@lifeaitools/rdc-skills 0.35.21 → 0.35.23

package/skills/behavior-audit/agents/openai.yaml

@@ -1,4 +1,4 @@
-interface:
-  display_name: "Behavior Audit"
-  short_description: "Audit Claude and Codex behavior against shared rules."
-  default_prompt: "Run a behavior audit for the last seven days and write the evidence bundle to a report directory."
+interface:
+  display_name: "Behavior Audit"
+  short_description: "Audit Claude and Codex behavior against shared rules."
+  default_prompt: "Run a behavior audit for the last seven days and write the evidence bundle to a report directory."

package/skills/deploy/SKILL.md

@@ -7,6 +7,13 @@ description: rdc:deploy (slug, [action]) — run one command; exit 0 means shipp
 > One command, then one line: the URL on exit 0, or the `DEPLOY-FAILED` block on
 > non-zero. No tool-call narration, no raw JSON dumps, no progress commentary.
 
+> **Sandbox contract:** This skill honors `RDC_TEST=1` per
+> `.rdc/guides/agent-bootstrap.md` § RDC_TEST Sandbox Contract. Under the flag
+> the run stays read-only — registry read plus health probe — and every
+> mutating step short-circuits: no deploy triggered, no Cloudflare cache purge,
+> no DNS write, no promote to main. `skills/tests/rdc-deploy.test.json` asserts
+> exactly those four absences, so this is an enforced contract, not a promise.
+
 # rdc:deploy — run the program, read the exit code
 
 **This skill is not a procedure. It is one command.**

package/skills/design/SKILL.md

@@ -36,6 +36,73 @@ RDC-owned design execution for Studio and LIFEAI interfaces. This skill is the S
 - `rdc:design critique <target>` — UX/design critique
 - `rdc:design polish <target>` — final visual and interaction pass
 - `rdc:design craft <feature>` — shape and build a token-aware UI feature
+- `rdc:design prototype <description>` — build a new UI surface for visual review
+- `rdc:design compare <brief>` — dispatch the same design brief to a Claude-side pass AND a Codex-side pass in parallel, then synthesize a structured comparison (any mode also accepts a trailing `--compare` flag to run the same dual-dispatch before finalizing that mode's output)
+
+## New-Surface Gate — PRODUCT.md / DESIGN.md Before Source Mutation
+
+> Promoted from `rdc:onramp` (Phase 1 disk-tree scaffold + Phase 4 brand-book rubric gate),
+> which is the existing model for this pattern — read `skills/onramp/SKILL.md` §Phase 1.5,
+> §Phase 4.6 if this section is ever unclear. Onramp scoped this to `places/<slug>/`; this
+> section is the same mechanism generalized to any new or reshaped product surface. Approved:
+> Dave, direct instruction, 2026-09-05/06 session — "the existing onramp version is the model;
+> it was simply scoped too narrowly."
+
+**Applies to:** `craft`, `prototype`, and any other mode about to build a **new or reshaped
+product surface** — a new app/site/package root, a new top-level route, or a rename/merge/
+split of an existing screen map. **Does not apply to** `edit`, `audit`, `critique`, `polish`,
+`tokens`, `palette`, `theme`, or `colorize` against an already-documented surface — those
+operate against an existing PRODUCT.md/DESIGN.md, they don't have to create one first.
+
+**Small-fix exemption — same boundary as `rdc:fixit`:** a change touching **fewer than 5
+files** that does **not** introduce a new route, a new top-level screen, or a new app/site
+root is exempt. Anything at or above that, or anything that adds a route/screen/app root
+regardless of file count, is in scope for this gate.
+
+### The gate
+
+Before writing or editing any source file for an in-scope task:
+
+1. **Locate the surface root** — the app/site/package directory, or the repo root for a
+   standalone project (e.g. `C:/Dev/rdc-cde/PRODUCT.md`, not a subdirectory).
+2. **Check for `PRODUCT.md` and `DESIGN.md` at that root.**
+3. **If either is absent, or present but stale against the requested surface** (its screen
+   map / information hierarchy does not cover the screen, route, or object the task is about
+   to add or change) — **refuse to proceed to source mutation.** Create or update the missing
+   or stale document(s) first, using the field lists below. This mirrors onramp's Phase 4
+   rubric row *"DESIGN.md — 24-spread outline, not a stub"* and the Phase 1 gate *"disk:
+   places/<slug>/ exists with all required files"* — a present-but-stub or present-but-stale
+   document fails the gate exactly like an absent one.
+4. Only after both documents exist and are current does the mode proceed to `craft`/
+   `prototype` source work.
+
+### PRODUCT.md — required fields
+
+| Field | Content |
+|---|---|
+| Operator | Who runs/uses this surface day to day |
+| Job-to-be-done | The concrete task the surface exists to complete |
+| Durable objects + authority | The entities the UI reads/writes and who/what owns each one |
+| Screen map | Every screen/route/panel, one row each, with its purpose |
+| Acceptance outcomes | What "this surface works" means, observably |
+
+### DESIGN.md — required fields
+
+| Field | Content |
+|---|---|
+| Dominant visual object | The one element the eye lands on first, per screen |
+| Information hierarchy | What outranks what, and why |
+| Interaction model | How the user acts on the dominant object and its neighbors |
+| Responsive transformation | What changes, collapses, or reflows at narrower widths |
+| Semantic color | Token names mapped to meaning (never raw hex) |
+| References | Existing surfaces/patterns this design borrows from, cited by path |
+| Anti-patterns | What this surface deliberately does NOT do |
+| Visual acceptance criteria | What a reviewer checks to call the visual design done |
+
+**Refusal wording** (use verbatim, adapted to the target): *"BLOCKED: `<surface>` has no
+PRODUCT.md/DESIGN.md at its root (or they are stale against this request). Per the
+new-surface gate, I'm writing/updating those first before touching source."* Then write them,
+then proceed.
 
