Build easier on-ramps (domain/audience/role lens selectors, progressive disclosure) that never replace or talk down to the source, keep a static fallback under any dynamic layer, preserve operators/negations/quantifiers, and report an EDCM-style body-vs-footnote tension reading as an illustrative heuristic (not an edcmbone metric runtime). |
-| `gonol-build/` | procedural | — | UCNS geometry / EDCM text construction discipline. Resolve current owning contracts and admissible scale options; preserve closure, atomic participation, occurrence identity, full-source receipts, and protocol-required replay. Refuses superseded omega/phi/psi, bone/flesh, and carrier-LCM language doctrine. |
+| `gonol-build/` | procedural | — | UCNS gonol objects/constructors/geometry + Stack language-construction research discipline. Resolve the owning Stack workspace, preserve closure, atomic participation, occurrence identity, constitutive relations, provenance, and required replay; EDCM is measurement/evaluation only. |
 | `ucns-option-selection/` | procedural | — | Fail-closed scoped UCNS option selection. Freezes candidates, authority, gates, evidence, and policies; requires complete evaluation, replay, purpose-relative comparison, explicit ratification, rollback, and non-transfer. Hard-gate failures cannot be compensated by scores. |
 | `epac-selection-display/` | procedural | — | Evidence-bound EPAC display selection. Resolves an exact provisional source, target, receipt, and available representation; preserves status, nonclaims, sealed-comparison boundaries, and `hmmm`; and keeps WebMCP as a read-only handoff rather than a research-code executor. |
 | `meta/` | procedural | — | Meta Energy Theory axioms. Extract and preserve Energy Theory axioms from resonances among small network architectures, with formula-backed examples and overlap grids; keep Energy Theory distinct from EDCMBONE flesh/bone and FLAR implementation detail. |
@@ -190,6 +191,7 @@ blocks first; do not hand-edit `llms.txt` as independent doctrine.
 
 ```bash
 python -m unittest discover -s tests
+bash tools/check_gonol_authority.sh
 python tools/check_skill_lib_drift.py --warnings-fail
 python tools/check_skill_compliance.py --warnings-fail
 python ratios/ratios_check.py --strict
@@ -211,8 +213,10 @@ There is a small stdlib Python editorial test suite. There is still no `package.
   skills.json semantics, per-skill spec coverage, SKILL.md frontmatter, README
   index coverage, collection-point schema/generator/visualizer coverage, universal parser
   behavior, llms-build behavior, and parser ratio bookends.
+- `tools/ai.sh` is the canonical Termux-side controller for the remote `a0` tmux coding-agent session; `tools/install_ai.sh` installs `ai.sh` into caller PATH, preferring Termux `$PREFIX/bin`.
 - The parsers are reference implementations; the test suite covers core parser
   behavior and library integration, not every consuming-runner contract.
+- `check_gonol_authority.sh` fails closed when active skill-lib surfaces restore EDCM construction ownership.
 - `check_skill_lib_drift.py` checks editorial agreement among skill directories, `skills.json`, `README.md`, `ORG_DISTRIBUTION.md`, `AGENTS.md`, `CLAUDE.md`, and generated `llms.txt`.
 - `check_skill_compliance.py` checks baseline `skill-build` invariants for each `SKILL.md`.
 - `ratios_check.py --strict` verifies opening/closing ratios seals for
@@ -251,7 +255,7 @@ There is a small stdlib Python editorial test suite. There is still no `package.
 9. Apply `char-compress` when compressing repo context: carry flesh, frozen bones, transforms, and hmmm; drop only safely regenerable scaffold.
 10. Treat `char-compress` as a skill-lib-owned compression procedure, but do not claim unearned theorem/status support or edcmbone metric status.
 11. Before promoting a word into canon, a theorem term, ontology primitive, schema field, encoding label, or cross-domain mapping, apply `domain-claims`: establish the domain-qualified sense and resolve collisions before attaching provenance; then apply `canon` to assess authority.
-12. Before constructing, reviewing, replaying, or extending UCNS gonols, apply `gonol-build`: resolve current UCNS geometry and EDCM admissible scale options, preserve closure and atomic participation, require declared occurrence-addressed function plans, and keep incomplete constructors visible as `hmmm`.
+12. Before constructing, reviewing, replaying, or extending language-gonol research, apply `gonol-build`: resolve current UCNS gonol-object/constructor/geometry authority and the exact owning Stack research workspace; EDCM is measurement/evaluation only. Preserve closure and atomic participation, require declared constitutive relations, and keep incomplete geometry visible as `hmmm`.
 13. Before selecting among UCNS options, apply `ucns-option-selection`: freeze the scoped decision boundary, enforce noncompensable eligibility and evidence gates, require explicit ratification, and preserve non-transfer, rollback, negative evidence, and `hmmm`.
 14. Before selecting and displaying an EPAC artifact, apply `epac-selection-display`: pin the provisional source, exact target, receipt, and available renderer; preserve status, nonclaims, sealed comparison, and `hmmm`; and keep WebMCP read-only.
 15. Before mutating `The-Interdependency/stack` structure, apply `stack-update` with `interdependent-work-graph`; update every affected authority/provenance projection, remove superseded claims, recompute the work-graph digest, and require the deterministic stack checker before merge.

ORG_DISTRIBUTION.md

@@ -38,7 +38,7 @@ Propagation PRs should cite this repository and the source commit SHA.
 * `plain-lens/` — plain-language, multi-lens companion views of dense canonical text
 * `thought-lens/` — raw-thought to audience-legible translation with claim-kernel fidelity and back-translation checks
 * `meta/` — consultation router for current METAPAT authority; no frozen doctrine copy
-* `gonol-build/` — UCNS geometry / EDCM text construction, declared scale options, closure, atomic participation, protocol-required replay, and honest continuation boundaries
+* `gonol-build/` — UCNS gonol objects/constructors/geometry + Stack language-construction research, closure, atomic participation, replay, and honest continuation boundaries; EDCM is measurement/evaluation only
 * `ucns-option-selection/` — fail-closed scoped UCNS option comparison, selection, ratification, non-transfer, rollback, and decision receipts
 * `epac-selection-display/` — exact provisional EPAC target and representation selection with receipt-backed display, status preservation, and a read-only WebMCP handoff boundary
 * `the-interdependency/` — org-wide workflow protocol and usage-guidance doctrine for The Interdependency projects

README.md

@@ -45,7 +45,7 @@ into [`llms.txt`](llms.txt) from self-declared `LLMS` blocks.
 | [`plain-lens/`](plain-lens/SKILL.md) | Plain-language, multi-lens companion views of dense canonical text — easier on-ramps that do not replace or talk down to the source. Domain/audience/role lens selectors, progressive-disclosure reading UX, static fallback for dynamic pages, and EDCM-style body-vs-footnote tension readings. Independent of msdmd. |
 | [`thought-lens/`](thought-lens/SKILL.md) | Translation compiler for raw, context-heavy thought. Recovers a claim kernel before rendering audience/surface-specific language, preserves claim force and operators, introduces coined terms only after their ordinary meaning lands, and back-checks for fidelity and remaining context debt. Independent of msdmd. |
 | [`meta/`](meta/SKILL.md) | Consultation router for current METAPAT. Retrieves the authoritative repository when a conceptual relation, distinction, boundary, transformation, scale, or cross-domain correspondence would constrain downstream work; carries no frozen METAPAT doctrine here. Independent of msdmd. |
-| [`gonol-build/`](gonol-build/SKILL.md) | UCNS geometry / EDCM text construction, closure, atomic participation at admissible scales, and scoped replay discipline. Resolves each owning contract; preserves exact source and occurrence identity; imposes no universal scale ladder. Independent of msdmd. |
+| [`gonol-build/`](gonol-build/SKILL.md) | UCNS gonol objects/constructors/geometry + Stack language-construction research discipline. Preserves closure, atomic participation, exact source/occurrence identity, provenance, honest `hmmm`, and scoped replay; EDCM is measurement/evaluation only. Independent of msdmd. |
 | [`ucns-option-selection/`](ucns-option-selection/SKILL.md) | Fail-closed UCNS option-selection rubric. Freezes scope, candidates, authority, gates, policies, and evidence boundaries; requires complete execution, falsification, replay, purpose-relative comparison, non-transfer, rollback, and explicit ratification; refuses compensating scores and universal promotion from local evidence. Independent of msdmd. |
 | [`epac-selection-display/`](epac-selection-display/SKILL.md) | Evidence-bound EPAC target selection and display. Resolves an exact provisional source, target, receipt, and available renderer; preserves standing, `selection_effect`, nonclaims, sealed-comparison boundaries, and `hmmm`; and keeps the WebMCP surface a read-only repository-and-skill handoff rather than an EPAC executor. Independent of msdmd. |
 | [`the-interdependency/`](the-interdependency/SKILL.md) | Workflow and protocol for code building, researching, GitHub maintenance and updates, EDCMBONE transcript assembly for analysis, and anything that touches The Interdependency organization or The Interdependent Way projects. Enforces structure preservation (neurodivergence-compatible), mandatory usage guidance in all artifacts, framework-aligned EDCMBONE analysis, and org-standard GitHub hygiene. Independent of msdmd. |
@@ -133,6 +133,24 @@ python tools/build_codex_plugin_skills.py --apply
 python tools/build_codex_plugin_skills.py --check
 ```
 
