open-multi-agent-kit 0.79.3 → 0.80.2

package/CHANGELOG.md

@@ -2,6 +2,44 @@
 
 ## [Unreleased]
 
+### Added
+
+- Reserved for changes after v0.80.2.
+
+## v0.80.2 — CI release gate stabilization (2026-06-16)
+
+### Fixed
+
+- Fixed run directory creation after stricter run artifact path validation.
+- Stabilized executor evidence gates so explicit output gates stay enforced while routing-only evidence hints do not block low-risk read/review lanes.
+- Preserved explicit runtime MCP/tool-calling capabilities when applying the provider authority matrix.
+- Required structured summary markers for summary/review evidence gates, preventing plain stdout from satisfying release evidence.
+
+## v0.80.1 — CI lint cleanup and registry-aligned patch release (2026-06-16)
+
+### Fixed
+
+- Removed release-blocking ESLint warnings/errors in native root loop, Headroom replay verifier, Freedomd runtime files, structured compaction, and session utilities.
+- Aligned package/docs/theme metadata for the `0.80.1` patch release after publishing `0.80.0`.
+
+## v0.80.0 — Freedomd sovereignty, default agent freedom, and release-ready GitHub docs (2026-06-16)
+
+### Added
+
+- Added Freedomd provider-sovereignty control plane: policy compiler, sovereignty scoring, retention gate, cutoff-resilient routing, degraded plans, local-first evidence envelopes, incident feeds, and provider exception artifacts.
+- Added default agent freedom behavior with strict guardrails opt-in via `OMK_STRICT_GUARDRAIL=1`; Freedomd routing remains explicitly controlled with `OMK_FREEDOMD_MODE=balanced|strict`.
+- Added Headroom control-plane hardening: loop-risk signals, replay verifier, compaction quality gate, artifact hashing, ontology relation gate, and adapter compaction consumption.
+- Added Freedomd README/release asset provenance and GitHub-facing documentation for provider independence, retention safety, and evidence portability.
+
+### Changed
+
+- Refreshed README, docs, roadmap, maturity notes, init templates, themes, and package metadata to align with `open-multi-agent-kit@0.80.0` and runtime contract family `v1.2`.
+- Wired Freedomd routing into the runtime-backed task runner and propagated sovereignty metadata through DAG/node routing, task context, and worker manifests.
+
+### Fixed
+
+- Closed remaining post-0.79.3 runtime hardening risks around prompt privacy, evidence gates, private stderr retention, provider health escalation, adapter cutoff handling, and graph ontology consistency.
+
 ## v0.79.3 — GLM 5.2 max thinking, scroll-safe TTY, and OSS CLI polish (2026-06-15)
 
 ### Added

package/MATURITY.md

@@ -1,7 +1,7 @@
 # OMK Command Maturity Matrix
 
 Last updated: 2026-05-24
-Current source version: v0.79.3
+Current source version: v0.80.2
 
