@monoes/monomindcli 1.14.6 → 1.15.0

package/.claude/skills/agent-browser-testing/SKILL.md

@@ -41,6 +41,40 @@ tools:
 
 ---
 
+## Commands at a Glance
+
+All 55 subcommands — grouped by what they do.
+
+| Group | Commands |
+|---|---|
+| **Open / Navigate** | `open`, `navigate back\|forward\|reload`, `pushstate`, `connect`, `close` |
+| **Inspect** | `snapshot`, `get url\|title\|text\|html\|value\|attr\|count\|box\|styles`, `errors`, `console` |
+| **Find** | `find role\|text\|label\|placeholder\|testid\|alttext\|title\|selector` |
+| **Assert** | `is visible\|enabled\|checked`, `isvisible`, `isenabled`, `ischecked` |
+| **Click / Tap** | `click`, `dblclick`, `tap`, `hover`, `focus` |
+| **Type / Fill** | `fill`, `type`, `keyboard type\|inserttext`, `keydown`, `keyup`, `press` |
+| **Form** | `select`, `check`, `uncheck`, `upload` |
+| **Scroll / Drag** | `scroll`, `scrollintoview`, `drag`, `swipe up\|down\|left\|right` |
+| **Mouse (raw)** | `mouse move\|down\|up\|wheel` |
+| **Wait** | `wait --text\|--not-text\|--url\|--selector\|--load\|--ms\|--fn\|--download` |
+| **Download** | `download <ref> <path>`, `wait --download <path>` |
+| **Screenshot / PDF** | `screenshot`, `pdf`, `highlight`, `diff url` |
+| **Clipboard** | `clipboard read\|write\|copy\|paste` |
+| **Dialogs** | `dialog accept\|dismiss\|status` |
+| **Tabs** | `tab list\|new\|close\|<targetId>` |
+| **Windows** | `window new [url]` (isolated context / incognito) |
+| **Frames** | `frame <ref>\|main` |
+| **Storage** | `storage local\|session`, `cookies list\|set\|clear` |
+| **Network** | `network route\|unroute\|headers\|capture\|requests\|request` |
+| **Emulation** | `set device\|media\|geo\|offline\|useragent\|viewport\|credentials`, `resize` |
+| **Session** | `state save\|load\|list\|show\|clear\|rename\|clean`, `open --session\|--state` |
+| **JS / Init** | `eval`, `eval --stdin`, `addinitscript`, `removeinitscript` |
+| **Record** | `record start\|stop\|restart\|status` |
+| **Performance** | `vitals`, `trace start\|stop\|status`, `profiler start\|stop\|heap`, `har start\|stop\|status` |
+| **Batch** | `batch "cmd1" "cmd2" ...` |
+
+---
+
 ## QA Execution Protocol
 
 When this skill is invoked, run ALL phases unless the user specifies a subset.
@@ -277,8 +311,10 @@ npx monomind browse find role button --name "Submit"
 npx monomind browse find text "Delete"            # by visible text
 npx monomind browse find label "Email"            # by label text
 npx monomind browse find testid "submit-btn"      # by data-testid
+npx monomind browse find alttext "Company logo"   # by alt attribute (images)
+npx monomind browse find title "More options"     # by title attribute (tooltips)
 npx monomind browse find role link --nth 2        # nth match
-npx monomind browse find role image --last         # last match
+npx monomind browse find role image --last        # last match
 npx monomind browse find placeholder "Search..."  # by placeholder text
 npx monomind browse find selector "div.modal"     # by CSS selector
 
