@ngockhoale/ukit 2.7.1 → 2.7.3

package/CHANGELOG.md

@@ -2,6 +2,78 @@
 
 All notable changes to UKit are documented here.
 
+## 2.7.3 - 2026-09-20
+
+AI freeze audit remediation — cycle C38 (TASK-001..007).
+
+- OpenCode residual sweep: `rg -i opencode` inventory classified into
+  keepers vs residuals; residual surfaces removed or steer-only, with a
+  zero-residual manifest test guarding regressions.
+- Unresolved-template-token repair: `mergeOmpConfig` + `diffPlan` detect
+  literal `{{token}}` left in rendered artifacts, repair known tokens,
+  report unknown ones via `unresolvedTokens` (never fabricate); install
+  prints a repair line like the C35 migration line.
+- `orchestration.allowUnsafeEval` (default `false`, optional-present):
+  only an explicit `true` renders `tools.approval.eval: allow` in
+  `.omp/config.yml` via the `omp.evalApproval` render variable; every
+  other state renders `deny` — fail-closed, never prompt. Migrations
+  never enable it; absence keeps the safe default.
+- `src/core/permissionPolicy.js` — pure semantic compiler
+  `compilePermissionPolicy({ permissionMode, allowUnsafeEval, host,
+  surfaces? })`: canonical deny set + precedence tiers
+  (hard-deny > protected-guard > host-deny > explicit-allow >
+  mode-fallback), models `interactive`/`safe-auto`/`unattended` across
+  claude/omp/codex, throws `UKIT_EPOLICY` on unknown mode. Render,
+  doctor, and the matrix share it as single source of truth.
+- `ukit doctor --permissions [--json]`: `src/core/permissionDoctor.js`
+  compiles the policy against live surfaces and reports FAIL classes
+  (unattended-prompt, deny-loss, unsafe-eval-mismatch,
+  unrendered-token) with exit 1; reachable prompts under
+  interactive/safe-auto report as INFO, absent host surfaces SKIP;
+  compact section in the default doctor run, full table under the flag.
+- No-freeze matrix (`tests/manifest/c38NoFreezeMatrix.test.js`):
+  safe→allow, destructive→deny+recovery, unknown-in-unattended never
+  asks, truncated payloads fail closed, merge idempotent, zero `{{`
+  residuals.
+- Canonical deny surfaces closed (R4.5 review): `templates/.claude/
+  settings.json` now denies the full canonical destructive set
+  (`git push -f`, `git clean -fd`, `git checkout .`, `git restore .`,
+  `rm -rf .`, `rm -rf`, `dd if=/dev/`, `*> /dev/sda`, `mkfs`,
+  fork-bomb) and the `gh pr create/comment` + `gh issue comment`
+  entries moved `ask`→`deny` — static surfaces can't express
+  ask-under-interactive-only, so deny is the only mode-safe posture
+  (unattended runs must never reach a prompt). `templates/.omp/
+  config.yml` gains `git push --force-with-lease*`. Report-only hosts
+  (codex) no longer count toward doctor `deny-loss`.
+- `ukit-bridge.js` (omp): `ask` verdicts now degrade to `block` with a
+  recorded cause + recovery route instead of surfacing an interactive
+  prompt under unattended mode (audit row 61).
+- Docs: `docs/UKIT_INTERNALS.md` gains a Permission Modes + Unsafe Eval
+  section — unattended ≠ all-destructive-allowed; a deny routes the
+  agent to the next safer alternative, and the host-internal-wait
+  guidance is unchanged.
+
+## 2.7.2 - 2026-09-20
+
+OpenCode adapter removal — cycle C37 (TASK-001..004).
+
+- Removed the OpenCode adapter lane end to end: adapter registry key +
+  aliases, `templates/adapter-presets/opencode/`, host-capability matrix
+  entry, doc-benchmark `opencode.yaml` fixture + BMK-09, manifest/manifest
+  wiring, `--tools` help listing, and root `opencode.json` dogfood file.
+- Steer-not-break detection: a leftover `opencode.json` is a user file —
+  install/status/doctor now detect it via `src/core/opencodeSteer.js` and
+  print one canonical warning advising manual removal (`rm opencode.json`);
+  the file is never written, merged, pruned, or auto-deleted.
+- Legacy tolerance: stale `agent`/`agents`/`compact.agentContext.targets`
+  values naming `opencode` are normalized away with a warning instead of
+  hard-failing; `--tools=opencode` / `--adapter opencode` warn + ignore /
+  fall back with exit 0; uninstall still removes a tracked `opencode.json`
+  on explicit user request; `opencode.json` stays in gitignore advice.
+- Packaging: `package.json` description/keywords/`files` and `.npmignore`
+  no longer reference the deleted preset; `packageArtifact.test.js` asserts
+  the tarball ships zero `adapter-presets/opencode` entries.
+
 ## 2.7.1 - 2026-09-20
 
 omp native-tier steering — cycle C36 (TASK-001..005).