 ## Required References
 
@@ -168,6 +235,52 @@ Project docs to read for Studio work:
    - For Studio: prefer route smoke checks, token API checks, and browser screenshots when UI changed.
    - For CLI prompt work: run `node {RDC_SKILLS_ROOT}/scripts/rdc-design-cli.mjs <command> <brief>` and inspect the generated report under `.rdc/reports/rdc-design-cli/`.
 
+## Design Compare — Claude vs Codex, Dispatched in Parallel
+
+Two AI engines are already wired for dual-dispatch across this fleet — Claude Code and
+Codex (OpenAI/ChatGPT models) — via `codex exec`. This is not a new integration; it is the
+same mechanism `scripts/lib/runner.mjs` already uses to spawn both `claude --print
+--output-format stream-json` and `codex exec --json` against a prompt, and the same one
+CDE's `apps/implementor-manager/src/codex-cli-implementor.ts` uses for a fresh read-only
+planning turn (`codex exec --sandbox read-only --json --output-schema <schema> <prompt>`).
+
+`rdc:design compare <brief>` (or any mode + `--compare`) runs that mechanism against a
+design brief instead of a code-change prompt:
+
+```powershell
+node {RDC_SKILLS_ROOT}/scripts/rdc-design-compare-cli.mjs "<design brief>" [--out <dir>]
+```
+
+This is a **real dispatch**, not a description of one: the script spawns a fresh
+`claude --print` process and a fresh `codex exec --sandbox read-only --json` process **in
+parallel**, both against the identical brief plus the design skill's own operating
+instructions (same construction as `rdc-design-cli.mjs`'s prompt assembly), and both told
+explicitly not to mutate files — this is a proposal pass, not an edit pass. It waits for
+both, extracts each engine's final message, and writes:
+
+```text
+{RDC_SKILLS_ROOT}/.rdc/reports/rdc-design-cli/compare-<timestamp>/
+  claude.md     — Claude's raw design proposal
+  codex.md      — Codex's raw design proposal
+  report.json   — both outputs + timing + exit codes
+```
+
+**After the script returns, the calling agent (not the script) writes the comparison** —
+read both `claude.md` and `codex.md` and produce a structured table, never concatenation:
+
+| Dimension | Claude proposal | Codex proposal | Verdict |
+|---|---|---|---|
+| Dominant visual object | ... | ... | which fits the brief |
+| Information hierarchy | ... | ... | which fits the brief |
+| Interaction model | ... | ... | which fits the brief |
+| Token/brand-system fit | ... | ... | which reuses existing tokens vs invents new ones |
+| Tradeoffs | ... | ... | cost of each approach |
+
+Recommend one, or a hybrid, with the reason stated — never "both are good options."
+
+If `codex` is unreachable (binary missing, non-zero exit), report that plainly and fall
+back to a single-engine pass; do not fabricate a second opinion.
+
 ## Command Menu
 