@@ -293,6 +329,28 @@ npx monomind browse is enabled @eN
 npx monomind browse is checked @eN
 ```
 
+### Window Management (Isolated Contexts)
+
+```bash
+# Open a new isolated browser window (like incognito — separate cookies/storage)
+npx monomind browse window new
+npx monomind browse window new https://app.example.com
+
+# Useful for multi-user testing (e.g. admin + regular user simultaneously):
+# 1. Connect once, then open isolated windows for each user role
+npx monomind browse connect --port 9222
+npx monomind browse window new                 # switches to new isolated context
+npx monomind browse open https://app.example.com/login
+npx monomind browse fill @e1 "admin@example.com"
+# ...
+
+# Tab management (within the same context)
+npx monomind browse tab                        # list all tabs
+npx monomind browse tab new https://example.com
+npx monomind browse tab close <tabId>
+npx monomind browse tab <tabId>               # switch to tab
+```
+
 ### Interaction
 
 ```bash
@@ -316,11 +374,15 @@ npx monomind browse keyup "Shift"
 npx monomind browse select @eN "Option A"        # select dropdown option
 npx monomind browse check @eN
 npx monomind browse uncheck @eN
-npx monomind browse scroll down --ref eN --amount 400   # scroll within element
+npx monomind browse scroll down 300                     # positional amount (shorthand)
 npx monomind browse scroll down --amount 600            # scroll page
+npx monomind browse scroll down --ref eN --amount 400   # scroll within element ref
+npx monomind browse scroll down --selector ".sidebar" 300  # scroll within CSS selector
 npx monomind browse scrollintoview @eN
 npx monomind browse drag @eN @eM                 # drag source to target
 npx monomind browse upload @eN ./file.pdf        # file input
+npx monomind browse download @eN ./report.pdf   # click element, capture file download
+npx monomind browse download ".export-btn" ./data.csv --timeout 15000
 npx monomind browse highlight @eN               # visual debug highlight
 ```
 
@@ -356,32 +418,126 @@ npx monomind browse wait --load domcontentloaded
 npx monomind browse wait --ms 500
 npx monomind browse wait --fn "window.__ready === true"
 npx monomind browse wait --text "Done" --timeout 10000
+npx monomind browse wait --download ./report.pdf    # wait for browser-triggered download to complete
+npx monomind browse wait --download ./data.csv --timeout 15000
+```
+
+### Element State Checks (Assertions)
+
+```bash
+npx monomind browse isvisible @e3          # true/false — is element visible?
+npx monomind browse isenabled @e2          # true/false — is element enabled (not disabled)?
+npx monomind browse ischecked @e1          # true/false — is checkbox/radio checked?
+
+# Also accepts CSS selectors
+npx monomind browse isvisible ".submit-btn"
+npx monomind browse isenabled "#email-input"
+npx monomind browse ischecked "input[name='agree']"
+
+# JSON output for scripting
+npx monomind browse isvisible @e4 --json   # → {"visible": true}
+```
+
+**Use in test scripts:**
+```bash
+# Run snapshot, then assert expected state
+npx monomind browse snapshot -i
+npx monomind browse click @e3             # click submit
+npx monomind browse wait --text "Success"
+npx monomind browse isvisible ".success-banner" --json
+```
+
+### Snapshot Regression Testing
+
+```bash
+# Save current snapshot as baseline
+npx monomind browse snapshot -i --save ./baselines/homepage.txt
+
+# Later: compare against baseline (colored diff: green=added, red=removed)
+npx monomind browse snapshot -i --diff ./baselines/homepage.txt
+# → Snapshot changed: +2 lines, -1 lines
+# + button "Sign Out" [@e12]
+# - button "Sign In" [@e8]
+
+# JSON output
+npx monomind browse snapshot -i --diff ./baselines/homepage.txt --json
+
+# Content-safety flags (for agentic pipelines with untrusted pages)
+npx monomind browse snapshot --content-boundaries          # wrap output in sentinel markers to prevent injection
+npx monomind browse snapshot --max-output 10000            # truncate to 10k chars (prevents context blowout)
+npx monomind browse snapshot --content-boundaries --max-output 8000  # both together
+
+# Compare two URLs side-by-side (navigates to each, diffs their snapshots)
+npx monomind browse diff url https://staging.example.com https://prod.example.com
+npx monomind browse diff url https://app.com/v1 https://app.com/v2 --interactive
+npx monomind browse diff url https://before.com https://after.com --json
+```
+
+### Mobile Testing (Touch Events)
+
+```bash
+# Tap with touch event — useful for mobile-emulated layouts
+npx monomind browse tap @e4
+npx monomind browse tap ".mobile-nav-toggle"
+
+# Swipe gestures — 10-step smooth touchMove sequence
+npx monomind browse swipe up                        # swipe up 300px from center
+npx monomind browse swipe down 500                  # swipe down 500px
+npx monomind browse swipe left --x 300 --y 400     # swipe left from (300,400)
+npx monomind browse swipe right 200 --x 50 --y 300 # swipe right 200px
+
+# Combine with device emulation
+npx monomind browse set device "iPhone 14"
+npx monomind browse snapshot -i
+npx monomind browse tap @e2
+npx monomind browse swipe up 400                    # scroll via swipe on mobile
 ```
 