 | Level | Meaning |
 |-------|---------|
@@ -72,5 +72,5 @@ Current source version: v0.79.3
 | Native runtime safety | OMK owns the root-orchestrator direction, but native chat must still lock turn-risk inference, approval/sandbox propagation, authority resolution, provider health probes, and DeepSeek read-only enforcement before stable provider-neutral claims. | Treat `docs/native-root-runtime-hardening.md` and `.omk/specs/native-orchestrator-phase1/` as the active hardening contract. |
 | MCP diagnostics | `mcp list/doctor/test` exist; invalid project/global MCP JSON now fails visibly through diagnostics without exposing config contents. | Add machine-readable MCP JSON and structured failure categories for command resolution, timeout, permission, and server health. |
 | Skills and harness templates | `omk skill` exposes current core/TypeScript/review packs, while init templates document project MCP scope, runtime skills, portable `.agents/skills`, and run-scoped harness manifests. | Keep external-inspired skills compact, source-linked, and non-vendored; verify install/sync through `skill-command` tests and package audit. |
-| Release docs and site | README, CHANGELOG, MATURITY, ROADMAP, getting-started docs, package audit, and release-gate commands now distinguish the v0.79.3 source target from registry-published claims while documenting alpha/experimental surfaces, parallel subagent orchestration, typed doctor repair plans, startup update prompts, native safety package readiness, current harness templates, packaged workflow skills, and the public project repository at `https://github.com/dmae97/open-multi-agent-kit`. | Treat `npm run release:check`, native safety packaging, tarball install smoke, GitHub Smoke Test, registry verification, and GitHub CI evidence on the exact commit as the publish/deploy gate before claiming v0.79.3 as released. |
+| Release docs and site | README, CHANGELOG, MATURITY, ROADMAP, getting-started docs, package audit, and release-gate commands now distinguish the v0.80.2 source target from registry-published claims while documenting alpha/experimental surfaces, parallel subagent orchestration, typed doctor repair plans, startup update prompts, native safety package readiness, current harness templates, packaged workflow skills, and the public project repository at `https://github.com/dmae97/open-multi-agent-kit`. | Treat `npm run release:check`, native safety packaging, tarball install smoke, GitHub Smoke Test, registry verification, and GitHub CI evidence on the exact commit as the publish/deploy gate before claiming v0.80.2 as released. |
 | Goal planner | Goal lifecycle exists, including continue, generated plan/evidence criteria, and verification. | Expand planner quality scoring and release evidence. |

package/README.md

@@ -1,25 +1,36 @@
-# OMK
+# OMK — Freedomd agent control plane
 
-Run multiple coding agents safely in one repo.
+**Local-first, multi-provider, evidence-based orchestration for coding agents.**
 
-OMK turns a coding task into a scoped DAG run: it routes the right model, limits tool access, requires evidence, and saves replayable artifacts before claiming success.
+OMK turns a coding goal into a scoped DAG run, routes it across compatible runtimes, limits authority per lane, records local evidence, and keeps the loop moving even when a provider is unavailable, retained, export-restricted, or policy-blocked.
 
-Use it when one agent is too loose, too risky, or too hard to audit.
+![OMK Freedomd runtime control plane](readmeasset/omk-freedomd-control-plane.webp)
 
 <p>
-  <a href="https://www.npmjs.com/package/open-multi-agent-kit"><img alt="npm version" src="https://img.shields.io/npm/v/open-multi-agent-kit?color=00D6FF"></a>
+  <a href="https://www.npmjs.com/package/open-multi-agent-kit"><img alt="npm version" src="https://img.shields.io/npm/v/open-multi-agent-kit?color=F2A900"></a>
   <a href="LICENSE"><img alt="license" src="https://img.shields.io/badge/license-MIT-blue.svg"></a>
-  <a href="https://github.com/dmae97/open-multi-agent-kit/blob/main/proof/PROOF_INDEX.md"><img alt="proof check" src="https://img.shields.io/badge/proof--check-source-00FFC2"></a>
+  <a href="https://github.com/dmae97/open-multi-agent-kit/blob/main/proof/PROOF_INDEX.md"><img alt="proof check" src="https://img.shields.io/badge/proof--check-source-2F9B4A"></a>
   <a href="https://github.com/dmae97/open-multi-agent-kit/discussions"><img alt="discussions" src="https://img.shields.io/badge/discussions-open-9D4EDD"></a>
 </p>
 
+## What changed in v0.80.2
+
+OMK now has a **Freedomd** control-plane direction:
+
+- **Provider sovereignty scoring** — route by health, authority, capability, retention, jurisdiction, cutoff risk, locality, and portability.
+- **Retention gate** — classify task data and decide `allow`, `redact`, `downgrade`, or `block` before a prompt leaves the local control plane.
+- **Cutoff-resilient routing** — provider incidents and policy blocks degrade to local authority, read-only review, artifact replay, or a clear diagnostic instead of silently stalling.
+- **Local-first evidence envelopes** — persist provider-independent evidence under `.omk/runs/<run-id>/freedomd/` with hashes for replay and audit.
+- **Default agent freedom** — high-autonomy agent behavior is default; strict guardrails are opt-in with `OMK_STRICT_GUARDRAIL=1`.
+- **Explicit Freedomd mode** — sovereignty routing is opt-in with `OMK_FREEDOMD_MODE=balanced` or `OMK_FREEDOMD_MODE=strict`.
+
 ## 30-second demo
 
 ```bash
 npx -y -p open-multi-agent-kit omk do "review this repo for release risk" --dry-run --json
 ```
 