 | Command | Purpose |
@@ -181,7 +294,9 @@ Project docs to read for Studio work:
 | `audit` | Technical and visual audit |
 | `critique` | UX/design review |
 | `polish` | Final detail pass |
-| `craft` | Shape and build a token-aware interface |
+| `craft` | Shape and build a token-aware interface — gated by New-Surface Gate above |
+| `prototype` | Build a new UI surface for visual review — gated by New-Surface Gate above |
+| `compare` | Dispatch the same design brief to Claude AND Codex in parallel, synthesize a comparison |
 
 ## CLI Helper
 
@@ -191,6 +306,7 @@ Use the local helper to see exactly how much instruction text the skill is gener
 node {RDC_SKILLS_ROOT}/scripts/rdc-design-cli.mjs studio "audit the Studio palette page"
 node {RDC_SKILLS_ROOT}/scripts/rdc-design-cli.mjs palette "generate a PRT palette workflow"
 node {RDC_SKILLS_ROOT}/scripts/rdc-design-cli.mjs --json theme "RDC earth-forward light theme"
+node {RDC_SKILLS_ROOT}/scripts/rdc-design-compare-cli.mjs "dashboard hero panel for a stewardship dashboard"
 ```
 
 The helper writes logs to:
@@ -199,7 +315,7 @@ The helper writes logs to:
 {RDC_SKILLS_ROOT}/.rdc/reports/rdc-design-cli/
 ```
 
-Each run includes character count, word count, approximate token count, references loaded, and the final prompt text.
+Each run includes character count, word count, approximate token count, references loaded, and the final prompt text. `rdc-design-compare-cli.mjs` writes its dual-engine outputs under `.rdc/reports/rdc-design-cli/compare-<timestamp>/` (see Design Compare above).
 
 ## Boundaries
 
@@ -209,3 +325,7 @@ Each run includes character count, word count, approximate token count, referenc
 - Do not persist Rampa output directly to production without Studio token mapping.
 - Do not fork or ship Palette Designer until its license is verified.
 - Do not use Studio's deprecated `/api/editor/render` route.
+- Do not skip the New-Surface Gate for `craft`/`prototype` work on a new or reshaped
+  surface, even when the requester did not mention PRODUCT.md/DESIGN.md.
+- Do not let `compare` mode's Codex-side pass mutate files — it runs `--sandbox read-only`
+  by design; if a prompt change would lift that, treat it as an architectural change.

package/skills/edit/SKILL.md

@@ -17,8 +17,10 @@ description: rdc:edit (target) — open the local website editor for a target
 **This skill is the designated handler `rdc-harness` dispatches to.**
 `packages/work/src/editors.mjs`'s `BY_CLASS` table maps `site-html`/`site-ts`
 product classes to editor id `'rdc:edit'` — when an agent calls
-`npx --package=@lifeai/rdc-harness rdc-harness edit <slug>` (the published
-package; see `guides/agent-bootstrap.md` for registry setup) against a website
+`npx -y --package=@lifeai/rdc-harness rdc-harness edit <slug>` (published on
+GitHub Packages — needs `@lifeai:registry=https://npm.pkg.github.com` in
+`.npmrc`; an `npm` 404 means that mapping is missing. Full setup:
+`{PROJECT_ROOT}/.rdc/guides/agent-bootstrap.md`) against a website
 target, its receipt names `editorId: 'rdc:edit'` and an `editableBoundary`.
 This skill IS that handler: resolve the same target, launch/reuse the editor
 host, use the harness's `editableBoundary` as the save boundary if the call

package/skills/open/SKILL.md

@@ -12,12 +12,14 @@ description: rdc:open ([slug]) — orient before working; answers where you are
 ## Run the script. Do not re-derive this by hand.
 
