Connect agents to Symbolic through MCP.

The Symbolic MCP server exposes every product as a plain language tool and returns the same organization scoped run, evidence, and status used by the console.

Connect

Add one remote server entry to your MCP client. Store the API key in the client environment, never in a committed file.

mcp.jsonjsonReady

{  "mcpServers": {    "symbolic": {      "url": "https://symbolic.sh/api/mcp",      "headers": {        "Authorization": "Bearer $SYMBOLIC_API_KEY",        "X-Symbolic-Organization": "org_123"      }    }  }}

Authentication

Required headers

Authorization: Bearer $SYMBOLIC_API_KEY

X-Symbolic-Organization: org_123

Every read and mutation remains scoped to the organization resolved from the authenticated principal.

Product tools

symbolic.flow.runOrganize a GitHub project into phases with explicit acceptance criteria. Review the evidence behind each gate and see what needs attention before the next phase.

symbolic.intent.runTurn product decisions into acceptance criteria, then compare the supplied implementation references with those decisions.

symbolic.context.runResolve repository rules, scope, and precedence into a focused instruction pack with its sources attached.

symbolic.guard.runCompare proposed actions with an explicit authority contract covering paths, commands, hosts, and budget.

symbolic.models.runBring candidate run evidence and compare accepted outcomes, cost, and task constraints using the same evaluation criteria.

symbolic.access.runCompare expected access with observations across roles, tenants, and actions to identify permission mismatches.

symbolic.blockers.runMap dependencies, owners, and work status to see what can proceed and what must be resolved first.

symbolic.handoffs.runPackage the objective, repository state, decisions, evidence, and blockers so another agent can see where to continue.

symbolic.states.runDefine roles, devices, data states, and constraints, then compare supplied observations with the resulting state map.

symbolic.migrations.runEvaluate migration scenarios, constraints, observations, and recovery evidence before deciding whether a change is ready.

symbolic.release.runCompare software claims with observed checks and review what is verified, blocked, failed, or still untested.

Example request

request.jsonjsonReady

{  "name": "symbolic.release.run",  "arguments": {    "projectId": "project_123",    "repository": "https://github.com/acme/product",    "target": "https://staging.acme.com",    "objective": "Verify that a new user can finish onboarding and see saved data after reload."  }}

The tool creates one organization scoped run.
Findings retain evidence provenance.
Run lookup uses symbolic.runs.get.