package/README.md

@@ -1,6 +1,6 @@
 # UKit
 
-UKit installs a project-local AI workspace (`.claude/`, adapters, docs, and index helpers) so your team can work with Claude Code, OpenAI Codex, OpenCode, and omp (Oh My Pi) from the same repo setup.
+UKit installs a project-local AI workspace (`.claude/`, adapters, docs, and index helpers) so your team can work with Claude Code, OpenAI Codex, and omp (Oh My Pi) from the same repo setup.
 
 ## Install
 
@@ -39,7 +39,6 @@ Fill in the generated docs baseline (`docs/PROJECT.md`, `docs/MEMORY.md`, `docs/
 
 **Adapters**
 - `.codex/settings.json` + `.codex/settings.local.json` — Codex config
-- `opencode.json` — OpenCode project config with UKit-safe guardrails
 - `.omp/AGENTS.md` — thin `@../AGENTS.md` import so omp reads the shared instructions
 - `.omp/RULES.md` — sticky hard rules for omp sessions
 - `.omp/config.yml` — omp model roles, tool approval, and bash patterns
@@ -75,7 +74,6 @@ When long sessions approach the compact threshold, UKit compacts only safe-zone
 |------|-------------|--------|--------|
 | Claude Code | `CLAUDE.md` | `.claude/skills/` (canonical) | `.claude/settings.json` |
 | OpenAI Codex | `AGENTS.md` | `.claude/skills/` (referenced) | `.codex/settings.json`, `.codex/settings.local.json`, `.codex/README.md` |
-| OpenCode | `AGENTS.md` | `.claude/skills/` (native Claude-compatible support) | `opencode.json` |
 | omp (Oh My Pi) | `.omp/AGENTS.md` (imports `AGENTS.md`) | `.claude/skills/` (read via omp's claude provider — no symlink) | `.omp/config.yml`, `.omp/RULES.md`, `.omp/README.md` |
 
 All supported tools share the same SKILL.md ecosystem so the workspace is authored once and reused everywhere.

package/manifests/hostCapabilities.yaml

@@ -27,16 +27,6 @@ hosts:
     owner-file-injection:    { state: observed,  note: 'AGENTS.md rendered for codex; parity asserted only at artifact level' }
     settings-env:            { state: tested,    evidence: tests/integration/installPipeline.test.js }
     context-files:           { state: tested,    evidence: tests/integration/installPipeline.test.js }
-  opencode:
-    skills:                  { state: observed,  note: 'reads .claude/skills via discovery; asserted indirectly by artifact tests' }
-    subagents:               { state: unknown }
-    hooks:                   { state: unknown }
-    session-start-injection: { state: unknown }
-    stop-gate:               { state: unknown }
-    slash-commands:          { state: observed,  note: 'ukit-* helper entrypoints declared in opencode.json' }
-    owner-file-injection:    { state: observed,  note: 'AGENTS.md loaded at session start; no injection test' }
-    settings-env:            { state: tested,    evidence: tests/integration/installPipeline.test.js }
-    context-files:           { state: observed,  note: 'AGENTS.md asserted by install pipeline; no opencode-specific parse test' }
   omp:
     skills:                  { state: observed,  note: 'omp reads .claude/skills via claude discovery provider' }
     subagents:               { state: tested,    evidence: tests/consistency/ompAgentParity.test.js }

package/manifests/instructionRules.yaml

@@ -412,22 +412,6 @@ rules:
       When the same file or symbol fails debugLoopThreshold times in one
       session, route the next attempt one tier higher, capped at smart.
     marker_in: [templates/CLAUDE.md, templates/AGENTS.md, CLAUDE.md, AGENTS.md]
-  - id: HOST-OC-01
-    title: OpenCode must explicitly read the triggered SKILL.md
-    kind: critical
-    statement: >-
-      OpenCode reads AGENTS.md at session start only and does not auto-load
-      skills; the model must explicitly read the triggered SKILL.md.
-    marker_in: [templates/AGENTS.md, AGENTS.md]
-    anchor: 'read the triggered SKILL.md'
-  - id: HOST-OWN-01
-    title: Codex/OpenCode read the root owner instructions
-    kind: critical
-    statement: >-
-      When running in Codex or OpenCode, read and follow the root
-      PROJECT_IMPORTANT.md before doing project work.
-    marker_in: [templates/AGENTS.md, AGENTS.md]
-    anchor: 'read and follow the root'
   - id: DURA-01
     title: DuraOne skill activates only when installed
     kind: supporting