 `rdc-harness` ships as a published package — `@lifeai/rdc-harness` on GitHub
-Packages, not npmjs (see `guides/agent-bootstrap.md` for the registry
-precondition). Invoke it through the installed `bin`, never a
-machine-specific checkout path:
+Packages, not npmjs. That needs `@lifeai:registry=https://npm.pkg.github.com`
+in `.npmrc`; if `npm` 404s on the package, that mapping is what is missing
+(auth token: the clauth `github` PAT — full setup in
+`.rdc/guides/agent-bootstrap.md`). Invoke it through the installed `bin`,
+never a machine-specific checkout path:
 
 ```bash
-npx --package=@lifeai/rdc-harness rdc-harness open <slug>
+npx -y --package=@lifeai/rdc-harness rdc-harness open <slug>
 ```
 
 Only if you are actively developing the harness itself (working ON
@@ -143,9 +145,10 @@ Downstream verbs re-resolve from the registry by slug.
 
 ## Steps
 
-0. **Call `rdc-harness open <slug>`.** Steps 1–2 are what it returns, not a
-   procedure to perform. Only hand-resolve if the harness is genuinely
-   unavailable, and say so explicitly when you do.
+0. **Call `npx -y --package=@lifeai/rdc-harness rdc-harness open <slug>`.**
+   Steps 1–2 are what it returns, not a procedure to perform. Only
+   hand-resolve if the harness is genuinely unavailable, and say so
+   explicitly when you do.
 1. **Position.** Read `sourceBoundary` and `boundaryNote` from the result — they
    already account for pooled repos and lanes. Relative paths from there are
    correct by construction; a hardcoded `C:/Dev/regen-root/...` from a lane

package/skills/release/SKILL.md

@@ -94,11 +94,13 @@ monorepo subtree, not a standalone repo like this one),
 "Tests/self-test passed" through "Local install/update executed" rows safely:
 real `npm pack`, isolated-prefix install (never the real global store), real
 verify, and `--live` explicitly gating the actual publish. Where applicable,
-`npx --package=@lifeai/rdc-harness rdc-harness deploy <slug> [--live]` (the
-published package; see `guides/agent-bootstrap.md` for registry setup) can
-supply that evidence directly instead of hand-rolling the same
-pack/install/verify cycle. It does **not** replace version bump, tag or push
-— the harness CLI does none of those.
+`npx -y --package=@lifeai/rdc-harness rdc-harness deploy <slug> [--live]`
+(published on GitHub Packages — needs
+`@lifeai:registry=https://npm.pkg.github.com` in `.npmrc`; an `npm` 404 means
+that mapping is missing. Full setup:
+`{PROJECT_ROOT}/.rdc/guides/agent-bootstrap.md`) can supply that evidence
+directly instead of hand-rolling the same pack/install/verify cycle. It does
+**not** replace version bump, tag or push — the harness CLI does none of those.
 
