Ops Pocket logo

Ops Pocket

Organization
Lifecycle-Innovations-Limited
ops-pocket

OPS on-demand: This skill should be used when the user asks to "pocket memos", "voice memo pipeline"…

Overview

PublisherLifecycle-Innovations-Limited
Repositoryclaude-ops
Skill nameops-pocket
Stars
543
Forks
64
Bundled files
5
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 5 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Lifecycle-Innovations-Limited on GitHub. Read the source before you install it.

Installation

Install the Ops Pocket AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Lifecycle-Innovations-Limited/claude-ops/tree/main/claude-ops/skills/ops-pocket
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Lifecycle-Innovations-Limited/claude-ops.git /tmp/claude-ops
mkdir -p .claude/skills
cp -r /tmp/claude-ops/claude-ops/skills/ops-pocket .claude/skills/ops-pocket
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ops Pocket in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Ops Pocket on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Ops Pocket is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

OPS ► POCKET

Load ops-rules before acting. Public repo (no personal data). Outbound: one draft → one approval → one send. If AskUserQuestion / Workflow are missing, follow Rule 10 in ops-rules (Hermes: numbered options / two-turn Telegram card; delegate_task).

Surfaces the Pocket voice-memo pipeline as a discoverable slash command. The pipeline was previously orphaned — 7 scripts, no entry point. This skill is the entry point.

For the full architecture reference (every state file, every script, every failure mode) see STATUS.md sibling to this file.

Runtime Context

Before any route runs, resolve:

  1. State dir: ${POCKET_STATE_DIR:-$HOME/.claude/state/pocket}. Every file mentioned below is relative to this dir.
  2. Plugin root: ${CLAUDE_PLUGIN_ROOT} — used to invoke pipeline scripts under scripts/.
  3. Daemon health snapshot: ${CLAUDE_PLUGIN_DATA_DIR:-$HOME/.claude/plugins/data/ops-ops-marketplace}/daemon-health.json — surface its pocket-* service status if present.

If $POCKET_STATE_DIR does not exist on disk, the pipeline has never been bootstrapped — route to setup automatically (no autoroute on status if the user explicitly asked for status).

Routing table

Parse $ARGUMENTS and route immediately. First token decides the route.

First argRoute
(empty) / statusStatus dashboard — daemon health + queue depths + cursors + counts
setupDelegate to skills/setup/channels/pocket.md if present; else inline steps
tasksPending + in-progress + recently-completed task list
testEnd-to-end synthetic completion test (covered below)
logs [N]Tail N lines (default 50) of notifier + run + executor logs
restartKickstart launchd notifier + (re)spawn pocket-exec tmux supervisor
whatsapp on / offToggle WhatsApp self-chat sink
email on / offToggle email self-mail sink

Anything else: print the routing table and exit.


Route — status (default)

The default route. Read state files, render a compact dashboard.

Read these — in parallel where possible:

bash
STATE_DIR="${POCKET_STATE_DIR:-$HOME/.claude/state/pocket}"

# Daemon plist liveness
launchctl list com.claude-ops.pocket-activity-notifier 2>/dev/null \
  | awk '/PID/{print "pid="$3} /LastExitStatus/{print "exit="$3}'

# Supervisor tmux session
tmux has-session -t pocket-exec 2>/dev/null && echo "tmux=up" || echo "tmux=down"
tmux list-windows -t pocket-exec 2>/dev/null | wc -l

# Queue depths (line counts)
wc -l < "$STATE_DIR/pending-triage.jsonl" 2>/dev/null
wc -l < "$STATE_DIR/tasks.jsonl" 2>/dev/null
wc -l < "$STATE_DIR/spawn-ledger.jsonl" 2>/dev/null
wc -l < "$STATE_DIR/supervisor-out-queue.jsonl" 2>/dev/null

