Skip to main content

LinkedCamp plugin

Recommended. The LinkedCamp connection, the skill, and approval prompts.

LinkedCamp skill

The skill on its own. You connect LinkedCamp yourself, and there are no approval prompts.
Both are published on ClawHub and talk to the same hosted server every other client uses:

​
Plugin or skill?

Use the plugin unless you have a reason not to.

​
Install

Requires OpenClaw 2026.6.1 or newer.
On OpenClaw main/dev builds with ClawHub as the default source, the clawhub: prefix is optional. Stable releases through 2026.10.4 need it.

​
Connect your account

The login prints a link. Open it, paste your API key into the LinkedCamp page, and approve. OpenClaw stores an encrypted token, not your key.If the Gateway runs on another machine and the browser cannot return to it, finish with:
In the Control UI, the same flow is Settings → MCP → linkedcamp → Sign in.
Your API key is in Settings → Integrations → API. See the quick start for plan requirements. Then check the connection:
and ask the agent: “Which LinkedCamp account am I connected to?”
Never paste your API key into a chat. The agent is told not to ask for it; if a key does end up in a conversation, rotate it in Settings → Integrations.

​
Approval prompts

With the plugin installed, the agent asks you to allow once or deny before it:
  • Reaches people — launching or resuming a campaign, sending a message or test email, scheduling or publishing a LinkedIn post, changing posting slots, turning on an automation.
  • Changes live outreach — a campaign’s audience, saving or editing a sequence, attaching an agent, editing an active agent, knowledge-base entry or persona, overwriting templates, saved replies or notes, skipping a scheduled step.
  • Removes data — deleting campaigns, leads, messages, agents, knowledge, personas, tags, blacklist rules, merge fields, blocked dates or reminders, and blacklisting a lead.
Reading data, building a draft campaign, adding leads, private draft posts, and inactive agents or automations go through without a prompt. Approve in the Control UI, with /approve, or in the chat the request was routed to.
With the skill alone there are no approval prompts. The agent is instructed to confirm in the conversation, but nothing enforces it.

​
Plugin settings

Set under plugins.entries.linkedcamp.config:

​
Hide tools entirely

To take tools away from the agent instead of approving them one by one:

​
What you can do

The agent reaches the full tool surface your plan and permissions allow. Tools appear in OpenClaw as linkedcamp__<tool> — for example linkedcamp__whoami or linkedcamp__campaign_list.
  • Campaigns: build, generate LinkedIn and Sales Navigator audience searches, author multi-step sequences, launch after confirmation
  • Leads, notes, tags and blacklists
  • The unified inbox and reply drafts
  • AI reply agents and knowledge bases
  • LinkedIn post scheduling
  • Campaign, account and funnel analytics, including agency sub-accounts
Try:
  • “Show me my LinkedCamp account and unread inbox counts.”
  • “Build a draft campaign for heads of customer support at B2B SaaS companies. Show me the audience and the sequence before anything sends.”
  • “Summarize my campaign performance for the last 30 days.”
  • “Find my unread conversations, prioritize interested prospects and draft a reply for each. Don’t send anything.”

​
What it will not do

  • Connect a LinkedIn account — that takes your LinkedIn password, so do it in LinkedCamp
  • CRM integrations, billing, team management or API-key changes
The agent will never ask for a LinkedIn password, session cookie or verification code.

​
Versions

The current release of both the plugin and the skill is 1.0.4.