Parallect Skill for OpenClaw

An OpenClaw skill that connects your autonomous agent to Parallect.ai — a multi-provider deep research platform that queries Perplexity, Gemini, OpenAI, Grok, and Anthropic in parallel, then synthesizes results into a unified report with cross-referenced citations and conflict resolution.

What it does

When installed, this skill teaches your OpenClaw agent to:

• Run deep research on any topic using multiple AI providers simultaneously
• Manage budgets thoughtfully — always discusses cost before spending your money
• Poll asynchronously — submits research, checks back with exponential backoff, and delivers results when ready
• Present findings intelligently — summarizes key insights first, then shares the full synthesis with citations
• Suggest follow-ons — offers next research directions from the synthesis engine
• Track spending — reports cost per query and cumulative session spend

Quick Start

1. Get your API key

Sign up at parallect.ai and go to Integrations (or Settings > API Keys) to create a key.

2. Install the skill

Option A: ClawHub (recommended)

clawhub install parallect

Option B: Manual

mkdir -p ~/.openclaw/skills/parallect/references
cp SKILL.md ~/.openclaw/skills/parallect/SKILL.md
cp references/*.md ~/.openclaw/skills/parallect/references/

3. Configure your API key

Add to ~/.openclaw/openclaw.json:

{
  "skills": {
    "entries": {
      "parallect": {
        "enabled": true,
        "env": {
          "PARALLECT_API_KEY": "par_live_YOUR_KEY_HERE"
        }
      }
    }
  }
}

4. Refresh OpenClaw

Ask your agent to "refresh skills" or restart the gateway.

Usage

Just talk to your agent naturally:

• "Research the latest advances in quantum computing"
• "Deep dive on the competitive landscape for AI code editors"
• "Look up what's happening with CRISPR gene therapy clinical trials"
• "Find out about the regulatory implications of autonomous vehicles in the EU"

The agent will check your balance, discuss budget, submit the research, poll for results, and deliver a synthesis with citations.

How it works

You: "Research X"
  │
  ▼
Agent checks balance ──→ Discusses budget tier with you
  │
  ▼
Agent calls research(query, tier) ──→ Gets jobId
  │
  ▼
Agent polls research_status(jobId) ──→ Reports progress
  │                                      "2 of 3 providers done..."
  ▼
Agent calls get_results(jobId) ──→ Delivers synthesis
  │
  ▼
Agent presents follow-on suggestions ──→ You pick one (or ask your own)

Budget Tiers

Tier	Max Cost	Providers	Best for
XXS	~$1	1	Quick factual lookups
XS	~$2	1-2	Brief overviews
S	~$5	2	Standard questions
M	~$15	3-4	Detailed research (default)
L	~$30	4-5	Comprehensive analysis
XL	~$60	All	Exhaustive multi-provider

The agent will suggest an appropriate tier based on your question's complexity and confirm before spending.

MCP Tools

This skill uses Parallect's hosted MCP server at https://parallect.ai/api/mcp/mcp with 9 tools:

Tool	Purpose
research	Submit a research query (async)
research_status	Poll job progress
get_results	Retrieve completed synthesis
follow_up	Pursue follow-on questions
list_threads	List recent research threads
get_thread	Get full thread context
balance	Check credit balance
usage	View spend analytics
list_providers	See available providers and tiers

Skill Frontmatter

The SKILL.md uses standard AgentSkills fields (name, description) plus OpenClaw-specific extensions:

Field	Purpose
user-invocable: true	Tells OpenClaw this skill can be triggered by natural language, not just explicit tool calls
metadata.openclaw.emoji	Display icon in OpenClaw's skill list
metadata.openclaw.primaryEnv	Identifies the primary environment variable for setup prompts
metadata.openclaw.requires.env	OpenClaw validates these env vars are set before enabling the skill

Links

• Parallect.ai — Dashboard, billing, research history
• MCP Docs — Full MCP server documentation
• API Reference — REST API and tool reference

License

MIT