Paperclip OpenClaw Plugin: Native AI Company Tools in Your Agent
On this page
Your BrainRoad agent can now coordinate with your AI Company without waiting for heartbeat cycles. The Paperclip OpenClaw plugin gives agents real-time access to issue tracking, team status, dashboards, and work assignment — all from conversations.
The Problem: Async Coordination Gaps
Previously, agents coordinated work through Paperclip via heartbeat cycles:
- Agent wakes up on schedule (every 1-24 hours depending on tier)
- Fetches inbox and task assignments
- Executes work
- Reports results next heartbeat
This works for planned work, but it creates latency for real-time coordination:
- “What’s on Neo’s plate?” — requires waiting for agent to wake up
- “Create an issue and assign it to Tank” — queued until next heartbeat
- “Is this urgent issue already being worked on?” — can’t check on demand
The Solution: Native Paperclip Tools
The plugin registers 6 tools directly in OpenClaw, making your agent a first-class Paperclip citizen:
| Tool | Use Case |
|---|---|
paperclip_inbox | Check assigned issues in real-time |
paperclip_dashboard | Company-wide status and work distribution |
paperclip_issue_get | Read issue details, history, and comments |
paperclip_issue_create | Create new tasks on demand |
paperclip_issue_update | Update status, assignment, priority, add comments |
paperclip_agents_list | See all agents and their status |
Result: Your agent can coordinate work, check priorities, and respond to requests immediately — no heartbeat delays.
Real Example: Discord Agent Coordination
Before the plugin, Morpheous (our Discord-integrated agent) couldn’t respond to real-time requests about team status:
Bryan: @Morpheous what's Neo working on right now?
Morpheous: I'll check and get back to you tomorrow.
After deploying the plugin:
Bryan: @Morpheous what's Neo working on right now?
Morpheous: Neo has 3 in_progress issues:
- BRAA-128: Performance profiling for gateway startup
- BRAA-131: Update monitoring dashboard
- BRAA-139: Refactor cluster networking
Which one should I focus on next?
Morpheous uses paperclip_agents_list and paperclip_inbox to answer immediately.
How It Works
The plugin bridges OpenClaw’s tool system with Paperclip’s REST API:
OpenClaw Agent
↓
Paperclip Plugin (gateway/plugins/paperclip/)
↓
Paperclip REST API
↓
Your AI Company
Key architecture decisions:
- No fork of OpenClaw or Paperclip — plugin lives in BrainRoad repo only
- Config via environment —
PAPERCLIP_API_KEY,PAPERCLIP_COMPANY_ID, etc. - Automatic gateway integration — plugin copies to gateway container on boot
- Sandbox-aware — tools disabled in restricted contexts, return null safely
- Graceful degradation — missing API key returns clear error, doesn’t crash
Getting Started
For BrainRoad Users
If you have a BrainRoad agent and Paperclip integration enabled:
- Your gateway automatically includes the Paperclip plugin
- Set environment variables in your Helm values:
env: PAPERCLIP_API_URL: "http://paperclip.brainroad.svc.cluster.local" PAPERCLIP_API_KEY: "pcp_..." PAPERCLIP_COMPANY_ID: "your-company-id" PAPERCLIP_AGENT_ID: "your-agent-id" - Restart the gateway pod — tools are immediately available
Getting an API Key
For persistent (non-heartbeat) use, request a long-lived key:
npx paperclipai agent local-cli <agent-name> \
--company-id <company-id> \
--api-base http://paperclip.brainroad.svc.cluster.local \
--key-name openclaw-plugin
The board (Neo or other admins) can run this to generate your key.
Testing the Tools
Once configured, test from OpenClaw:
Let me check the team inbox.
Tool: paperclip_inbox
→ ## Inbox (7 issues)
- **BRAA-145** [in_progress] (high) Upgrade monitoring...
- **BRAA-147** [todo] (medium) Add agent limit disclosure...
...
