Parallel MCP Server
for AI Agents
Connect your AI agent to StackOne's Parallel MCP server and give it 29 MCP tools out of the box. Auth, tool execution, and security all managed.
StackOne 
Coverage
29 Agent Actions
Create, read, update, and delete across Parallel — and extend your agent's capabilities with custom actions.
Authentication
Agent Tool Authentication
Per-user OAuth in one call. Your Parallel MCP server gets session-scoped tokens with zero credentials stored on your infra.
Agent Auth →
Security
Agent Protection
Every Parallel tool response scanned for prompt injection in milliseconds — 88.7% accuracy, all running on CPU.
Prompt Injection Defense →
Performance
Max Agent Context. Min Cost.
Free up to 96% of your agent's context window to enhance reasoning and reduce cost, on every Parallel call.
Tools Discovery →What is the Parallel MCP Server?
A Parallel MCP server lets AI agents read and write Parallel data through the Model Context Protocol — Anthropic's open standard for connecting LLMs to external tools. StackOne's Parallel MCP server ships with 29 pre-built actions, fully extensible via the Connector Builder — plus managed authentication, prompt injection defense, observability, and agent execution runtime. Connect it from MCP clients like Claude Desktop, Claude Code, Cursor, Goose, and VS Code, or from agent frameworks like OpenAI Agents SDK, LangChain, and Vercel AI SDK.
All Parallel MCP Tools
Every action from Parallel's API, ready for your agent. Create, read, update, and delete — scoped to exactly what you need.
Task Runs
- Create Task Run
Create a new task run for structured web research. If the response is null or missing an id, the creation failed silently — do not treat it as success.
- Retrieve Task Run
Retrieve a task run by ID
Task Groups
- Create Task Group
Create a new task group for batch task execution
- Retrieve Task Group
Retrieve a task group by ID
Monitors
- Create Monitor
Create a web content monitor
- List Monitors
List monitors with optional filtering
- Retrieve Monitor
Retrieve a single monitor by ID
- Update Monitor
Update an existing monitor
Other (21)
- Add Runs To Task Group
Add task runs to an existing task group
- Create FindAll Run
Create an asynchronous FindAll run to discover entities. For plain company or person lookups, use parallel_entity_search instead (synchronous, one call) — choose this action only when the request explicitly needs a generator tier, an exclude list, webhooks, event streaming, or post-match enrichment.
- Add Enrichment To FindAll Run
Add enrichment to extract structured data from FindAll matches. Can be called immediately after run creation — no need to wait for the run to complete or for matches to appear.
- Search
Perform a web search with AI-powered relevance filtering
- Retrieve Task Run Input
Retrieve the input of a task run
- Retrieve Task Run Result
Retrieve the result of a completed task run. If the response contains type "error" with message "Run still active.", the run has not finished yet — that is a normal still-running signal, not a failure; wait briefly and call this action again.
- Retrieve Task Group Run
Retrieve a specific run within a task group
- Retrieve FindAll Run Status
Retrieve the status of a FindAll run
- Retrieve FindAll Run Result
Retrieve results of a FindAll run. Returns a snapshot at the time of the request — works on active runs too, so call it to get partial results instead of waiting for completion.
- Get FindAll Run Schema
Retrieve the schema of a FindAll run
- List Monitor Events
List events detected by a monitor
- Extract
Extract content from one or more URLs
- Chat Completions
Generate a chat completion with web research capabilities
- Stream Task Group Events
Stream execution events for a task group. When called via MCP tool_call, a null response is expected (SSE connection lifetime limits), not an error — the runs keep executing; poll with parallel_retrieve_task_group instead.
- Stream FindAll Events
Stream events for a FindAll run. When called via MCP tool_call, a null response is expected (SSE connection lifetime limits), not an error — the run keeps executing; poll with parallel_retrieve_findall_run_status instead.
- Cancel FindAll Run
Cancel an active FindAll run
- Extend FindAll Run
Extend the match limit of a FindAll run
- Fast Entity Search
Perform a fast synchronous entity search. Preferred for simple company or person lookups. When the request specifies a generator tier (base/core/pro), an exclude list, webhooks, streaming, or enrichment, use parallel_create_findall_run instead — though this action is also the right supplement for immediate preliminary results while a FindAll run is still matching.
- Ingest FindAll Run
Generate a structured FindAll run spec (entity type, match conditions, generator, match limit) from a natural language objective. Does not start a run or return a findall_id — pass the returned spec to parallel_create_findall_run to execute the search.
- Trigger Monitor Run
Manually trigger a monitor execution
- Cancel Monitor
Cancel an active monitor
Set Up Your Parallel MCP Server in Minutes
One endpoint. Any framework. Your agent is talking to Parallel in under 10 lines of code.
Agent Frameworks
JSON{
"mcpServers": {
"stackone": {
"command": "npx",
"args": [
"-y",
"mcp-remote@latest",
"https://api.stackone.com/mcp?x-account-id=<account_id>",
"--header",
"Authorization: Basic <YOUR_BASE64_TOKEN>"
]
}
}
}Platform Resources
MCP Code Mode: Keeping Tool Responses Out of Agent Context
Anthropic's code_execution processes data already in context. Custom MCP code mode keeps raw tool responses in a sandbox. 14K tokens vs 500.
11 min
Comparing BM25, TF-IDF, and Hybrid Search for MCP Tool Discovery
Benchmarking BM25, TF-IDF, and hybrid search for MCP tool discovery across 916 tools. The 80/20 TF-IDF/BM25 hybrid hits 21% Top-1 accuracy in under 1ms.
10 min
Indirect Prompt Injection Defense for MCP Tools: A Technical Guide
MCP tools that read emails, CRM records, and tickets are indirect prompt injection vectors. Here's how we built a two-tier defense that scans tool results in ~11ms.
12 min
MCP vs A2A: Architecture, Security, and When to Use Each
MCP vs A2A: what each protocol standardizes, how they differ, their shared security risks including indirect prompt injection, and when to use one, both, or a hybrid architecture.
12 min
MCP vs API: What 200+ Connector Builds Taught Us
MCP wraps APIs, it doesn't replace them. After building 200+ connectors that serve both, here's when each approach wins.
14 min read
Parallel MCP Server FAQ
Does StackOne have a Parallel MCP server?
Parallel MCP server vs direct API integration — what's the difference?
How does Parallel authentication work for AI agents?
origin_owner_id.Are Parallel MCP tools vulnerable to prompt injection?
What is the context bloat of a Parallel agent and how do I avoid it?
Can I limit which actions my Parallel agent can access?
Can I create custom agent actions for my Parallel MCP server?
When should I NOT use a Parallel MCP server?
What AI frameworks and AI clients does the StackOne Parallel MCP server support?
Put your AI agents to work
All the tools you need to build and scale AI agent integrations, with best-in-class connectivity, execution, and security.