-OMK should produce a bounded plan before changing files:
+A dry run produces bounded planning artifacts before changing files:
 
 ```txt
 goal
@@ -28,13 +39,13 @@ goal
 flow
   goal -> input envelope -> DAG -> scoped lanes -> evidence -> verify
 
-artifacts (dry run)
+artifacts
   .omk/runs/<run-id>/input-envelope.json
   .omk/runs/<run-id>/dag.json
   .omk/runs/<run-id>/dag-compile-report.json
 ```
 
-Executing (not dry-run) adds loop artifacts such as `.omk/runs/<run-id>/loop-state.json` and `.omk/runs/<run-id>/loop-decisions.jsonl`. Artifact names can evolve with the runtime contract; the invariant is that OMK persists reviewable run state before a worker claims completion.
+Executing a run adds loop state, route decisions, evidence observations, and replayable artifacts under `.omk/runs/<run-id>/`.
 
 ## Install
 
@@ -45,7 +56,7 @@ omk doctor
 omk chat
 ```
 
-For local verification from a checkout:
+Local checkout verification:
 
 ```bash
 npm ci
@@ -54,58 +65,94 @@ node dist/cli.js do "explain this repo" --dry-run --json
 npm run verify:no-kimi
 ```
 
-## Use OMK if you
+## Configure Freedomd
 
-- use Codex, Claude Code, OpenCode, DeepSeek, Qwen, OpenRouter, MiMo, or local models in the same repo;
-- want agents to produce evidence before saying “done”;
-- need read/write/shell authority separated by task;
-- want replayable `.omk/runs/` artifacts for review;
-- need project-scoped MCP, skills, hooks, and provider lanes instead of ambient global tool access.
+Balanced mode prefers provider diversity and records sovereignty diagnostics:
 
-## How OMK differs
+```bash
+OMK_FREEDOMD_MODE=balanced omk chat
+```
 
-| Tool type | Good at | Missing piece | OMK adds |
-| --- | --- | --- | --- |
-| Codex / Claude Code | Strong single-agent execution | Audit and routing layer | DAG, evidence, replay |
-| OpenCode | Terminal coding loop | Multi-provider governance | Provider authority + gates |
-| MCP servers | Tool access | Task-level control | Scoped MCP per lane |
-| CI | Post-hoc verification | Agent-time enforcement | Evidence before completion |
+Strict mode prefers local/self-hosted paths and blocks retained/export-restricted providers more aggressively:
+
+```bash
+OMK_FREEDOMD_MODE=strict OMK_MAX_RETENTION_DAYS=0 omk chat
+```
+
+Provider incident overrides are local and auditable:
+
+```json
+{
+  "schemaVersion": "omk.provider-incidents.v1",
+  "incidents": [
+    {
+      "providerId": "anthropic",
+      "kind": "export-control",
+      "severity": "block",
+      "reason": "operator policy override",
+      "updatedAt": "2026-06-16T00:00:00.000Z"
+    }
+  ]
+}
+```
+
+Save that as `.omk/provider-incidents.json`, or set `OMK_PROVIDER_INCIDENTS` to the same JSON. Optional public feeds can be supplied with `OMK_FREEDOMD_FEED_URLS`; feed failures never block routing and local overrides win.
+
+## Use OMK if you
+
+- use Codex, OpenCode, CommandCode, Kimi, DeepSeek, GLM, Qwen, OpenRouter, MiMo, or local models in the same repo;
+- want provider availability, retention, jurisdiction, and exportability treated as runtime risks;
+- need read/write/shell/merge authority separated by task;
+- want agents to produce replayable evidence before saying “done”;
+- need project-scoped MCP, skills, hooks, and graph memory instead of ambient global tool access;
+- need degraded local/review/replay modes when frontier providers are blocked.
 
 ## What OMK controls
 