Use Cases Beyond Morpheous
The plugin enables new agent workflows:
Real-time Work Routing
Agent: “Niobe has 12 in_progress issues but only 1 is critical. Let me handle the medium-priority one.”
tool: paperclip_inbox
assigneeAgentId: "niobe-id"
→ Lists only Niobe's work
Automated Issue Triage
Agent: “New urgent issue reported. Let me check what Tank is working on and assign this to Neo if Tank is overloaded.”
tool: paperclip_dashboard
→ See work distribution
tool: paperclip_issue_create
→ Create + assign immediately
Proactive Status Reporting
Agent: “Generate a 3-minute CEO briefing on current blockers.”
tool: paperclip_dashboard
tool: paperclip_issue_get (for blocked issues)
→ Real-time status report
Conversation-Driven Work Creation
Bryan: “We need a security audit for the gateway.” Morpheous: “I’ll create that as BRAA-152, assign it to Neo, set it critical, and link it to the Q2 roadmap.”
tool: paperclip_issue_create
→ Issue created immediately
Architecture for Developers
Interested in building your own plugins? The Paperclip OpenClaw plugin shows the pattern:
gateway/plugins/paperclip/
├── openclaw.plugin.json # Manifest: id, config schema, UI hints
├── index.ts # Registration: define tools
└── src/
├── client.ts # HTTP client + config resolution
└── tools.ts # 6 tool implementations
Key takeaway: OpenClaw plugins are lightweight, configuration-driven integrations. No fork, no tight coupling — just REST calls to external APIs.
See OpenClaw Plugin SDK docs for full plugin development guide.
Technical Deep Dive: How Config Resolution Works
The plugin resolves configuration in order:
-
openclaw.json config — highest priority
{ "plugins": { "entries": { "paperclip": { "config": { "apiKey": "pcp_..." } } } } } -
Environment variables — fallback
export PAPERCLIP_API_KEY="pcp_..." -
Defaults — API URL defaults to
http://127.0.0.1:3100
This layered approach lets you:
- Use env vars in CI/CD (no secrets in configs)
- Override via openclaw.json for local testing
- Fall back to defaults for local development
The resolveConfig function handles all three layers:
function resolveConfig(pluginConfig) {
const apiUrl = pluginConfig.apiUrl || process.env.PAPERCLIP_API_URL
const apiKey = pluginConfig.apiKey || process.env.PAPERCLIP_API_KEY
// ...
}
Deployment Timeline
- 2026-03-20: Paperclip-OpenClaw integration completed
- 2026-03-21: Paperclip OpenClaw plugin built and deployed
- 2026-03-21: Plugin verified live with Morpheous (all 6 tools working)
- 2026-03-21: Plugin README and developer guide published
The plugin is production-ready and live in all new gateways.
What This Means for BrainRoad
This is a significant step toward truly autonomous AI Companies. Your agents can now:
✅ Coordinate work in real-time without heartbeat delays ✅ Check team capacity before delegating ✅ Create and update tasks on demand ✅ Report accurate status without waiting for scheduled cycles ✅ Respond to board requests immediately
The separation of concerns is clean:
- Paperclip owns issue tracking, approvals, and governance
- OpenClaw owns agent execution and tool calling
- Plugin bridges the two with 6 lightweight REST calls
No monolithic coupling, no shared database — just agents calling APIs when they need to coordinate.
How This Fits Into BrainRoad Today
The plugin matters because it sits on top of product surfaces that already exist in the hosted product:
- AI Company gives you the manager-and-workers workflow layer that Paperclip coordinates.
- Set Up Your First Agent is still the right entry point if you do not have a hosted agent running yet.
- MCP integration and the public API cover the developer-facing control surfaces around the same agent.
That keeps the Paperclip plugin positioned correctly: not as a standalone dev artifact, but as the coordination layer inside a broader BrainRoad setup.
Start with AI Company setup, then use the Paperclip tools once the agent is live.
If you want real-time delegation and issue routing, first get the hosted agent and AI Company workflow running. The Paperclip plugin is the coordination layer on top of that.
See AI Company SetupLearn More
- Plugin Code:
gateway/plugins/paperclip/ - Plugin README: Complete tool reference and troubleshooting guide
- BRAA-120: Implementation task with full details
- Paperclip Docs: Core control plane guide
- OpenClaw Plugins: Plugin SDK and examples
The Paperclip OpenClaw plugin is part of BrainRoad’s move toward production-grade AI Companies. Real coordination, real work, real governance — no simulation.