 ### Screenshots & Visual Evidence
 
 ```bash
 npx monomind browse screenshot before.png
-npx monomind browse screenshot --full page.png   # full-page
+npx monomind browse screenshot --full page.png             # full-page
+npx monomind browse screenshot --annotate out.png          # overlay @eN numbered labels (requires prior snapshot -i)
+npx monomind browse screenshot --hide-scrollbars clean.png # hide native scrollbars via CSS injection
 npx monomind browse screenshot --format webp --quality 90 out.webp
-npx monomind browse screenshot --json            # returns path
+npx monomind browse screenshot --json                      # returns path as JSON
 npx monomind browse pdf ./output.pdf
 npx monomind browse pdf --landscape ./report.pdf
 ```
 
+**`--annotate` pattern:** always run `snapshot -i` first to populate refs, then `screenshot --annotate` to produce a screenshot with numbered labels at each element's position. Numbers match the `@eN` refs from the snapshot, so you can point to elements by number in bug reports.
+
+> **Note:** `--annotate` uses viewport-relative coordinates — do not combine with `--full`. On full-page captures, badges will be misaligned below the fold.
+
+```bash
+npx monomind browse snapshot -i
+npx monomind browse screenshot --annotate ./issue-001.png
+# → labels [1] [2] [3]... appear at each interactive element
+```
+
 ### Console, Errors & JavaScript
 
+> **Console capture is always-on.** As soon as you `open` or `connect`, monobrowse hooks `Runtime.consoleAPICalled`, `Log.entryAdded`, and `Runtime.exceptionThrown` automatically. No `capture start` needed — messages accumulate in memory and `console`/`errors` read from that buffer at any time.
+
 ```bash
-npx monomind browse console                      # all console messages
-npx monomind browse console --clear              # clear console history
-npx monomind browse console --errors-only        # only error-level messages
-npx monomind browse errors                       # JS exceptions only
-npx monomind browse errors --clear              # clear exception history
-npx monomind browse errors --json               # machine-readable output
+npx monomind browse console                      # all log/warn/error/info messages since connect
+npx monomind browse console --errors-only        # filter to error level only
+npx monomind browse console --clear              # reset the buffer
+npx monomind browse console --json               # machine-readable [{type, text, timestamp}]
+npx monomind browse errors                       # uncaught JS exceptions (Runtime.exceptionThrown)
+npx monomind browse errors --json                # [{text, url, lineNumber, columnNumber, timestamp}]
+npx monomind browse errors --clear
+
 npx monomind browse eval "document.title"
 npx monomind browse eval "document.querySelectorAll('button').length"
 npx monomind browse eval "window.__store.getState()" --json
-npx monomind browse addinitscript "window.__TEST__ = true"
+# Multiline JS via heredoc (--stdin reads from stdin)
+cat <<'EOF' | npx monomind browse eval --stdin
+  const items = document.querySelectorAll('li');
+  JSON.stringify([...items].map(i => i.textContent.trim()))
+EOF
+npx monomind browse addinitscript "window.__TEST__ = true"   # runs on every page load
 npx monomind browse removeinitscript <id>
 ```
 