 ```text
-Goal → DAG plan → scoped lanes → evidence bundle → verify gate → replay / inspect
+Goal -> DAG -> lane grants -> runtime routing -> evidence envelope -> verify -> replay
 ```
 
-- **Routing**: choose a compatible provider/runtime for the task.
-- **Authority**: keep read, write, shell, and merge powers explicit.
+- **Routing**: choose compatible provider/runtime lanes with capability and sovereignty scores.
+- **Authority**: keep read, write, patch, shell, MCP, and merge power explicit.
 - **Evidence**: require command output, diff, artifact, metric, or review proof.
-- **Replay**: save run artifacts under `.omk/runs/<run-id>/` for review.
+- **Replay**: save run artifacts under `.omk/runs/<run-id>/` for audit.
+- **Fallback**: degrade to local/review/replay modes when policy or provider health blocks execution.
 - **Scope**: keep MCP servers, skills, hooks, and memory bounded per lane.
 
-## Examples
+## How OMK differs
 
-- [Codex MCP evidence run](https://github.com/dmae97/open-multi-agent-kit/tree/main/examples/codex-mcp-evidence-run): project-scoped MCP setup plus evidence-gated DAG dry run.
-- [Provider fallback](https://github.com/dmae97/open-multi-agent-kit/tree/main/examples/provider-fallback): `--provider auto` routing with fallback planning.
-- [Proof index](https://github.com/dmae97/open-multi-agent-kit/blob/main/proof/PROOF_INDEX.md): source-controlled proof bundles checked by `npm run proof:check`.
+| Surface | Common failure | OMK adds |
+| --- | --- | --- |
+| Single-agent coding CLIs | Strong execution but weak audit | DAG, lane grants, local evidence, replay |
+| Cloud frontier providers | Policy/retention/export cutoff risk | Freedomd sovereignty scoring and degraded fallback |
+| MCP servers | Ambient tool access | Per-lane MCP/tool/shell authority |
+| CI | Post-hoc verification | Agent-time evidence gates before completion |
+| Local LLMs | Lower capability but high sovereignty | Local-first fallback and review/replay modes |
 
-## Maturity and safety claims
+## Release and safety claims
 
-Current source version: `open-multi-agent-kit@0.79.3`.
+Current source version: `open-multi-agent-kit@0.80.2`.
 
-- Public package name: `open-multi-agent-kit`.
-- Runtime contract family: `v1.2` means contract family, not a stable npm `1.x` release.
+- Runtime contract family: `v1.2`.
 - Release channel: `pre-1.0`.
-- OS-level sandboxing is planned, not claimed; child env hardening and approval gates are adapter-specific.
-- registry verification: treat npm `latest` claims as valid only after tagged CI and registry checks pass.
+- OS-level sandboxing is **not** claimed; safety relies on authority gates, approval/sandbox policy, scoped capabilities, evidence, and local artifacts.
+- npm `latest` claims are valid only after tagged CI and registry verification both pass.
+- README hero generation used the local Codex OAuth `gpt-image-2` workflow; the official `omk image` OpenAI Images API path still requires an OpenAI Platform project API key supplied only at runtime.
 
-See [versioning](docs/versioning.md), [provider maturity](docs/provider-maturity.md), and [SECURITY.md](SECURITY.md).
+See [versioning](docs/versioning.md), [claims](docs/claims.md), [provider maturity](docs/provider-maturity.md), and [SECURITY.md](SECURITY.md).
+
+## Examples
+
+- [Codex MCP evidence run](https://github.com/dmae97/open-multi-agent-kit/tree/main/examples/codex-mcp-evidence-run): project-scoped MCP setup plus evidence-gated DAG dry run.
+- [Provider fallback](https://github.com/dmae97/open-multi-agent-kit/tree/main/examples/provider-fallback): `--provider auto` routing with fallback planning.
+- [Proof index](https://github.com/dmae97/open-multi-agent-kit/blob/main/proof/PROOF_INDEX.md): source-controlled proof bundles checked by `npm run proof:check`.
 