# Completed reports
ls "$STATE_DIR/executor-results/"*.done.json 2>/dev/null | wc -l

# Last pull cursor
cat "$STATE_DIR/cursor.txt" 2>/dev/null

# Last 3 notifications (from out-queue-sent.jsonl)
tail -n 3 "$STATE_DIR/out-queue-sent.jsonl" 2>/dev/null

# Per-component health
for h in .activity-notifier-health .out-queue-health .email-bridge-health .health; do
  cat "$STATE_DIR/$h" 2>/dev/null
done

Output format (desktop):

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 OPS ► POCKET — [timestamp]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

DAEMONS
  watcher          [running|stalled]    last tick [Xs ago]
  triage           [running|idle]
  executor watchdog [running|stalled]   tmux pocket-exec=[up|down] (N windows)
  activity-notifier [running|stalled]   pid [N]
  out-queue        [running|disabled|error]
  email-bridge     [running|disabled|error]

QUEUES
  pending-triage   [N items]            (awaiting ACT/SKIP decision)
  tasks.jsonl      [N items]            (canonical task store)
  spawn-ledger     [N spawns]
  out-queue        [N pending]
  executor-results [N .done.json]

CURSORS
  watcher last pull   [ISO timestamp from cursor.txt]
  last completion     [latest .done.json mtime]

CHANNELS
  whatsapp         [on|off|misconfigured]   chat_jid=[masked tail]
  email            [on|off|misconfigured]   to=[masked]

LAST 3 NOTIFICATIONS
  [N. kind type | task_id | sent_at]
──────────────────────────────────────────────────────

Mobile mode (Rule 7 — $SSH_CONNECTION/$SSH_CLIENT/$SSH_TTY set OR $OPS_MOBILE=1 OR $COLUMNS<80): plain text lines, no boxes, no emoji prefixes:

pocket: watcher running, exec tmux=up, notifier pid=N
queues: triage 0, tasks 12 (3 active), out 0
last pull: 5m ago. last done: 12m ago.
channels: whatsapp on, email off.

After the dashboard, use AskUserQuestion (max 4 — Rule 1) ONLY if there is something actionable (stalled daemon, pending-triage > 0, etc.). On a clean board, exit silently.


Route — setup

Delegate to the setup channel doc when it exists:

bash
SETUP_DOC="$CLAUDE_PLUGIN_ROOT/skills/setup/channels/pocket.md"
[[ -f "$SETUP_DOC" ]] && cat "$SETUP_DOC"

If pocket.md does not exist, print the inline onboarding sequence:

  1. Confirm pocket-cli (or equivalent Pocket API auth) is on PATH.
  2. Drop a whatsapp-config.json and/or email-config.json under $STATE_DIR (see STATUS.md for schemas).
  3. Run bash $CLAUDE_PLUGIN_ROOT/scripts/install-pocket-notifier.sh to register the launchd LaunchAgent.
  4. Verify with /ops:pocket status.

Per Rule 4 (background by default during setup) and Rule 2 (never delegate commands to the user) — run the install script via Bash with run_in_background: true. Per Rule 3, never auto-skip channel selection — if whatsapp-config.json and email-config.json are both missing, ask via AskUserQuestion which to configure first ([WhatsApp] / [Email] / [Both] / [Skip]).


Route — tasks

Render a single combined list. No AskUserQuestion — read-only by default.

bash
STATE_DIR="${POCKET_STATE_DIR:-$HOME/.claude/state/pocket}"

# Pending triage (awaiting ACT/SKIP)
tail -n 50 "$STATE_DIR/pending-triage.jsonl" 2>/dev/null

# Live tasks (canonical store)
tail -n 50 "$STATE_DIR/tasks.jsonl" 2>/dev/null

# In-flight workers (windows in pocket-exec tmux, excluding _idle + supervisor)
tmux list-windows -t pocket-exec -F '#{window_name}' 2>/dev/null \
  | grep -vE '^(_idle|supervisor)$'

# Recently completed (.done.json sorted by mtime, last 10)
ls -t "$STATE_DIR/executor-results/"*.done.json 2>/dev/null | head -10

Format:

PENDING TRIAGE (N)
  [id] [created] [title]

ACTIVE TASKS (N)
  [task_id] [status] [title] [started]

IN-FLIGHT WORKERS (N)
  [tmux window name] [up since]

RECENTLY COMPLETED (last 10)
  [task_id] [duration] [report file]

Route — test (synthetic end-to-end)

Smoke-tests the full notifier → out-queue → both bridges path without going through the watcher or triage. Same flow as the manual session-level test that was run when the notifier shipped.