 ## Resolution Order
 

package/skills/tests/onramp.test.json

@@ -1,101 +1,101 @@
-{
-  "manifest_version": 1,
-  "skill": "rdc:onramp",
-  "description": "Tier-2 self-test for rdc:onramp enrollment skill",
-  "tier": 2,
-  "sandbox": true,
-  "env": {
-    "RDC_TEST": "1"
-  },
-  "invocation": {
-    "slug": "onramp-selftest",
-    "name": "Onramp Self Test",
-    "flags": ["--archetype", "working-landscapes", "--dry-run"]
-  },
-  "fixture": {
-    "prompt": "rdc:onramp onramp-selftest --name \"Onramp Self Test\" --archetype working-landscapes --dry-run",
-    "env": { "RDC_\u0054EST": "1" },
-    "slow": true
-  },
-  "assertions": {
-    "exit_code": 0,
-    "stdout_contains": ["Preflight", "Resolve drift"]
-  },
-  "acceptance": {
-    "output_contains": ["Preflight", "Resolve drift", "disk-ahead", "already-enrolled"],
-    "output_not_contains": ["hand-edit _context.md", "raw insert"]
-  },
-  "legacy_assertions": [
-    {
-      "type": "exit_code",
-      "expect": 0,
-      "description": "Skill exits cleanly under RDC_TEST with --dry-run"
-    },
-    {
-      "type": "stdout_contains",
-      "pattern": "Preflight",
-      "description": "Checklist begins with Preflight step"
-    },
-    {
-      "type": "stdout_contains",
-      "pattern": "Resolve drift",
-      "description": "Drift resolution step is emitted"
-    },
-    {
-      "type": "stdout_contains",
-      "pattern": "\\[RDC_TEST\\]",
-      "description": "Sandbox short-circuit marker present — confirms external calls were skipped"
-    },
-    {
-      "type": "file_absent",
-      "path": "places/onramp-selftest/_context.md",
-      "description": "Scaffolder MUST NOT write _context.md — projection-drift violation gate"
-    }
-  ],
-  "door_contracts": [
-    {
-      "name": "enroll_place_signature",
-      "description": "enroll_place RPC must accept (text, text, jsonb, text) and return jsonb",
-      "type": "rpc_introspection",
-      "project_id": "uvojezuorjgqzmhhgluu",
-      "query": "SELECT p.proname, pg_get_function_identity_arguments(p.oid) AS args, pg_get_function_result(p.oid) AS returns FROM pg_proc p WHERE p.proname = 'enroll_place';",
-      "expect_args_contains": ["text", "jsonb"],
-      "expect_returns": "jsonb"
-    },
-    {
-      "name": "insert_work_item_accepts_project_node_id",
-      "description": "insert_work_item must accept p_project_node_id parameter",
-      "type": "rpc_introspection",
-      "project_id": "uvojezuorjgqzmhhgluu",
-      "query": "SELECT proname, pg_get_function_identity_arguments(oid) FROM pg_proc WHERE proname = 'insert_work_item';",
-      "expect_args_contains": ["p_project_node_id"]
-    }
-  ],
-  "disk_ahead_test": {
-    "description": "disk-ahead state must STOP and open a reconciliation work item",
-    "setup": "mkdir -p places/disk-ahead-test",
-    "invocation_slug": "disk-ahead-test",
-    "invocation_name": "Disk Ahead Test",
-    "assertions": [
-      {
-        "type": "stdout_contains",
-        "pattern": "disk-ahead",
-        "description": "Skill identifies the disk-ahead state"
-      },
-      {
-        "type": "stdout_contains",
-        "pattern": "STOP",
-        "description": "Skill STOPs on disk-ahead"
-      },
-      {
-        "type": "exit_nonzero",
-        "description": "Skill exits non-zero on disk-ahead (not a clean enrollment)"
-      }
-    ],
-    "teardown": "rmdir places/disk-ahead-test 2>/dev/null || true"
-  },
-  "idempotency_test": {
-    "description": "Running rdc:onramp twice for already-enrolled should produce zero new writes",
-    "note": "Under RDC_TEST=1 this is verified by checking that [RDC_TEST] skipping markers appear (not real insert calls) and the checklist reports already-enrolled"
-  }
-}
+{
+  "manifest_version": 1,
+  "skill": "rdc:onramp",
+  "description": "Tier-2 self-test for rdc:onramp enrollment skill",
+  "tier": 2,
+  "sandbox": true,
+  "env": {
+    "RDC_TEST": "1"
+  },
+  "invocation": {
+    "slug": "onramp-selftest",
+    "name": "Onramp Self Test",
+    "flags": ["--archetype", "working-landscapes", "--dry-run"]
+  },
+  "fixture": {
+    "prompt": "rdc:onramp onramp-selftest --name \"Onramp Self Test\" --archetype working-landscapes --dry-run",
+    "env": { "RDC_\u0054EST": "1" },
+    "slow": true
+  },
+  "assertions": {
+    "exit_code": 0,
+    "stdout_contains": ["Preflight", "Resolve drift"]
+  },
+  "acceptance": {
+    "output_contains": ["Preflight", "Resolve drift", "disk-ahead", "already-enrolled"],
+    "output_not_contains": ["hand-edit _context.md", "raw insert"]
+  },
+  "legacy_assertions": [
+    {
+      "type": "exit_code",
+      "expect": 0,
+      "description": "Skill exits cleanly under RDC_TEST with --dry-run"
+    },
+    {
+      "type": "stdout_contains",
+      "pattern": "Preflight",
+      "description": "Checklist begins with Preflight step"
+    },
+    {
+      "type": "stdout_contains",
+      "pattern": "Resolve drift",
+      "description": "Drift resolution step is emitted"
+    },
+    {
+      "type": "stdout_contains",
+      "pattern": "\\[RDC_TEST\\]",
+      "description": "Sandbox short-circuit marker present — confirms external calls were skipped"
+    },
+    {
+      "type": "file_absent",
+      "path": "places/onramp-selftest/_context.md",
+      "description": "Scaffolder MUST NOT write _context.md — projection-drift violation gate"
+    }
+  ],
+  "door_contracts": [
+    {
+      "name": "enroll_place_signature",
+      "description": "enroll_place RPC must accept (text, text, jsonb, text) and return jsonb",
+      "type": "rpc_introspection",
+      "project_id": "uvojezuorjgqzmhhgluu",
+      "query": "SELECT p.proname, pg_get_function_identity_arguments(p.oid) AS args, pg_get_function_result(p.oid) AS returns FROM pg_proc p WHERE p.proname = 'enroll_place';",
+      "expect_args_contains": ["text", "jsonb"],
+      "expect_returns": "jsonb"
+    },
+    {
+      "name": "insert_work_item_accepts_project_node_id",
+      "description": "insert_work_item must accept p_project_node_id parameter",
+      "type": "rpc_introspection",
+      "project_id": "uvojezuorjgqzmhhgluu",
+      "query": "SELECT proname, pg_get_function_identity_arguments(oid) FROM pg_proc WHERE proname = 'insert_work_item';",
+      "expect_args_contains": ["p_project_node_id"]
+    }
+  ],
+  "disk_ahead_test": {
+    "description": "disk-ahead state must STOP and open a reconciliation work item",
+    "setup": "mkdir -p places/disk-ahead-test",
+    "invocation_slug": "disk-ahead-test",
+    "invocation_name": "Disk Ahead Test",
+    "assertions": [
+      {
+        "type": "stdout_contains",
+        "pattern": "disk-ahead",
+        "description": "Skill identifies the disk-ahead state"
+      },
+      {
+        "type": "stdout_contains",
+        "pattern": "STOP",
+        "description": "Skill STOPs on disk-ahead"
+      },
+      {
+        "type": "exit_nonzero",
+        "description": "Skill exits non-zero on disk-ahead (not a clean enrollment)"
+      }
+    ],
+    "teardown": "rmdir places/disk-ahead-test 2>/dev/null || true"
+  },
+  "idempotency_test": {
+    "description": "Running rdc:onramp twice for already-enrolled should produce zero new writes",
+    "note": "Under RDC_TEST=1 this is verified by checking that [RDC_TEST] skipping markers appear (not real insert calls) and the checklist reports already-enrolled"
+  }
+}