 ## Community
 
 - Ask setup and provider questions in [Discussions](https://github.com/dmae97/open-multi-agent-kit/discussions).
 - File reproducible bugs in [Issues](https://github.com/dmae97/open-multi-agent-kit/issues).
-- If you want safer multi-agent coding runs, star the repo to follow the `v0.8` release line.
+- Star the repo to follow the `0.80.x` pre-1.0 Freedomd release line.
 
 ## License
 

package/ROADMAP.md

@@ -1,15 +1,17 @@
 # Roadmap
 
-Current source version: v1.1.18
-Last updated: 2026-05-24
+Current source version: `open-multi-agent-kit@0.80.2` (`pre-1.0`; runtime contract family `v1.2`)
+Last updated: 2026-06-15
 
-## 2026-05-24 runtime hardening status
+## 2026-06-15 release truth status
 
-Latest pushed source on `new-origin/main` is `6305e2b62185c11549f59e2340936769a3027cdd`. This supersedes the earlier native-root pivot commit in the same line. The architecture direction remains OMK-as-root with Kimi as the default coding adapter, but the current line is still hardening-gated:
+Current source and npm target is `open-multi-agent-kit@0.80.2` on the `v1.2` runtime contract family and `pre-1.0` release channel.
 
-- GitHub Actions Smoke Test is green on `6305e2b`.
-- GitHub Actions CI is red on Windows jobs on `6305e2b`; do not publish/tag v1.1.18 until this is fixed.
-- The active architecture backlog is now tracked in `docs/native-root-runtime-hardening.md`, `docs/native-root-runtime-algorithms.md`, and `.omk/specs/native-orchestrator-phase1/`.
+- Main CI and main Smoke are green on the `0.80.2` release commit.
+- Tag Release and tag Smoke are green for `v0.80.2`.
+- npm `latest` is registry-verified as `open-multi-agent-kit@0.80.2`.
+- OS-level sandboxing is still planned, not claimed; current safety relies on authority gates, approval policy, evidence, replay, and scoped runtime capabilities.
+- The active architecture backlog is tracked in `docs/native-root-runtime-hardening.md`, `docs/native-root-runtime-algorithms.md`, and `.omk/specs/native-orchestrator-phase1/`.
 
 ## v1.1.9 reality
 
@@ -17,7 +19,7 @@ Provider routing and graph viewing are no longer purely future work:
 
 - `omk run`, `omk parallel`, and DAG replay expose `--provider auto|kimi`.
 - `omk provider` / `omk deepseek` manage DeepSeek enablement, key setup, availability checks, and default fallback to the most mature adapter.
-- DeepSeek is an opportunistic read-only/advisory worker; Kimi remains the most mature adapter, orchestrator, writer, merger, and final authority.
+- DeepSeek and similar API providers remain read/review/advisory by default; write/shell/merge authority is selected by runtime-mode capability, health, approval policy, and sandbox constraints.
 - `omk graph view` generates an HTML view from `.omk/memory/graph-state.json`.
 - `omk goal` has a persisted lifecycle, continue loop, generated plan/evidence criteria, and verification flow.
 
@@ -67,13 +69,14 @@ Provider routing and graph viewing are no longer purely future work:
 ### P0: release and contract gates
 
 - Done: YAML validation now runs in local `verify` plus CI/smoke workflows.
-- Done: package dry-pack, package audit, tarball smoke, native safety build, and release matrix gates were re-verified against v1.1.17 artifacts.
-- Required before v1.1.18 publish/tag: regenerate the native safety binary, pass package audit, pass smoke-pack/tarball install smoke, and pass `npm run release:check` on the exact intended release diff.
-- Required before v1.1.18 publish/tag: GitHub Actions CI and Smoke Test must both pass on the exact intended commit.
-- Done: provider/deepseek and screenshot JSON command contracts gained hermetic regression tests.
-- Done: current AGENTS/init templates and packaged workflow skills were aligned with the active skills/MCP/agents/harness surface, including all generated agent MCP/skills/hooks flags and parallel subagent orchestration guidance.