Steps:

  1. Pre-flight: verify $STATE_DIR/whatsapp-config.json OR email-config.json has enabled: true. If neither: print not configured — run /ops:pocket setup and exit.
  2. Generate a synthetic task_id like test-$(date +%s).
  3. Append a tasks.jsonl entry for the synthetic task (so the notifier can title-resolve it):
    bash
    echo "{\"task_id\":\"$TID\",\"title\":\"synthetic pocket test\",\"created_at\":\"$(date -u +%FT%TZ)\",\"verdict\":\"ACT\"}" \
      >> "$STATE_DIR/tasks.jsonl"
  4. Drop a synthetic .done.json into executor-results/:
    bash
    cat > "$STATE_DIR/executor-results/$TID.done.json" <<EOF
    {"task_id":"$TID","status":"done","completed_at":"$(date -u +%FT%TZ)","report_file":"$TID.out.md"}
    EOF
    echo "# Synthetic test report\n\nThis is a test artifact from /ops:pocket test." \
      > "$STATE_DIR/executor-results/$TID.out.md"
  5. Manually invoke the notifier once (don't wait 60s for launchd):
    bash
    "$CLAUDE_PLUGIN_ROOT/scripts/ops-pocket-activity-notifier.py" --once 2>&1 | tail -20
    (If --once is not supported, run the script with POCKET_NOTIFIER_ONESHOT=1 env var, or just python3 ops-pocket-activity-notifier.py.)
  6. Drain both bridges in parallel:
    bash
    "$CLAUDE_PLUGIN_ROOT/scripts/ops-pocket-out-queue.py" 2>&1 | tail -10
    "$CLAUDE_PLUGIN_ROOT/scripts/ops-pocket-email-bridge.py" 2>&1 | tail -10
  7. Verify delivery — for each enabled channel:
    • WhatsApp: tail $STATE_DIR/out-queue-sent.jsonl for an entry with the synthetic task_id. (Do NOT use mcp__whatsapp__list_messages — the bridge's own sent ledger is the source of truth.)
    • Email: tail $STATE_DIR/email-bridge.log for SENT on the synthetic task_id.
  8. Cleanup:
    bash
    rm -f "$STATE_DIR/executor-results/$TID.done.json" \
          "$STATE_DIR/executor-results/$TID.out.md"
    Leave the tasks.jsonl line (append-only log; harmless).

Report:

POCKET ► TEST
  task_id: [tid]
  whatsapp: [SENT|SKIPPED-disabled|FAIL: reason]
  email:    [SENT|SKIPPED-disabled|FAIL: reason]
  duration: [Xs]

If any channel fails, surface the last error line from the corresponding *.log and .health file.


Route — logs [N]

bash
N="${ARG_2:-50}"
STATE_DIR="${POCKET_STATE_DIR:-$HOME/.claude/state/pocket}"

for f in activity-notifier.log run.log executor.log out-queue.log email-bridge.log; do
  echo "==> $f <=="
  tail -n "$N" "$STATE_DIR/$f" 2>/dev/null || echo "(missing)"
done

For Rule 7 mobile mode, default N=20 and skip per-file headers.


Route — restart

bash
# 1. Kickstart the launchd activity-notifier
launchctl kickstart -k "gui/$(id -u)/com.claude-ops.pocket-activity-notifier" 2>&1

# 2. (Re)spawn the pocket-exec supervisor via the executor watchdog cron script
"$CLAUDE_PLUGIN_ROOT/scripts/ops-cron-pocket-executor.py" 2>&1 | tail -10

If the launchd label is missing (notifier never installed), prompt via AskUserQuestion: [Install notifier now] / [Skip]. On install: run scripts/install-pocket-notifier.sh (Rule 4: run_in_background: true).


