Model Context Protocol (MCP) Integration

Rescile natively supports the Model Context Protocol (MCP), allowing AI agents (like Claude Desktop, Cursor, or autonomous agents) to connect securely, reason over your graph, audit compliance, and execute validated mutations via intent contracts.

The MCP server operates over HTTP Server-Sent Events (SSE) at /mcp and exposes the following tools:

Graph Query & GraphRAG Tools

AI agents can invoke these tools to extract precise architecture and parameter data:

• getResource: Fetch specific properties of a single known node (type, name).
• listResources: Paginated list of resources filtered by type and exact properties.
• getNeighbors: Retrieve upstream and downstream dependencies from a central node.
• findPath: Find connection paths between two arbitrary resources.
• localSearch: GraphRAG ego-centric context retrieval leveraging auto-generated community summaries.
• globalSearch: GraphRAG dataset-wide map-reduce queries.
• semanticSearch: Sub-string fuzzy matching on resource descriptions and properties.
• listResourceTypes: Exposes the entire graph schema (node types, property schemas, edge relations).

Dependency & Blast-Radius Traversal

• getBlastRadius: Traces downstream and upstream dependency cascades from a central resource:
  • Supports filtering by specific edge relations (e.g. ['LOOKS_UP', 'EXTENDS_DDIC', 'DEPENDS_ON']).
  • Supports custom traversal depth (max_depth: 3).
  • Returns affected nodes, direct impact count, indirect impact count, and full traversal paths.

Compliance Findings & Audit Tools

• listComplianceFindings: Lists active compliance findings and policy violations across the property graph.
  • Optional resource_type filter (e.g. 'compliance_violation', 'finding'). If omitted, matches all compliance-generated finding nodes.
  • Optional severity filter ('critical', 'warning', 'info').
  • Optional rule substring filter.
  • Returns structured finding codes, severity, description messages, and autonomous repair metadata:
    • remediation_contract: Name of the intent contract to call for automated healing.
    • suggested_inputs: Pre-calculated inputs map tailored to resolve this specific violation.
• runAudit: Triggers an immediate compliance audit check on the graph.
  • Returns overall passed boolean, count of critical_findings and warning_findings, and detailed list of active findings.

Intent Contracts & Mutation Governance

To prevent silent data corruption, Rescile enforces contract-driven mutations:

• listContracts: Lists all intent contracts defined in the workspace (contracts/*.toml), including their descriptions, typed input schemas (with bounds, regex patterns, and formats), and target assets or input files.
• invokeContract (Requires --mcp-write): Executes an intent contract by name with structured inputs.
  • Validates input values against the contract’s schema (pattern, bounds, format, enum_lookup).
  • Executes atomic multi-write operations with rollback protection.
  • Re-evaluates the graph immediately and returns affected asset files.
• saveAsset (Requires --mcp-write and --allow-raw-assets): Raw CSV asset writer. Disabled by default in favor of typed invokeContract mutations.

Action Execution Tools

• executeAction: Triggers actions defined in module actions/*.toml (such as test runners, dry-run linters, or deployment checks).
  • Can optionally block and wait for execution logs, exit codes, and stdout/stderr output.

Pre-packaged Prompts

Agents can invoke predefined prompts to streamline interactions:

• get_resource_details: Analyzes a resource and its direct dependencies.
• graphrag_local_search / graphrag_global_search: Pre-structures requests for deep graph analysis.