-- Remaining: lock runtime safety gates for native turn risk, approval/sandbox propagation, authority-provider resolution, provider health probes, and DeepSeek read-only routing.
-- Remaining: lock broader provider fallback metadata with tests for rate limit, timeout, and default fallback variants.
+- Done: package dry-pack, package audit, tarball smoke, release matrix gates, GitHub Release, and npm registry verification were re-verified for `0.80.2`.
+- Done: `version:check` now validates current-version Markdown claims in README, docs, ROADMAP, and init templates.
+- Done: native turn risk defaults are safer: explicit read-only constraints override write keywords, ambiguous turns fall back to `ask`, and write/shell/merge turns require evidence gates.
+- Done: tool authority has staged `shadow|warn|enforce` modes, with native turn dispatch blocking in enforce mode.
+- Done: runtime routing is health-aware for async execution paths and records `executeTask` decision traces.
+- Done: headroom compaction can feed the effective runtime context capsule.
+- Required before any future publish/tag: pass `npm run release:check`, tag CI/smoke, tarball install smoke, package audit, and registry verification on the exact intended release diff.
+- Remaining: expand provider health from available/unavailable to binary/auth/model/quota/rate-limit/latency vectors across all adapters.
 - Remaining: define minimum machine-readable CLI envelopes for the rest of the automation-critical commands.
 
 ### P1: observability and diagnostics
@@ -95,8 +98,8 @@ Provider routing and graph viewing are no longer purely future work:
 
 ### Provider routing maturity
 
-- Keep Kimi as the most mature adapter and main orchestrator, planner, merger, and final synthesis runtime.
-- Use provider hints for explorer, reviewer, QA, planner, and documentation roles only when preflight is healthy and task risk is low.
+- Keep authority at the runtime-mode level: API runtimes are advisory/read/review unless a runtime-mode contract grants write/shell/merge authority.
+- Use provider hints for explorer, reviewer, QA, planner, and documentation roles only when runtime health is acceptable and task risk is low.
 - Record provider attempts, route confidence, fallback reason, and final authority in run evidence.
 
 ### Graph and memory maturity

package/dist/brand/night-city.theme.json

@@ -17,32 +17,129 @@
         "muted": "#9DB3C7",
         "gray": "#758FA8"
     },
-    "backgrounds": ["dark", "surface"],
+    "backgrounds": [
+        "dark",
+        "surface"
+    ],
     "semantics": {
-        "route.active": { "color": "cyan", "glyph": "▸", "usage": "text" },
-        "route.fallback": { "color": "amber", "glyph": "↻", "usage": "text" },
-        "route.dead": { "color": "red", "glyph": "✕", "usage": "text" },
-        "evidence.pass": { "color": "mint", "glyph": "✓", "usage": "text" },
-        "evidence.fail": { "color": "red", "glyph": "✗", "usage": "text" },
-        "evidence.pending": { "color": "amber", "glyph": "◐", "usage": "text" },
-        "dag.lane.running": { "color": "cyan", "glyph": "▶", "usage": "text" },
-        "dag.lane.blocked": { "color": "amber", "glyph": "⏸", "usage": "text" },
-        "dag.lane.done": { "color": "mint", "glyph": "●", "usage": "text" },
-        "dag.lane.queued": { "color": "muted", "glyph": "◌", "usage": "text" },
-        "telemetry.info": { "color": "cream", "glyph": "·", "usage": "text" },
-        "telemetry.warn": { "color": "amber", "glyph": "▲", "usage": "text" },
-        "telemetry.error": { "color": "red", "glyph": "✖", "usage": "text" },
-        "control.accent": { "color": "purple", "glyph": "◆", "kind": "chrome", "usage": "indicator" },
-        "control.dim": { "color": "gray", "glyph": "┊", "kind": "chrome", "usage": "text" },
