@odla-ai/cli 0.11.1 → 0.11.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@odla-ai/cli",
-  "version": "0.11.1",
+  "version": "0.11.2",
   "description": "Agent-operable CLI for odla provisioning, calendar consent and sync lifecycle, System AI administration, Worker secrets, security jobs, and smoke checks.",
   "license": "MIT",
   "homepage": "https://odla.ai/docs/packages/cli",

package/skills/odla/SKILL.md

@@ -55,6 +55,28 @@ State which path you're taking and what you'll build in one line; get a nod.
    store a Google authorization code/refresh token. Open only the state-bound
    platform/Google consent URL returned by odla; calendar actions remain unavailable in the read-only slice.
 
+## Tooling sources
+
+- **odla**: this skill plus the installed `@odla-ai/*` packages are the
+  complete, offline contract (README + exported declarations/JSDoc, rendered
+  at `https://odla.ai/docs/packages/<pkg>`). Never reconstruct odla behavior
+  from the web or training memory.
+- **Third-party tools (wrangler, Clerk) — the reverse**: their CLIs and config
+  formats evolve, so never work from memorized setup steps. When you reach a
+  wrangler or Clerk step, fetch the vendor's current agent docs and follow
+  those:
+  - Cloudflare: docs index `https://developers.cloudflare.com/llms.txt`; every
+    docs page has a markdown twin at `<page>/index.md` (wrangler config:
+    `https://developers.cloudflare.com/workers/wrangler/configuration/index.md`).
+    With the human's OK (it edits global agent config), you may also follow
+    `https://developers.cloudflare.com/agent-setup/prompt.md` to add
+    Cloudflare's own skills + MCP servers to your harness.
+  - Clerk: agent-first CLI setup `https://clerk.com/docs/cli.md`; docs index
+    `https://clerk.com/docs/llms.txt`.
+
+  Offline? Say so, continue with the vendor steps written in these references,
+  and flag that they may be stale.
+
 ## The flow
 
 You reached this skill because the human ran `npx @odla-ai/cli setup`. Then:

package/skills/odla/references/build.md

@@ -25,15 +25,19 @@ For static Appointment Schedule parity, add the public per-env
 ## 2. Build the Worker shell
 
 `npm i` only the SDKs you need and use the local `references/sdks.md` as the
-integration map. Create the Worker entrypoint and `wrangler.jsonc` before any
-deployed-secret transfer. With o11y, install `@odla-ai/o11y`, add
-`"nodejs_compat"` to `compatibility_flags`, and wrap the entrypoint with
-`withObservability`.
+integration map. `npm i -D wrangler` so the project pins its own copy instead
+of relying on an implicit `npx` fetch. Create the Worker entrypoint and
+`wrangler.jsonc` before any deployed-secret transfer — take the current config
+syntax from Cloudflare's live docs (SKILL.md "Tooling sources"), not memory.
+With o11y, install `@odla-ai/o11y`, add `"nodejs_compat"` to
+`compatibility_flags`, and wrap the entrypoint with `withObservability`.
 
 Each installed package's README and exported TypeScript declarations/JSDoc are
 the version-matched API contract; resolve public entry points through its
-`package.json` `exports`. The npm artifacts contain everything required for
-this flow; private service source or online documentation is not setup context.
+`package.json` `exports`. For everything odla, the npm artifacts contain the
+whole setup context — private service source or web documentation is never
+required. Third-party tooling (wrangler, Clerk) is the reverse: use the live
+vendor docs, never memorized steps.
 
 ## 3. Provision  ⏸ device-code approval
 

package/skills/odla-migrate/SKILL.md

@@ -124,3 +124,20 @@ Read only the current phase file as directed above. After `npm install`, prefer
 each installed package's README and exported TypeScript declarations/JSDoc over
 training memory; resolve entry points through its `package.json` `exports`.
 Those artifacts describe the version this project actually uses.
+
+Third-party tools (wrangler, Clerk) are the reverse: their CLIs and config
+formats evolve, so never work from memorized setup steps. When a phase reaches
+a wrangler or Clerk step, fetch the vendor's current agent docs and prefer
+their syntax over any literal command written in a phase file:
+
+- Cloudflare: docs index `https://developers.cloudflare.com/llms.txt`; every
+  docs page has a markdown twin at `<page>/index.md` (wrangler config:
+  `https://developers.cloudflare.com/workers/wrangler/configuration/index.md`).
+  With the human's OK (it edits global agent config), you may also follow
+  `https://developers.cloudflare.com/agent-setup/prompt.md` to add Cloudflare's
+  own skills + MCP servers to your harness.
+- Clerk: agent-first CLI setup `https://clerk.com/docs/cli.md`; docs index
+  `https://clerk.com/docs/llms.txt`.
+
+Offline? Say so, continue with the vendor steps written in the phase files,
+and flag that they may be stale.

package/skills/odla-migrate/references/phase-1-static.md

@@ -10,7 +10,11 @@ in this session).
 ## Steps
 
 1. `npm i -D wrangler`
-2. Write `wrangler.jsonc` by hand using this documented shape:
+2. Write `wrangler.jsonc` by hand. The shape below is the odla-shaped
+   baseline; verify field names and current syntax against Cloudflare's live
+   doc first
+   (`https://developers.cloudflare.com/workers/wrangler/configuration/index.md`
+   — see SKILL.md "Context bootstrap"):
    - `name`: kebab-case app id; `main`: `src/worker.ts`
    - `compatibility_date` (today), `compatibility_flags: ["nodejs_compat"]`
    - `assets: { "directory": "<buildDir>", "binding": "ASSETS",

package/skills/odla-migrate/references/phase-3-auth.md

@@ -17,6 +17,11 @@ dashboard. Do NOT assume an app exists: a stale publishable key can linger
 on the registry's app_auth record from a deleted/recreated app —
 `clerk apps list` is the source of truth for what's real.
 
+Clerk's CLI evolves. Before the first `npx clerk` command, fetch the current
+doc — `https://clerk.com/docs/cli.md` — and prefer its syntax over the literal
+flags below. The steps here are the odla-shaped sequence; the live doc is
+authoritative for commands and flags.
+
 ## Steps
 
 1. Provision the Clerk app with the CLI (after the human's `clerk auth