@@ -402,6 +558,11 @@ npx monomind browse network capture start
 npx monomind browse network capture stop
 npx monomind browse network capture clear
 npx monomind browse network requests --json
+npx monomind browse network requests --filter "api/users"            # filter by URL substring
+npx monomind browse network requests --method POST                   # filter by HTTP method
+npx monomind browse network requests --status-code 200               # filter by status code
+npx monomind browse network requests --type xhr                      # filter by resource type
+npx monomind browse network request <requestId>                      # full detail for one request
 npx monomind browse network cookies             # cookies via network layer
 ```
 
@@ -465,12 +626,23 @@ npx monomind browse state load <name>
 npx monomind browse state list
 npx monomind browse state show        # inspect current active session
 npx monomind browse state clear       # clear current active session
+npx monomind browse state rename <old> <new>           # rename a saved session
+npx monomind browse state clean --older-than 7         # delete sessions older than N days (default: 7)
 ```
 
 ### Performance & Diagnostics
 
+> **DevTools panel coverage:**
+> ✅ Console panel — `console` / `errors` (always-on, see above)
+> ✅ Network panel — `network requests` + filters, `network request <id>`, `har start/stop --bodies`
+> ✅ Performance panel — `trace start/stop` (exports Chrome DevTools-compatible `.json`)
+> ✅ Memory panel — `profiler heap ./heap.json` (heap snapshot)
+> ✅ Application panel — `storage`, `cookies`
+> ✅ Sources / JS eval — `eval`, `eval --stdin`, `addinitscript`
+> ❌ Not available: source-map resolution for error stacks, live network waterfall, Lighthouse audit scores, Total Blocking Time metric
+
 ```bash
-# Core Web Vitals
+# Core Web Vitals (LCP < 2.5s PASS, 2.5–4s WARN, >4s FAIL)
 npx monomind browse vitals --wait 3000 --json
 
 # CPU profiler
@@ -480,11 +652,12 @@ npx monomind browse profiler stop --json
 # Heap snapshot
 npx monomind browse profiler heap ./heap.json
 
-# Chrome trace
+# Chrome trace (load in chrome://tracing or Perfetto)
 npx monomind browse trace start --screenshots
 npx monomind browse trace stop ./trace.json
+npx monomind browse trace status
 
-# HAR (full HTTP archive with bodies)
+# HAR (full HTTP archive with request/response bodies)
 npx monomind browse har start
 npx monomind browse har status
 npx monomind browse har stop --bodies --json   # --bodies only valid on stop
@@ -493,9 +666,7 @@ npx monomind browse har stop --bodies --json   # --bodies only valid on stop
 npx monomind browse record start --format jpeg --quality 80
 npx monomind browse record status
 npx monomind browse record stop --json
-
-# Trace status
-npx monomind browse trace status
+npx monomind browse record restart ./take2.mp4   # atomically stop+start (no page reload)
 ```
 
