I Built a 24/7 Freelance Operations Agent with OpenClaw — It Runs My Business While I Sleep

No Python. No Node. Three skill files, one Lobster workflow, and my consultancy now runs on autopilot.

---

What I Built

A fully autonomous freelance operations agent I call Claw-Ops that handles the three biggest time drains in running a dev consultancy:

1. Morning dev briefing — PR triage from GitHub, action items pulled from Slack, delivered to Telegram at 8AM
2. Client status drafting — every Friday, reads merged PRs + closed Linear tickets, writes the weekly update email, sends me the draft to approve with one tap
3. Inbox zero on follow-ups — detects unanswered client emails older than 48 hours, drafts a polite follow-up, asks me on Telegram whether to send it

Before this: ~9 hours a week on context-switching, manual status writing, and email chasing.
After: I review a Telegram message in the morning and say yes or no.

---

The Stack

• OpenClaw (gateway running on a Mac Mini, always on)
• Claude via Anthropic API (the brain)
• Telegram (my interface — DMs and one-tap approvals)
• Three custom skills + one Lobster workflow (no external services beyond what I already used)

Zero new subscriptions. Zero new dashboards. Everything surfaces in the app I already have open: Telegram.

---

Skill 1: pr-morning-digest

# pr-morning-digest

When triggered (or on schedule), check all GitHub repos accessible via the gh CLI.

Find pull requests that match ANY of these:
- Open more than 24 hours with no review activity
- CI checks failing
- Labeled "urgent" or "blocking"

For each match, output:
- Repo name
- PR title and number
- Current status (awaiting review / CI failing / etc.)
- Direct link

Format as a clean priority list, most urgent first.
Send the result to Telegram.
If nothing matches, send: "All clear — no PRs need attention."

Tools: shell (gh CLI)

---

Skill 2: slack-action-extractor

# slack-action-extractor

When triggered, read the last 100 messages from the #engineering Slack channel
using the Slack CLI or API.

Identify messages that:
- Mention my name or @handle
- Contain task language ("can you", "please", "need you to", "by EOD", "by Friday")
- Are from the last 24 hours

For each identified message:
- Extract the action item in one sentence
- Note who asked and when
- Note any deadline mentioned (or flag as "no deadline specified")

Append results to ~/tasks.md under today's date heading.
If new items were found, send a Telegram summary.
If nothing new, do nothing.

Tools: shell (Slack CLI), file write

---

Skill 3: client-status-drafter

# client-status-drafter

Every Friday at 4PM, do the following:

1. Run `gh pr list --state merged --limit 20` to get this week's merged PRs
2. Read closed Linear tickets from the past 7 days via the Linear CLI
3. From memory, recall the client name and their current project context
4. Draft a plain-English weekly status email:
   - 2-3 sentences on what shipped
   - 1 sentence on what's in progress
   - 1 sentence on anything that needs client input or decision
   - Professional but warm tone — no bullet points, just prose
5. Save draft to ~/drafts/status-{YYYY-MM-DD}.md
6. Send me the draft on Telegram with two options:
   - Reply "send" to send it via Gmail immediately
   - Reply "edit [your changes]" to revise before sending

Tools: shell (gh, linear, gmail CLI), file write, memory read

---

The Lobster Workflow That Chains Everything

name: claw-ops-daily
schedule: "0 8 * * 1-5"

steps:
  - id: prs
    skill: pr-morning-digest
    output: pr_summary

  - id: slack
    skill: slack-action-extractor
    output: action_items

  - id: brief
    action: telegram_send
    message: |
      Good morning. Here's your briefing:

      **PRs needing attention:**
      {{pr_summary}}

      **New action items from Slack:**
      {{action_items}}

  - id: status_draft
    skill: client-status-drafter
    condition: "day_of_week == friday"
    input:
      context: "{{pr_summary}} {{action_items}}"

---

The Setup (Reproducible in Under an Hour)

Step 1: Install OpenClaw

curl -fsSL https://openclaw.ai/install.sh | bash
openclaw config set anthropic_api_key=YOUR_KEY

Step 2: Create the skill files

mkdir -p ~/.openclaw/skills
nano ~/.openclaw/skills/pr-morning-digest.md
nano ~/.openclaw/skills/slack-action-extractor.md
nano ~/.openclaw/skills/client-status-drafter.md

Step 3: Create the Lobster workflow

nano ~/.openclaw/workflows/claw-ops-daily.yaml

Step 4: Connect Telegram

openclaw connect telegram
# Follow the QR code flow — takes 90 seconds

Step 5: Start the gateway and test

openclaw start
# Message your Telegram bot: "Run claw-ops-daily now"

Total time from zero to running: 47 minutes.

---

What Actually Happens in Production

Monday 8:02 AM — Telegram:

Good morning. Here's your briefing:

PRs needing attention:
• [URGENT] backend/api — PR #312 "Fix auth token refresh" — CI failing, open 31hrs
• frontend/dashboard — PR #308 "New onboarding flow" — awaiting review, open 26hrs

New action items from Slack:
• Sarah asked you to look at the deployment config by EOD today
• No deadline: Jordan mentioned the staging DB needs a migration check

Friday 4:03 PM — Telegram:

Draft ready: ~/drafts/status-2026-04-25.md

Preview:
"This week we shipped the auth token refresh fix and the new onboarding
flow, both now live in production. We're currently working through the
staging database migration — expect that wrapped by Tuesday. One thing
we'd love your input on: the API rate limit thresholds for the premium
tier. Happy to jump on a quick call Monday if useful."

Reply "send" to send via Gmail or "edit [changes]" to revise.

I reply "send." Done.

---

The Result

Task	Before	After
Morning PR triage	25 min manual GitHub scanning	45 sec reading Telegram
Slack action item harvesting	~40 min/day of tab switching	Automated, instant summary
Weekly client status	90 min writing + context gathering	30 sec reviewing + one-tap send
Email follow-up chasing	45 min/week	Zero — agent handles it
Total	~9 hrs/week	~20 min/week

---

Honest Notes

What worked immediately: The PR digest and Telegram integration were flawless out of the box.

What needed iteration: The client status email tone took two rounds of refinement in agents.md. Adding "Write like a human who has a strong working relationship with this client. Warm, direct, no corporate filler." to the agent personality fixed it completely.

Security note: I run with --sandbox disabled because I need full shell access for the gh and linear CLIs. This is a deliberate choice with understood risk. Start sandboxed, then selectively enable only the tools each skill genuinely needs.

---

Fork It

All three skill files and the Lobster workflow are in this GitHub Repository:
https://github.com/FarooqShabbir/claw-ops

The entire point of the skills format is that you're not copying code — you're copying intent. Edit the instructions to match your stack, and the agent adapts automatically.

---

Submitted for the OpenClaw Writing Challenge on DEV — OpenClaw in Action prompt.