+## VM coding-agent launcher
+
+The canonical `ai.sh` launcher lives at `tools/ai.sh`. It runs on Termux,
+reaches the `a0` VM through the SSH host alias `a0`, and manages the remote
+`a0` tmux session. Install it into the caller PATH without copying the
+implementation:
+
+```bash
+bash tools/install_ai.sh
+```
+
+On Termux the installer symlinks `ai.sh` into `$PREFIX/bin`; elsewhere it falls
+back to `~/.local/bin/ai.sh`. `ai.sh status` reports remote pane state,
+`ai.sh restart deepcode` repairs a missing/dead DeepCode window, and pane output
+persists on the VM under `~/.local/state/a0/logs`. `ai.sh keys` only propagates
+already-present VM login-environment keys into tmux and reports
+`present`/`missing`, never values.
+
 ## Maintenance tools
 
 Pure-stdlib helper scripts live in [`tools/`](tools/README.md). The small
@@ -208,7 +226,8 @@ authoritative spec.
 Skills come in two kinds. Pick the right one for what you're adding.
 
 **Metadata-block skills** apply the `msdmd` convention to a new block
-name (`doc-build`, `cap-build`, `deps-build`, `owner-build`, `test-build`, `meta-module-build`, `risk-boundary-build`, `ratios`, `manifest`, `llms-build`, and `typed-meta-frontend` are the existing examples).
+name (`doc-build`, `cap-build`, `deps-build`, `owner-build`, `test-build`, `meta-module-build`, `risk-boundary-build`,
+`ratios`, `manifest`, `llms-build`, and `typed-meta-frontend` are the existing examples).
 To add one:
 
 1. Pick a `<BLOCK_NAME>` (e.g. `DOCS`, `CAPABILITIES`, `OWNERS`, `LLMS`).