 ### Batch Execution
@@ -594,6 +765,67 @@ npx monomind browse snapshot -i           # should show error state
 
 ---
 
+## Dogfood / Exploratory QA Mode
+
+When asked to "dogfood", "QA", "find bugs", "exploratory test", or "bug hunt" a web app:
+
+### Setup
+
+```bash
+mkdir -p ./dogfood-output/screenshots ./dogfood-output/videos
+```
+
+### Workflow
+
+```
+1. Orient    → open URL, initial annotated screenshot, snapshot -i
+2. Explore   → visit each major section, test interactive elements
+3. Document  → screenshot + record for each issue found (as you go, not at end)
+4. Wrap up   → re-check console errors, write summary
+```
+
+### At each page
+
+```bash
+npx monomind browse snapshot -i                                              # find interactive elements
+npx monomind browse screenshot --annotate ./dogfood-output/screenshots/<page>.png  # annotated overview
+npx monomind browse errors                                                   # check for JS errors
+npx monomind browse console                                                  # check console output
+```
+
+### Documenting an issue (repro-first)
+
+For **interactive/behavioral bugs** (require clicking or typing to reproduce):
+
+```bash
+# 1. Start recording before reproducing
+npx monomind browse record start
+# 2. Walk through the steps (use type not fill during recording — shows keystrokes)
+npx monomind browse screenshot ./dogfood-output/screenshots/issue-001-step1.png
+npx monomind browse type @eN "input value"
+npx monomind browse click @eM
+npx monomind browse screenshot --annotate ./dogfood-output/screenshots/issue-001-result.png
+# 3. Stop recording
+npx monomind browse record stop --json   # returns path to frames JSON
+```
+
+For **static/visible-on-load bugs** (typos, layout, placeholder text):
+
+```bash
+npx monomind browse screenshot --annotate ./dogfood-output/screenshots/issue-001.png
+# Single annotated screenshot is sufficient — no recording needed
+```
+
+### `type` vs `fill` during recording
+
+Use `type` (character-by-character) when recording a video so keystrokes are visible in the recording. Use `fill` (instant clear + set) for speed outside recording.
+
+### Reporting
+
+Aim for 5–10 well-documented issues. Depth of evidence (annotated screenshots, step-by-step repro) beats quantity. Append each issue as you find it — never batch for the end.
+
+---
+
 ## Task Walkthrough Mode
 
 When helping a user accomplish a task in a live UI:
@@ -669,10 +901,61 @@ npx monomind memory search --query "login flow" --namespace ui-testing
 
 ---
 
+## Electron App Automation
+
+Automate any Electron desktop app (VS Code, Slack, Discord, Figma, Notion, Spotify) via CDP — same snapshot-interact workflow as for web pages.
+
+### Launch with CDP port
+
+```bash
+# macOS — quit the app first if already running
+open -a "Slack"   --args --remote-debugging-port=9222
+open -a "Visual Studio Code" --args --remote-debugging-port=9223
+open -a "Discord" --args --remote-debugging-port=9224
+open -a "Figma"   --args --remote-debugging-port=9225
+open -a "Notion"  --args --remote-debugging-port=9226
+
+# Linux
+slack --remote-debugging-port=9222
+code  --remote-debugging-port=9223
+
+# Windows
+"C:\...\slack.exe" --remote-debugging-port=9222
+```
+
+### Connect and interact
+
+```bash
+# Connect to a specific port
+npx monomind browse connect --port 9222
+
+# Auto-discover any running Chromium-based app on ports 9222 or 9229
+npx monomind browse connect --auto-connect
+
+# Standard workflow from here
+npx monomind browse snapshot -i
+npx monomind browse click @e5
+npx monomind browse screenshot --annotate slack-desktop.png
+```
+
+`--auto-connect` probes ports 9222 and 9229 in order and connects to the first responding Chrome/Electron instance.
+
+### Tab and webview management
+
+Electron apps often have multiple windows or embedded webviews:
+
+```bash
+npx monomind browse tab                  # list all tabs/webviews (shows targetId for each)
+npx monomind browse tab <targetId>       # switch to tab by its targetId
+npx monomind browse tab new              # open a new tab
+```
+
+---
+
 ## Activation Checklist
 
 When this skill fires:
-- [ ] Get the URL (ask if not provided)
+- [ ] Get the URL (ask if not provided — or use `--auto-connect` for Electron apps)
 - [ ] Run `open` + `errors` baseline
 - [ ] Run all 7 QA phases (or subset the user requested)
 - [ ] Screenshot key states: initial, post-action, errors, mobile