package/skills/tests/rdc-env.test.json

@@ -1,12 +1,12 @@
-{
-  "manifest_version": 1,
-  "skill": "rdc:env",
-  "description": "Environment status is read-only and reports installed version.",
-  "fixture": { "prompt": "rdc:env status", "env": { "RDC_\u0054EST": "1" }, "slow": false },
-  "assertions": { "exit_code": 0, "stdout_contains": ["Environment"] },
-  "acceptance": {
-    "output_contains": ["version", "status"],
-    "output_not_contains": ["force", "delete environment"]
-  },
-  "teardown": { "reset_branch": true }
-}
+{
+  "manifest_version": 1,
+  "skill": "rdc:env",
+  "description": "Environment status is read-only and reports installed version.",
+  "fixture": { "prompt": "rdc:env status", "env": { "RDC_\u0054EST": "1" }, "slow": false },
+  "assertions": { "exit_code": 0, "stdout_contains": ["Environment"] },
+  "acceptance": {
+    "output_contains": ["version", "status"],
+    "output_not_contains": ["force", "delete environment"]
+  },
+  "teardown": { "reset_branch": true }
+}

package/skills/tests/rdc-new-model.test.json

@@ -1,12 +1,12 @@
-{
-  "manifest_version": 1,
-  "skill": "rdc:new-model",
-  "description": "New-model registration begins with the canonical gate.",
-  "fixture": { "prompt": "rdc:new-model example-model", "env": { "RDC_\u0054EST": "1" }, "slow": true },
-  "assertions": { "exit_code": 0, "stdout_contains": ["New Model"] },
-  "acceptance": {
-    "output_contains": ["canonical", "registration"],
-    "output_not_contains": ["raw insert", "skip gate"]
-  },
-  "teardown": { "reset_branch": true }
-}
+{
+  "manifest_version": 1,
+  "skill": "rdc:new-model",
+  "description": "New-model registration begins with the canonical gate.",
+  "fixture": { "prompt": "rdc:new-model example-model", "env": { "RDC_\u0054EST": "1" }, "slow": true },
+  "assertions": { "exit_code": 0, "stdout_contains": ["New Model"] },
+  "acceptance": {
+    "output_contains": ["canonical", "registration"],
+    "output_not_contains": ["raw insert", "skip gate"]
+  },
+  "teardown": { "reset_branch": true }
+}

