MCP Agent Access

Professional

MCP Agent Access lets Codex, Claude Code, Claude.ai Cowork, Claude Desktop, Cursor, ChatGPT, and API clients use Algo Trade Analytics as a durable quant research workbench.

Your external agent brings its own model, context window, and token budget. Algo Trade Analytics supplies the trading-specific substrate: Case Studies, Pine strategy context, backtest tools, evidence gates, research notebooks, background jobs, and candidate handoff back into Algo Lab.

Location: Dashboard > MCP Access

MCP Agent Access is included with the Professional plan. Basic users can use the native Algo Trade Analytics workflows for Algo Lab, AI Pine editing, Backtest vs Alerts, and Alerts vs Fills. External MCP clients require Professional, beta access, or an Professional access or an admin-granted beta entitlement.

MCP is research access, not trade execution

The Algo Trade Analytics MCP server is not a broker connection. It cannot place trades, modify your broker account, or bypass evidence gates. Treat MCP output as research evidence that still needs review before any live trading decision.

Why MCP Matters

Use Your Agent Harness

Drive research from Codex, Claude Code, Cowork, Cursor, ChatGPT, or another client while keeping Algo Trade Analytics as the system of record.

Keep Durable Evidence

Store hypotheses, protocols, trials, lessons, promoted candidates, and linked reports in Case Studies instead of losing them inside one chat.

Run Algo Trade Analytics Tooling

Let the agent call Algo Trade Analytics context, Case Study, backtest, evidence, sweep, walk-forward, bootstrap, and research-job tools.

Stay Scoped

Use OAuth and least-privilege scopes. Manual bearer tokens remain available only as a fallback for clients that cannot complete OAuth.

MCP vs In-App Algo Agent

Path	Best for	Model/token owner	Research state
In-app Algo Agent	A guided Algo Trade Analytics-native research loop inside Algo Lab.	Algo Trade Analytics hosted model usage and Algo Trade Analytics credits.	Stored in the selected Case Study.
MCP Agent Access	Advanced local or hosted agent harnesses that should drive Algo Trade Analytics tools directly.	Your external client spends its own tokens; Algo Trade Analytics credits still cover Algo Trade Analytics-hosted AI/server work and paid tool compute.	Stored in Algo Trade Analytics Case Studies, notebooks, jobs, and candidate records.

Use the in-app Algo Agent when you want Algo Trade Analytics to manage the agent loop for you. Use MCP when you want a local or external agent to plan, inspect, and drive Algo Trade Analytics research tools while Algo Trade Analytics preserves the evidence trail.

Typical Research Loop

1. Create or select a Case Study in Algo Lab so the agent has a durable research ledger.

2. Connect an MCP client through OAuth from the MCP Access page.

3. Start with ata_context so the client verifies the bound Algo Trade Analytics user, market-data readiness, available actions, and guidance.

4. Load the Case Study and current research contract before calling paid or heavy tools.

5. Run evidence work through direct tools or durable background jobs when the check may take time.

6. Read meta.clientGuidance and meta.researchState after every tool call before choosing the next action.

7. Review outcomes in Algo Lab before accepting staged code, promoting a candidate, or using the result outside Algo Trade Analytics.

What MCP Is Not

• It is not a trading bot or broker execution channel.
• It is not a way to bypass Algo Trade Analytics credits for Algo Trade Analytics-hosted AI or server-side work.
• It is not proof that a strategy is safe for live capital.
• It is not TradingView’s official Pine runtime.
• It is not a replacement for reviewing staged code and linked evidence.

MCP Docs

Client Setup Connect Codex, Claude, Cursor, ChatGPT, Studio, and manual-token clients

OAuth and CIMD How Algo Trade Analytics OAuth discovery, protected-resource metadata, and Client ID Metadata Documents work

Tools and Workflows The MCP tool families, recommended call order, and background research jobs

Security, Scopes, and Credits Entitlements, scopes, token handling, revocation, and credit boundaries

Troubleshooting Fix OAuth, client setup, missing tools, revoked connections, and research-job issues

Related

Algo Agent Algo Trade Analytics's native research loop inside Algo Lab

Case Study Tab The ledger MCP research writes back into

Algo Lab The in-app workspace that MCP research connects back to