package/.claude/skills/mastermind/runorg.md

@@ -113,7 +113,8 @@ curl -s -X POST "${CTRL_URL}/api/mastermind/event" \
     --arg org "$orgName" \
     --arg goal "$goal" \
     --arg proj "$REPO_ROOT" \
-    '{type:"org:start",session:$session,org:$org,goal:$goal,project:$proj,ts:(now*1000|floor)}')" || true
+    --argjson ci "${CHECKPOINT_INTERVAL_MS:-600000}" \
+    '{type:"org:start",session:$session,org:$org,goal:$goal,project:$proj,checkpointInterval:$ci,ts:(now*1000|floor)}')" || true
 ```
 
 ### 1.6.4 — Execute first iteration inline
@@ -147,12 +148,22 @@ If `LOOP_STATUS == "active"` or `LOOP_STATUS == "paused"`:
    LOOP_PROMPT=$(cat "$run_prompt_file")
    ```
 
-2. Call `ScheduleWakeup` with:
+2. Emit `org:stop` to signal this iteration is pausing:
+   ```bash
+   curl -s -X POST "${CTRL_URL}/api/mastermind/event" \
+     -H "Content-Type: application/json" \
+     -d "$(jq -cn \
+       --arg org "$orgName" \
+       --arg runId "run-$(date -u +%Y%m%dT%H%M%S)" \
+       '{type:"org:stop",org:$org,runId:$runId,ts:(now*1000|floor),reason:"scheduled-iteration-complete"}')" || true
+   ```
+
+3. Call `ScheduleWakeup` with:
    - `delaySeconds`: `$poll_interval_seconds`
    - `reason`: `"<orgName>: next scheduled poll (every <poll_interval_minutes> min)"`
    - `prompt`: full contents of `$LOOP_PROMPT`
 
-3. Update `next_run` in org JSON:
+4. Update `next_run` in org JSON:
    ```bash
    next_ts=$(( $(date +%s) + poll_interval_seconds ))
    next_iso=$(date -u -r "$next_ts" +%Y-%m-%dT%H:%M:%SZ 2>/dev/null \
@@ -305,33 +316,37 @@ jq -cn \
   '{type:"run:start",runId:$runId,org:$org,orgName:$org,goal:$goal,bossRole:$bossRole,project:$proj,ts:(now*1000|floor)}' \
   >> "$runFile"
 
-# Print resolved values so Step 4 can embed them in the boss prompt
-echo "ORG_VARS: orgName=${orgName} runId=${runId} sessionId=${sessionId} goal=${goal} bossRole_id=${bossRole_id} bossRole_title=${bossRole_title} bossRole_agent_type=${bossRole_agent_type} board_id=${board_id} todo_col=${todo_col} doing_col=${doing_col} done_col=${done_col} checkpointMin=${checkpointMin} memNs=${memNs} CTRL_URL=${CTRL_URL} MONO_DIR=${MONO_DIR} runFile=${runFile} REPO_ROOT=${REPO_ROOT}"
-```
-
-After running the script above, read the `ORG_VARS:` line from its output and hold those values for Step 4 (embed them into the boss prompt as literals — **do not re-run bash to look them up**).
-
----
-
-## Step 3 — Emit org:start (separate Bash call, re-derives from output)
-
-After the combined Step 2+3 script runs, emit `org:start` using the values printed by `ORG_VARS:`. Replace each `<X>` with the literal value from that line:
+# Emit session:start first — creates the chat tab session record and writes active-session.json
+# so capture-handler can tag all subsequent agent:spawn / agent:complete events with this session
+curl -s -X POST "${CTRL_URL}/api/mastermind/event" \
+  -H "Content-Type: application/json" \
+  -d "$(jq -cn \
+    --arg session "$sessionId" \
+    --arg org "$orgName" \
+    --arg prompt "$goal" \
+    '{type:"session:start",session:$session,org:$org,prompt:$prompt,ts:(now*1000|floor)}')" || true
 
-```bash
-curl -s -X POST "<CTRL_URL>/api/mastermind/event" \
+# Emit org:start — must run here while CTRL_URL/runId/orgName are in scope
+curl -s -X POST "${CTRL_URL}/api/mastermind/event" \
   -H "Content-Type: application/json" \
   -d "$(jq -cn \
-    --arg session "<sessionId>" \
-    --arg org "<orgName>" \
-    --arg runId "<runId>" \
-    --arg goal "<goal>" \
-    --arg proj "<REPO_ROOT>" \
-    '{type:"org:start",session:$session,org:$org,runId:$runId,goal:$goal,project:$proj,ts:(now*1000|floor)}')" || true
+    --arg session "$sessionId" \
+    --arg org "$orgName" \
+    --arg runId "$runId" \
+    --arg goal "$goal" \
+    --arg proj "$REPO_ROOT" \
+    --argjson ci "${CHECKPOINT_INTERVAL_MS:-600000}" \
+    '{type:"org:start",session:$session,org:$org,runId:$runId,goal:$goal,project:$proj,checkpointInterval:$ci,ts:(now*1000|floor)}')" || true
+
+# Print resolved values so Step 3 can embed them in the boss prompt
+echo "ORG_VARS: orgName=${orgName} runId=${runId} sessionId=${sessionId} goal=${goal} bossRole_id=${bossRole_id} bossRole_title=${bossRole_title} bossRole_agent_type=${bossRole_agent_type} board_id=${board_id} todo_col=${todo_col} doing_col=${doing_col} done_col=${done_col} checkpointMin=${checkpointMin} memNs=${memNs} CTRL_URL=${CTRL_URL} MONO_DIR=${MONO_DIR} runFile=${runFile} REPO_ROOT=${REPO_ROOT}"
 ```
 