-        "control.fg": { "color": "cream", "kind": "chrome", "usage": "text" },
-        "control.bg": { "color": "dark", "kind": "background", "usage": "background" }
+        "route.active": {
+            "color": "cyan",
+            "glyph": "▸",
+            "usage": "text"
+        },
+        "route.fallback": {
+            "color": "amber",
+            "glyph": "↻",
+            "usage": "text"
+        },
+        "route.dead": {
+            "color": "red",
+            "glyph": "✕",
+            "usage": "text"
+        },
+        "evidence.pass": {
+            "color": "mint",
+            "glyph": "✓",
+            "usage": "text"
+        },
+        "evidence.fail": {
+            "color": "red",
+            "glyph": "✗",
+            "usage": "text"
+        },
+        "evidence.pending": {
+            "color": "amber",
+            "glyph": "◐",
+            "usage": "text"
+        },
+        "dag.lane.running": {
+            "color": "cyan",
+            "glyph": "▶",
+            "usage": "text"
+        },
+        "dag.lane.blocked": {
+            "color": "amber",
+            "glyph": "⏸",
+            "usage": "text"
+        },
+        "dag.lane.done": {
+            "color": "mint",
+            "glyph": "●",
+            "usage": "text"
+        },
+        "dag.lane.queued": {
+            "color": "muted",
+            "glyph": "◌",
+            "usage": "text"
+        },
+        "telemetry.info": {
+            "color": "cream",
+            "glyph": "·",
+            "usage": "text"
+        },
+        "telemetry.warn": {
+            "color": "amber",
+            "glyph": "▲",
+            "usage": "text"
+        },
+        "telemetry.error": {
+            "color": "red",
+            "glyph": "✖",
+            "usage": "text"
+        },
+        "control.accent": {
+            "color": "purple",
+            "glyph": "◆",
+            "kind": "chrome",
+            "usage": "indicator"
+        },
+        "control.dim": {
+            "color": "gray",
+            "glyph": "┊",
+            "kind": "chrome",
+            "usage": "text"
+        },
+        "control.fg": {
+            "color": "cream",
+            "kind": "chrome",
+            "usage": "text"
+        },
+        "control.bg": {
+            "color": "dark",
+            "kind": "background",
+            "usage": "background"
+        }
     },
     "components": {
-        "statusCard": { "border": "control.accent", "title": "control.fg", "body": "telemetry.info", "meta": "control.dim" },
-        "routerTable": { "active": "route.active", "fallback": "route.fallback", "dead": "route.dead", "header": "control.accent" },
-        "evidenceGate": { "pass": "evidence.pass", "fail": "evidence.fail", "pending": "evidence.pending" },
-        "dagLanes": { "running": "dag.lane.running", "blocked": "dag.lane.blocked", "done": "dag.lane.done", "queued": "dag.lane.queued" },
-        "logStream": { "info": "telemetry.info", "warn": "telemetry.warn", "error": "telemetry.error", "dim": "control.dim" }
+        "statusCard": {
+            "border": "control.accent",
+            "title": "control.fg",
+            "body": "telemetry.info",
+            "meta": "control.dim"
+        },
+        "routerTable": {
+            "active": "route.active",
+            "fallback": "route.fallback",
+            "dead": "route.dead",
+            "header": "control.accent"
+        },
+        "evidenceGate": {
+            "pass": "evidence.pass",
+            "fail": "evidence.fail",
+            "pending": "evidence.pending"
+        },
+        "dagLanes": {
+            "running": "dag.lane.running",
+            "blocked": "dag.lane.blocked",
+            "done": "dag.lane.done",
+            "queued": "dag.lane.queued"
+        },
+        "logStream": {
+            "info": "telemetry.info",