package/manifests/platform.full.yaml

@@ -124,8 +124,8 @@ items:
       - docs-project
       - docs-memory
     mergeStrategy: overwrite_with_backup
-    # Same set as root-claude-md: AGENTS.md is the same body with a different H1 plus one
-    # OpenCode-only section, so it interpolates the same variables. Kept in sync by
+    # Same set as root-claude-md: AGENTS.md is the same body with a different H1, so it
+    # interpolates the same variables. Kept in sync by
     # tests/consistency/contextDocsParity.test.js.
     variables:
       - project.name
@@ -1805,17 +1805,3 @@ items:
     packs:
       - core
 
-  # --- OpenCode ---
-
-  - id: multi-opencode-config
-    type: config
-    sourceTemplate: adapter-presets/opencode/opencode.template.json
-    targetPath: opencode.json
-    requires: []
-    mergeStrategy: overwrite_with_backup
-    variables:
-      - project.name
-      - runtime.packageManager
-    enabledByDefault: true
-    packs:
-      - core

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@ngockhoale/ukit",
-  "version": "2.7.1",
-  "description": "Install/update an index-first AI workspace for Claude Code, OpenAI Codex, OpenCode, and omp (Oh My Pi).",
+  "version": "2.7.3",
+  "description": "Install/update an index-first AI workspace for Claude Code, OpenAI Codex and omp (Oh My Pi).",
   "license": "MIT",
   "type": "module",
   "keywords": [
@@ -9,7 +9,6 @@
     "claude",
     "codex",
     "claude-code",
-    "opencode",
     "omp",
     "oh-my-pi",
     "skills",
@@ -37,8 +36,7 @@
     "templates/CLAUDE.md",
     "templates/AGENTS.md",
     "templates/.gitignore",
-    "manifests/",
-    "templates/adapter-presets/opencode/opencode.template.json"
+    "manifests/"
   ],
   "publishConfig": {
     "access": "public"

package/scripts/perf/audit-perf.mjs

@@ -3,8 +3,8 @@
  * audit-perf.mjs — TASK-233 (SPEC §2 FR-001) performance audit generator.
  *
  * Measures every SPEC §2 area from four independent sources and writes
- * `docs/AI_REPORT/perf-findings.json` (machine-readable, consumed by TASK-234,
- * TASK-235 and TASK-236) plus `docs/AI_REPORT/perf-measure.md` (raw appendix).
+ * `scripts/perf/perf-findings.json` (machine-readable, consumed by TASK-234,
+ * TASK-235 and TASK-236) plus `perf-measure.md` next to it (raw appendix).
  * Running this again with the same arguments IS the "measure it the same way
  * both times" requirement from SPEC §5 — TASK-236 re-runs it for before/after.
  *
@@ -735,7 +735,7 @@ function renderMeasureDoc(context) {
   lines.push('```bash');
   lines.push('node scripts/perf/audit-perf.mjs \\');
   lines.push(`  --telemetry-dir "${telemetry.dir || path.join(root, '.ukit', 'storage', 'cache', 'hook-latency')}" \\`);
-  lines.push(`  --out "${root}/docs/AI_REPORT/perf-findings.json"`);
+  lines.push(`  --out "${root}/scripts/perf/perf-findings.json"`);
   lines.push('node --test tests/handoff/c33/perfFindings.test.js');
   lines.push('```');
   lines.push('');
@@ -848,7 +848,7 @@ function main() {
   const options = parseArgs(process.argv.slice(2));
   const root = options.root;
   const telemetryDir = options.telemetryDir || path.join(root, '.ukit', 'storage', 'cache', 'hook-latency');
-  const outFile = options.out || path.join(root, 'docs', 'AI_REPORT', 'perf-findings.json');
+  const outFile = options.out || path.join(root, 'scripts', 'perf', 'perf-findings.json');
   const measureFile = path.join(path.dirname(outFile), 'perf-measure.md');
 
   const telemetry = loadTelemetry(telemetryDir);