+After running the script above, read the `ORG_VARS:` line from its output and hold those values for Step 3 (embed them into the boss prompt as literals — **do not re-run bash to look them up**).
+
 ---
 
-## Step 4 — Spawn Boss Agent
+## Step 3 — Spawn Boss Agent
 
 Using the literal values from the `ORG_VARS:` output (not shell variables — they don't persist), spawn the boss agent via Task tool:
 
@@ -508,6 +523,11 @@ OPERATING LOOP:
             -d "$(jq -cn --arg s "${sessionId}" --arg o "${orgName}" --arg rid "${runId}" \
               --arg from "<role_id>" --arg msg "Completed: <one-sentence summary of output>" \
               '{type:"org:comms",session:$s,org:$o,runId:$rid,from:$from,to:"boss",msg:$msg,ts:(now*1000|floor)}')" || true
+          # When your task is fully complete — signal offline:
+          curl -s -X POST "${CTRL_URL}/api/mastermind/event" -H "Content-Type: application/json" \
+            -d "$(jq -cn --arg s "${sessionId}" --arg o "${orgName}" --arg rid "${runId}" \
+              --arg from "<role_id>" \
+              '{type:"org:agent:offline",session:$s,org:$o,runId:$rid,from:$from,ts:(now*1000|floor)}')" || true
           # TOKEN TRACKING — estimate and persist token usage to state.json for the Budgets dashboard:
           # Count approximate output tokens from your work (1 word ≈ 1.3 tokens):
           _tok_out=$(echo -n "<paste a representative sample of your final output here>" | wc -w)
@@ -636,3 +656,29 @@ run_id: "<runId>"
 ## Step 8 — Brain Write (standalone only)
 
 If `caller` is not "command", follow _protocol.md Brain Write Procedure for domain `ops`.
+
+---
+
+## Reporting Artifacts (optional)
+
+When you create or modify a file that represents output of this org run, emit:
+
+```bash
+curl -s -X POST "${CTRL_URL}/api/mastermind/event" \
+  -H "Content-Type: application/json" \
+  -d "$(jq -cn \
+    --arg s "${sessionId}" \
+    --arg o "${orgName}" \
+    --arg rid "${runId}" \
+    --arg from "${AGENT_ROLE}" \
+    --arg label "Short description" \
+    --arg path "${FILE_PATH}" \
+    --arg mime "text/markdown" \
+    '{type:"org:artifact",session:$s,org:$o,runId:$rid,from:$from,artifact:{label:$label,type:"file",path:$path,mimeType:$mime},ts:(now*1000|floor)}')" || true
+```
+
+The dashboard chat will show this as an artifact card with a "View" button. Fill in:
+- `${AGENT_ROLE}` — the literal role id of the agent emitting the artifact
+- `${FILE_PATH}` — the absolute or repo-relative path to the created/modified file
+- `label` — a short human-readable description (e.g., "Weekly report", "Analysis output")
+- `mimeType` — `text/markdown`, `application/json`, `text/plain`, etc.

