roadflow MCP server

• Registry nameio.roadflow/roadflowverified in the MCP registry

The registry name sits under roadflow.io. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

• Stateanswerschecked 2026-10-26
• Tools43
• Server calls itselfroadflow
• Server version1.0.0
• Protocol version2025-06-18

On 2026-10-26 the server answered initialize and tools/list without a sign-in. First answer in 180 ms.

• Its tools do plan roadmap steps across parallel lanes and goals; share view-only roadmap links with stakeholders, but not sync roadmap items from a Slack list.
• Checked on 1 of the 7 days it takes to judge reliability.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

• INITE MCP scoreafter a weeka score needs 7 daily checks: 1 so far
• What an agent can do with it2 of 3 use cases
• Answeredmeasuring1 of the 7 daily checks it takes to print a share
• Median first answer761 ms
• Space in an agent's context≈ 4,533 tokensthe definitions of all its tools (43), which an agent carries on every turn
• Tool definitionsC · 73 of 100
• Typed results0 of 43 toolsdeclared outputSchema: results come typed, read once

• Parameters without a type: in 8 of 43
• Parameters without a description: in 18 of 43
• Not saying whether it only reads or can destroy: 43 of 43

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

• roadflow_new

  Create an ephemeral in-memory roadflow workspace (default). Optional seed JSON and/or markup. No cloud id until create_shared.

• roadflow_status

  Return mode, ids, URLs, and road/step counts for the current MCP session workspace.

• roadflow_get

  Return the current workspace as JSON and/or markup.

• roadflow_apply_json

  Replace workspace state with normalized JSON (serializeState shape). In shared edit mode, saves to the cloud automatically.

• roadflow_apply_markup

  Replace roads via friendly markup (same as the editor panel). Shared edit mode auto-saves.

• roadflow_open_shared

  Load a cloud map by edit or view id into this session. View links are read-only.

• roadflow_create_shared

  Persist the current workspace to the cloud (permanent share). Returns editUrl (play/edit/share further) and viewUrl (visual canvas, read-only). Prefer viewUrl when the user only needs to see the map.

• roadflow_export_json

  Export the current workspace as JSON (serializeState).

• roadflow_export_markup

  Export the current workspace as friendly markup.

• roadflow_export_csv

  Export steps as CSV (road, goal, lane, step, notes, type, status, position, date).

• roadflow_list_versions

  List shared-map version checkpoints (newest first). Relative indexes: -1 = newest. Requires a shared edit workspace (not view-only).

• roadflow_get_version

  Load one shared-map version by relative index (e.g. -1 newest, -2 previous). Returns JSON state. Does not change the live map or workspace unless load=true.

• roadflow_discard

  Clear the in-memory workspace for this MCP session (does not delete cloud maps).

• artifacts_new

  Create an ephemeral in-memory Artifacts graph workspace. Optional seed JSON (serializeState shape: nodes + labels). No cloud id until artifacts_create_shared.

• artifacts_status

  Return mode, ids, /artifacts/?id= URLs, and node/label counts.

• artifacts_get

  Return the current Artifacts workspace as JSON.

• artifacts_apply_json

  Replace (or merge) the Artifacts graph with normalized JSON. Shared edit mode auto-saves to the cloud.

• artifacts_open_shared

  Load a cloud Artifacts graph by edit or view id. View links are read-only.

• artifacts_create_shared

  Persist the current Artifacts workspace to the cloud. Returns editUrl and viewUrl under /artifacts/.

• artifacts_export_json

  Export the current Artifacts workspace as JSON (serializeState).

• artifacts_list_versions

  List shared Artifacts version checkpoints (newest first). Requires a shared edit workspace.

• artifacts_get_version

  Load one shared Artifacts version by relative index (-1 newest). Optional load=true replaces the MCP workspace (does not save).

• artifacts_discard

  Clear the in-memory Artifacts workspace for this MCP session (does not delete cloud graphs or the Plans workspace).

• model_new

  Create a Model project in MySQL (parameter register). Allocates a view id (mv-…) when omitted and returns viewUrl. Optional seed: title, calendar, streams, axes, phases, parameterSets, chains, parameters, observations, edges. Does not…

• model_open

  Bind this MCP session to an existing Model project (edit or view id). View ids are read-only.

• model_status

  Counts, rev, and /model/?id= URLs for the bound project (or pass id).

• model_get

  Hydrate records from MySQL. include=summary (default counts), dictionaries, parameters, observations, edges, or all. Prefer model_upsert_observation to change one fact — do not treat this as a blob to rewrite.

• model_discard

  Unbind the Model project from this MCP session. Does not delete MySQL rows.

• model_patch_project

  Update title, calendar, viewId, streams, axes, phases, parameterSets, and/or chains. Send If-Match via rev. Replaces each provided dictionary in full.

• model_list_parameters

  List parameter catalog rows. Optional filters: streamId, role, layer, primaryOwner.

• model_get_parameter

  One catalog row plus its observations (facts).

• model_upsert_parameter

  Create or patch one parameter definition (not observations). Pass rev for If-Match on update.

• model_upsert_parameters

  Create or patch many catalog rows (cap 400). Same body shape as model_upsert_parameter. continueOnError returns per-row errors; otherwise one transaction.

• model_delete_parameter

  Delete a parameter and its observations (cascade).

• model_list_observations

  List facts. Filter by parameterId, setId, evidenceStatus. Sets are a column (Base/Downside/Upside), not an axis.

• model_upsert_observation

  Create or replace one fact (parameterId + scope + setId). Example: Base P2 facility MW. Do not write CALC-computed values; only IN facts or CALC overrides (valueOverride). Pass rev for If-Match.

• model_upsert_observations

  Create or replace many facts (cap 400). Same body shape as model_upsert_observation. Rejects CALC outputs unless valueOverride. continueOnError returns per-row errors; otherwise one transaction.

• model_delete_observation

  Delete one fact by parameterId + scope/scopeKey + setId.

• model_list_edges

  Manual lineage edges only (SOURCE/FLOW/INTERFACE). FORMULA edges are derived from formulas on load — not stored.

• model_upsert_edge

  Create or patch a manual SOURCE, FLOW, or INTERFACE edge. FORMULA is rejected.

• model_delete_edge

  Delete a manual edge by edgeId.

• model_evaluate

  Read-only computeModel. Does not write MySQL. View ids allowed. Default period is calendar start (not today). periodId and range* are mutually exclusive. Cell cap 2000.

• model_list_dictionaries

  Streams, axes (members), phases, parameter sets, chains, and calendar. Period members may be derived from calendar.