Local stdio MCP server for personal Power Automate cloud flows — reverse-engineered over the undocumented api.flow.microsoft.com service API, authenticated with your own account (no admin, no paid quota).
Strong foundation with clear naming, comprehensive schemas, and good descriptions. All 7 tools have explicit registrations with Zod schemas converted to JSON Schema. Tool names follow verb_noun pattern (list_, get_, set_, create_or_update_). Descriptions are substantive (50-200 chars range). All parameters have type definitions and descriptions. Write tools implement preview/confirm pattern (dryRun, confirm flags). Main gaps: output schemas not documented in code; no pagination support for list_flows/get_flow_runs despite potentially large result sets; error handling relies on generic errorResult wrapper without recovery guidance.
Create a new flow or update an existing one from a definition object. dryRun is the default — pass dryRun:false to actually write. Get the definition shape from get_flow.
Full flow definition (triggers, actions, parameters) and connection references.
Recent run history for a flow: status, code and timing per run.
Per-action breakdown of one run — the debugging view. Follows inputs/outputs links for failed actions by default (or a named action, or all).
List every Power Platform environment the signed-in account can see.
List cloud flows in an environment (summary fields, not the full definition).
Output schemas not documented. Tool descriptions state what is returned (e.g., 'status, code and timing per run') but no formal schema is visible in code. LLMs cannot plan downstream tool calls without knowing response structure.
No pagination support for list_flows and get_flow_runs. Both accept 'top' parameter but no cursor/offset or total_count returned. Large result sets will blow context window; description should state result limit and pagination strategy.
Error handling is generic. errorResult() wraps all exceptions as plain text. No recovery guidance (e.g., 'User not found. Try list_environments() first.'). LLM cannot distinguish retryable from fatal errors.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2026-07-28+ | v2 |
Turn a flow on (start) or off (stop). Previews by default; pass confirm:true to apply.
create_or_update_flow parameter 'definition' is typed as generic 'object' with no schema constraints. LLM cannot validate structure before calling. Should reference get_flow output schema or provide a formal definition shape.
Tool composition: create_or_update_flow requires 'definition' and 'connectionReferences' objects but no guidance on how to construct them. Description says 'Get the definition shape from get_flow' but no formal schema link. Agents must infer structure from examples.