package/.claude/skills/monodesign/SKILL.md

@@ -160,13 +160,18 @@ If someone could look at this interface and say "AI made that" without doubt, it
 
 ## Antipattern Detection
 
-Before finalizing any design, run the antipattern detector. It checks for 27 known design antipatterns across two categories:
+Before finalizing any design, run the antipattern detector. It checks for 46 known design antipatterns across two categories:
 
 - **slop** — AI tells that signal lack of intentional design (purple palettes, side-tabs, identical card grids, italic-serif heroes, overused fonts)
 - **quality** — Design principle violations (spacing, hierarchy, readability, contrast, motion, typography)
 
 To run detection on HTML/CSS files:
 ```bash
+monomind design detect <file-or-dir>
+```
+
+Or use `npx impeccable detect` directly as an alternative:
+```bash
 npx impeccable detect <file-or-dir>
 ```
 
@@ -177,8 +182,31 @@ npm install -g impeccable
 npx impeccable@latest detect <file-or-dir>
 ```
 
+For machine-readable output (e.g. CI pipelines):
+```bash
+monomind design detect --json <file-or-dir>
+```
+
 For the full antipattern catalog with detection rules and remediation, see [reference/antipatterns-catalog.md](reference/antipatterns-catalog.md).
 
+## OKLCH Palette Seed
+
+Before composing a brand palette, seed the anchor color from the curated OKLCH library (129 hand-curated seeds):
+
+```bash
+monomind design palette                              # random seed (hue-zone weighted)
+monomind design palette --from "my-product-name"    # deterministic seed from product name
+monomind design palette --id seed-021               # specific seed by ID
+monomind design palette --json                      # machine-readable output
+```
+
+Each seed returns:
+- **Anchor color** — the single OKLCH value your primary brand color derives from
+- **Mood** — one evocative phrase describing the register this hue inhabits
+- **Strategy** — a concrete composition hint (bg choice, accent pairing, surface treatment)
+
+The seed is the start, not the recipe. Compose the full 5-role palette (bg, surface, ink, accent, muted) using your brief and the color-strategy guidance in this skill. The same seed can produce a dark-mode jazz club or a light-mode clinical brand depending on what the brief calls for.
+
 ## UX Quality Gate
 
 Antipatterns catch aesthetic and visual issues. For implementation-level UX quality — accessibility, touch interaction, form feedback, navigation patterns, animation timing, performance — load:
@@ -290,6 +318,9 @@ Shadows max at 0.15 alpha. No heavy drop shadows.
 | `components [target]` | Build | Build component systems, token foundations, theme toggles | [reference/component-system.md](reference/component-system.md) |
 | `research [target]` | Evaluate | UX research, usability testing, personas, journey mapping | [reference/ux-research.md](reference/ux-research.md) |
 | `images [target]` | Build | Generate design images, hero photography, inclusive prompts | [reference/image-prompts.md](reference/image-prompts.md) |
+| `hooks [action]` | Config | Manage the impeccable design detector hook per project | [reference/hooks.md](reference/hooks.md) |
+| `detect [file-or-dir]` | Evaluate | Run design anti-pattern detection on HTML/CSS files | (inline) |
+| `palette [--from name]` | Build | Seed a brand palette from the OKLCH library (129 seeds) | (inline) |
 
 ### Routing rules