package/skills/tests/rdc-refactor.test.json

@@ -1,29 +1,29 @@
-{
-  "manifest_version": 1,
-  "skill": "rdc:refactor",
-  "description": "Refactor preserves evidence and isolates all dispatched writers.",
-  "fixture": {
-    "prompt": "rdc:refactor hook-scope --takeover \"corrective consolidation\"",
-    "env": { "RDC_\u0054EST": "1" },
-    "slow": true
-  },
-  "assertions": {
-    "exit_code": 0,
-    "commits_made": { "min": 0 },
-    "stdout_contains": ["RDC Refactor"]
-  },
-  "acceptance": {
-    "output_contains": [
-      "consolidate_work_items_into_epic",
-      "unique leased worktree",
-      "collaboration manifest",
-      "validator"
-    ],
-    "output_not_contains": [
-      "raw update",
-      "shared working directory",
-      "archive all"
-    ]
-  },
-  "teardown": { "reset_branch": true }
-}
+{
+  "manifest_version": 1,
+  "skill": "rdc:refactor",
+  "description": "Refactor preserves evidence and isolates all dispatched writers.",
+  "fixture": {
+    "prompt": "rdc:refactor hook-scope --takeover \"corrective consolidation\"",
+    "env": { "RDC_\u0054EST": "1" },
+    "slow": true
+  },
+  "assertions": {
+    "exit_code": 0,
+    "commits_made": { "min": 0 },
+    "stdout_contains": ["RDC Refactor"]
+  },
+  "acceptance": {
+    "output_contains": [
+      "consolidate_work_items_into_epic",
+      "unique leased worktree",
+      "collaboration manifest",
+      "validator"
+    ],
+    "output_not_contains": [
+      "raw update",
+      "shared working directory",
+      "archive all"
+    ]
+  },
+  "teardown": { "reset_branch": true }
+}

package/skills/tests/rdc-regen-media.test.json

@@ -1,29 +1,29 @@
-{
-  "manifest_version": 1,
-  "skill": "rdc:regen-media",
-  "description": "Regen Media uses local keyless Codex gpt-image-2 as the primary image-generation path and avoids live generation under RDC_TEST.",
-  "fixture": {
-    "prompt": "rdc:regen-media generate a 2048x1152 homepage hero image",
-    "env": { "RDC_TEST": "1" }
-  },
-  "assertions": {
-    "exit_code": 0,
-    "commits_made": { "min": 0 },
-    "stdout_contains": ["image_gen", "gpt-image-2"]
-  },
-  "acceptance": {
-    "output_contains": [
-      "image_gen",
-      "gpt-image-2",
-      "--enable image_generation",
-      "RDC_TEST"
-    ],
-    "output_not_contains": [
-      "OPENAI_API_KEY=",
-      "generate_gpt_image is the default",
-      "upload completed",
-      "raw MCP"
-    ]
-  },
-  "teardown": { "reset_branch": true }
-}
+{
+  "manifest_version": 1,
+  "skill": "rdc:regen-media",
+  "description": "Regen Media uses local keyless Codex gpt-image-2 as the primary image-generation path and avoids live generation under RDC_TEST.",
+  "fixture": {
+    "prompt": "rdc:regen-media generate a 2048x1152 homepage hero image",
+    "env": { "RDC_TEST": "1" }
+  },
+  "assertions": {
+    "exit_code": 0,
+    "commits_made": { "min": 0 },
+    "stdout_contains": ["image_gen", "gpt-image-2"]
+  },
+  "acceptance": {
+    "output_contains": [
+      "image_gen",
+      "gpt-image-2",
+      "--enable image_generation",
+      "RDC_TEST"
+    ],
+    "output_not_contains": [
+      "OPENAI_API_KEY=",
+      "generate_gpt_image is the default",
+      "upload completed",
+      "raw MCP"
+    ]
+  },
+  "teardown": { "reset_branch": true }
+}