linkedin-outreach-skill

LinkedIn Local Outreach Skill

A local-first LinkedIn outreach skill for AI agents. Scout warm leads from LinkedIn activity, keep a local review database, send low-volume connection requests from your own visible browser session, sync accepted connections, and send first-degree DMs only after explicit approval.

This skill is designed for a user's own LinkedIn account on their own Mac or workstation. It does not ship proxies, rented accounts, credential login, CAPTCHA bypasses, stealth patches, or any shared database.

Install As A Claude Code Skill

npx skills add Bomx/linkedin-outreach-skill

Or clone manually:

git clone https://github.com/Bomx/linkedin-outreach-skill.git
cd linkedin-outreach-skill
python3 -m pip install -r requirements.txt
python3 -m playwright install chromium

Claude reads SKILL.md for the actual workflow.

What This Repo Includes

What This Repo Does Not Include

Your local data is intentionally excluded:

• no LinkedIn cookies or browser profiles
• no storage_state.json
• no leads.json
• no signals.json
• no action logs, run logs, screenshots, or exports
• no personal templates or message history

By default, runtime state is stored outside the repo at:

~/.linkedin-outreach/

You can override that location:

export LINKEDIN_OUTREACH_DATA_DIR=/absolute/path/to/linkedin-outreach-data

The .gitignore blocks common local-state paths as a second layer of protection, but the safest pattern is still: never put session data or lead databases inside a public repo.

Setup

1. Login Locally

python3 scripts/linkedin_outreach.py login --timeout 600

A visible Chromium window opens. Sign into LinkedIn manually. The script saves the local browser profile under ~/.linkedin-outreach/session/.

2. Check The Environment

python3 scripts/linkedin_outreach.py doctor
python3 scripts/linkedin_outreach.py status

3. Add Reusable Signals

High-intent SEO buyer posts:

python3 scripts/linkedin_outreach.py signal-add \
  --name "seo-buyer-intent-posts" \
  --type post_intent \
  --query "looking for SEO consultant" \
  --intent buyer_intent \
  --source-label "SEO buyer intent post"

Active SEO agency owners:

python3 scripts/linkedin_outreach.py signal-add \
  --name "active-seo-agency-owners" \
  --type post_intent \
  --query "SEO agency founder" \
  --source-label "Active SEO agency owner" \
  --limit 50 \
  --scrolls 6

Warm competitor/person engagement:

python3 scripts/linkedin_outreach.py signal-add \
  --name "competitor-engagers" \
  --type activity \
  --target-url "https://www.linkedin.com/in/example/" \
  --source-label "Engaged with watched profile" \
  --limit 30

Typical Workflow

Scout without sending:

python3 scripts/linkedin_outreach.py scout-search \
  --query "SEO agency founder" \
  --source-label "Active SEO agency owner" \
  --signal-name "active-seo-agency-owners" \
  --limit 50 \
  --scrolls 6 \
  --no-connect

Review leads:

python3 scripts/linkedin_outreach.py review --list --limit 30
python3 scripts/linkedin_outreach.py review --queue li_abc123 li_def456

Dry-run queued connection requests:

python3 scripts/linkedin_outreach.py connect --queued --limit 3

Actually send only after explicit approval:

python3 scripts/linkedin_outreach.py connect \
  --queued \
  --limit 3 \
  --execute \
  --no-note \
  --navigation-mode random \
  --min-delay 45 \
  --max-delay 90

Before DMs, sync accepted connections:

python3 scripts/linkedin_outreach.py sync-connections --contacted --limit 12

Send one first-degree DM:

python3 scripts/linkedin_outreach.py dm \
  --lead-id li_abc123 \
  --message "Hi {first_name}, thanks for connecting. Thought this might be useful..." \
  --execute

Safety Defaults

• Outbound commands default to dry-run.
• Connection requests send without a note by default.
• Custom connection notes are allowed only for exactly one selected lead.
• The browser is visible for login and runs locally.
• The script stops on LinkedIn checkpoints, CAPTCHA, restrictions, or unusual security prompts.
• Stop conditions record a 24-hour local safety cooldown in ~/.linkedin-outreach/db/safety_state.json.
• Browser-heavy commands refuse to start during an active cooldown unless the operator manually resolves LinkedIn and passes --force-cooldown-override.
• Suppressed profiles in ~/.linkedin-outreach/db/suppressions.json are skipped during future lead upserts.
• Already sent, pending, connected, or skipped leads are not reopened for duplicate connection requests.
• sync-connections should run before DMs so accepted requests are marked connected.

CLI Reference

python3 scripts/linkedin_outreach.py --help
python3 scripts/linkedin_outreach.py scout-search --help
python3 scripts/linkedin_outreach.py connect --help
python3 scripts/linkedin_outreach.py dm --help

Every command emits one final machine-readable line:

RESULT: {"status": "ok", ...}

Agents should treat the RESULT: line as the source of truth.

Privacy And Security

• No shared database. Leads, actions, templates, and signals live on the user's machine.
• No cookies in Git. LinkedIn sessions are saved under ~/.linkedin-outreach/session/, outside this repo by default.
• No proxy or stealth system. This is deliberate local automation for normal, low-volume personal-account workflows.
• No telemetry. The script writes local JSON files and drives the user's visible browser session.

Want The Rest Of Your SEO Engine Automated?

This skill helps with LinkedIn lead scouting and outreach. Distribb handles the broader SEO system:

• keyword research with buyer-intent signals
• content publishing to WordPress, Webflow, and Shopify
• original-data research pages
• internal linking and content repurposing
• backlink exchange through a high-DR partner network

-> Try Distribb free at distribb.io

Built by Borja Obeso, founder of Distribb.