impulso 0.19.1

package/skills/plannotator-annotate/SKILL.md

@@ -0,0 +1,45 @@
+---
+name: plannotator-annotate
+description:
+  Open Plannotator's annotation UI for a markdown file, HTML file, URL, or
+  folder and then respond to the returned annotations.
+disable-model-invocation: true
+---
+
+# Plannotator Annotate
+
+Use when user wants document annotated in Plannotator.
+
+## Protocol
+
+1. Set `TARGET` to the file/URL/folder to annotate. Never empty.
+
+2. Launch plannotator and get the URL:
+
+   ```bash
+   impulso-hub plannotator launch annotate <TARGET>
+   ```
+
+   This returns a JSON line with `url`, `remoteUrl`, `port`, and `pid`.
+   Plannotator runs in the background.
+
+3. **IMMEDIATELY relay URL:** "Open Plannotator at `<URL>` (remote: `<Remote>`).
+   Annotate `<TARGET>`, then tell me when done." WAIT for user to confirm.
+
+4. After user confirms, wait for plannotator to finish and get the decision:
+
+   ```bash
+   impulso-hub plannotator wait
+   ```
+
+   Returns JSON:
+   `{"decision":"approved"|"dismissed"|"annotated","feedback":"..."}`.
+
+5. Apply the decision:
+   - `approved` → acknowledge, proceed.
+   - `dismissed` → session closed, say so, continue.
+   - `annotated` → address the feedback text.
+
+6. Launch timeout (impulso-hub exits 2): check stderr, ask whether to retry.
+
+Run commands yourself. Never ask user to paste shell commands into chat.

package/skills/plannotator-last/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: plannotator-last
+description:
+  Open Plannotator on the latest rendered assistant message and use the returned
+  annotations to revise that message or continue.
+disable-model-invocation: true
+---
+
+# Plannotator Last
+
+Use when user wants latest assistant response annotated in Plannotator.
+
+No commentary before launch — preamble becomes the thing annotated.
+
+## Protocol
+
+1. Launch plannotator immediately (no preamble):
+
+   ```bash
+   impulso-hub plannotator launch last
+   ```
+
+   Returns JSON with `url`, `remoteUrl`, `port`, `pid`. Non-blocking.
+
+2. **IMMEDIATELY relay URL:** "Open Plannotator at `<URL>` (remote: `<Remote>`).
+   Annotate the last message, then tell me when done." WAIT for user.
+
+3. After user confirms:
+
+   ```bash
+   impulso-hub plannotator wait
+   ```
+
+   Returns `{"decision":"approved"|"dismissed"|"annotated","feedback":"..."}`.
+
+4. Apply the decision:
+   - `approved` → acknowledge, proceed.
+   - `dismissed` → session closed, say so, continue.
+   - `annotated` → address the feedback text.
+
+5. Launch timeout (exit 2): check stderr, ask whether to retry.
+
+Run commands yourself. Never ask user to paste shell commands into chat.

package/skills/plannotator-review/SKILL.md

@@ -0,0 +1,44 @@
+---
+name: plannotator-review
+description:
+  Open Plannotator's browser-based code review UI for the current worktree or a
+  pull request URL, then act on the feedback that comes back.
+disable-model-invocation: true
+---
+
+# Plannotator Review
+
+Use when user wants code changes reviewed in Plannotator.
+
+## Protocol
+
+1. Determine the review target. User gave PR URL → set `TARGET` to the PR URL.
+   Worktree review → leave `TARGET` empty.
+
+2. Launch plannotator and get the URL:
+
+   ```bash
+   impulso-hub plannotator launch review [TARGET]
+   ```
+
+   Returns JSON with `url`, `remoteUrl`, `port`, `pid`. Non-blocking.
+
+3. **IMMEDIATELY relay URL:** "Open Plannotator at `<URL>` (remote: `<Remote>`).
+   Review the changes, then tell me when done." WAIT for user to confirm.
+
+4. After user confirms, wait for plannotator to finish:
+
+   ```bash
+   impulso-hub plannotator wait
+   ```
+
+   Returns `{"decision":"approved"|"dismissed"|"annotated","feedback":"..."}`.
+
+5. Apply the decision:
+   - `approved` → acknowledge, proceed.
+   - `dismissed` → session closed, say so, continue.
+   - `annotated` → address the feedback text.
+
+6. Launch timeout (exit 2): check stderr, ask whether to retry.
+
+Run commands yourself. Never ask user to paste shell commands into chat.