Route — whatsapp on|off and email on|off

Both routes mutate a small JSON config file. Never overwrite the whole file — patch in place.

bash
STATE_DIR="${POCKET_STATE_DIR:-$HOME/.claude/state/pocket}"
CFG="$STATE_DIR/whatsapp-config.json"   # or email-config.json
ACTION="${ARG_2}"                        # on | off
mkdir -p "$STATE_DIR"

if [[ ! -f "$CFG" ]]; then
  # Bootstrap empty config with enabled=false so subsequent `on` is meaningful.
  echo '{"enabled":false}' > "$CFG"
fi

case "$ACTION" in
  on)  python3 -c "import json,sys,pathlib; p=pathlib.Path('$CFG'); d=json.loads(p.read_text()); d['enabled']=True; p.write_text(json.dumps(d,indent=2))" ;;
  off) python3 -c "import json,sys,pathlib; p=pathlib.Path('$CFG'); d=json.loads(p.read_text()); d['enabled']=False; p.write_text(json.dumps(d,indent=2))" ;;
  *)   echo "usage: /ops:pocket {whatsapp|email} {on|off}"; exit 2 ;;
esac

# Echo new state
cat "$CFG"

Required keys per channel (validated by status):

  • whatsapp-config.json — chat_jid (WhatsApp JID of the self-chat, e.g. <digits>@s.whatsapp.net).
  • email-config.json — to (self-email address).

If on is requested but the required key is missing, switch state to enabled: false and route the user to setup via AskUserQuestion.


Rule reminders

  • Rule 1: Max 4 options per AskUserQuestion. None of the routes above need more.
  • Rule 2: Never delegate commands to the user — every command runs through Bash.
  • Rule 3: Never auto-skip channels in setup — always ask explicitly.
  • Rule 4: Background by default in setup (run_in_background: true).
  • Rule 5: This skill performs NO destructive ops — no deletes, no terminates, no force-kills beyond launchctl kickstart -k (which is a graceful restart, not a wipe).
  • Rule 6: No outbound comms staged from this skill except the synthetic test message in test, which goes to the operator's OWN self-chat / self-email (already enabled by the user) — not to third parties. Per-message approval rule is satisfied by the test being a no-args opt-in.
  • Rule 7: Mobile mode renders plain text only — see status example.

Native tool usage

  • AskUserQuestion: only when there is a real choice to make (stalled daemon → restart? missing channel config → bootstrap?). Never required on a green status.
  • Skill: setup route may delegate to the setup channel doc.

See also

  • STATUS.md — full architecture: 8 scripts, every state file, lifecycle, troubleshooting.
  • scripts/ops-cron-pocket-watcher.py, scripts/ops-pocket-triage.py, scripts/ops-cron-pocket-executor.py, scripts/ops-pocket-activity-notifier.py, scripts/ops-pocket-out-queue.py, scripts/ops-pocket-email-bridge.py, scripts/ops-pocket-whatsapp-bridge.py.
  • scripts/install-pocket-notifier.sh, scripts/com.claude-ops.pocket-activity-notifier.plist.
  • templates/pocket-supervisor-prompt.md — the prompt the long-lived supervisor runs under.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Ops Pocket AI skill do?

OPS on-demand: This skill should be used when the user asks to "pocket memos", "voice memo pipeline"…

Why use Ops Pocket on TypingMind?

Because you install it once and use it with any model. Ops Pocket is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Ops Pocket in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Lifecycle-Innovations-Limited/claude-ops/tree/main/claude-ops/skills/ops-pocket. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Ops Pocket?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Ops Pocket?

As many as you like. As long as a model supports skills, you can use Ops Pocket with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Ops Pocket AI skill free?

Yes. It is published on GitHub by Lifecycle-Innovations-Limited under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