+            "warn": "telemetry.warn",
+            "error": "telemetry.error",
+            "dim": "control.dim"
+        }
     },
     "fallback16": {
         "route.active": "brightCyan",
@@ -75,6 +172,7 @@
             "readmeasset/omk-evidence-ledger.svg",
             "readmeasset/omk-logo-mark.svg",
             "readmeasset/omk-provider-lanes.svg"
-        ]
+        ],
+        "omkVersion": "0.80.2"
     }
 }

package/dist/brand/rust-forge.theme.json

@@ -210,7 +210,7 @@
     "meta": {
         "provenance": "Standalone oxidized rust-forge palette for the OMK orchestration console. Warm coal/iron/anvil backgrounds, rust-orange + molten + copper + ember heat accents, oxide/quench/verdigris patina greens for pass/done, crimson for fail/dead, steel/scale/ash grays, and bone foreground. Forge-specific heat/anvil/quench semantics make OMK control surfaces distinct from provider branding while preserving NO_COLOR glyph safety.",
         "tierAssumptions": "16-color tier gated against stock VGA/xterm ANSI values; terminals with remapped ANSI-16 palettes are outside the contract — glyphs are the safety net. NO_COLOR note: control.dim uses ┊ so logStream info (telemetry.info ·) and dim rows stay distinguishable without color.",
-        "omkVersion": "0.78.6",
+        "omkVersion": "0.80.2",
         "independence": "OMK is the control plane; providers and models are adapter lanes, never the product identity or final authority."
     }
 }

package/dist/commands/chat/native-root-loop.d.ts

@@ -1,8 +1,9 @@
-import type { TaskRunner } from "../../contracts/orchestration.js";
+import type { TaskResult, TaskRunner } from "../../contracts/orchestration.js";
 import type { RuntimeBootstrap } from "../../runtime/runtime-bootstrap.js";
 import type { ChatLayout } from "./utils.js";
 import { type DagNode } from "../../orchestration/dag.js";
 import type { CliRenderer } from "../../cli/ui/renderer.js";
+import type { TaskRunContext } from "../../contracts/worker-context.js";
 import type { OmkBrandThemeName, OmkTuiMotion } from "../../brand/theme.js";
 import type { TuiView } from "../../tui/model.js";
 export interface NativeRootLoopInput {
@@ -44,8 +45,18 @@ export declare function createNativeRootSessionState(input: {
     bootstrap: RuntimeBootstrap;
     executionPrompt?: string;
 }): NativeRootSessionState;
-export type NativeTurnRisk = "read" | "write" | "shell" | "merge";
-export declare function inferNativeTurnRisk(prompt: string): NativeTurnRisk;
+export type NativeTurnRisk = "read" | "write" | "shell" | "merge" | "ask";
+export interface NativeTurnRiskTrace {
+    readonly risk: NativeTurnRisk;
+    readonly confidence: number;
+    readonly matchedSignal: string;
+    readonly readOnlyOverride: boolean;
+    readonly positiveSignals: readonly string[];
+    readonly negativeScopes: readonly string[];
+    readonly requestedOps: readonly NativeTurnRisk[];
+    readonly excludedOps: readonly NativeTurnRisk[];
+}
+export declare function inferNativeTurnRisk(prompt: string): NativeTurnRiskTrace;
 export declare function buildNativeRootLoopTurnNode(input: {
     bootstrap: RuntimeBootstrap;
     prompt: string;
@@ -54,5 +65,18 @@ export declare function buildNativeRootLoopTurnNode(input: {
     skillNames?: readonly string[];
     hookNames?: readonly string[];
     executionPrompt?: string;
+    runId?: string;
 }): DagNode;
+export declare function executeNativeRootTurn(input: {
+    taskRunner: TaskRunner;
+    node: DagNode;
+    env: Record<string, string>;
+    signal: AbortSignal;
+    heartbeatEnabled: boolean;
+    renderer?: CliRenderer;
+    mcpAllowlist?: readonly string[];
+    skillNames?: readonly string[];
+    hookNames?: readonly string[];
+    runContext?: TaskRunContext;
+}): Promise<TaskResult>;
 export declare function runNativeOmkRootLoop(input: NativeRootLoopInput): Promise<number>;