Skip to main content
Glama

Server Details

Live, editable canvas for AI agents — author graphs via MCP, then watch and reroute runs live.

Ownership verified
Status
Healthy
Uptime
100.0% over the last 24 hours
Last Tested
Transport
Streamable HTTP · MCP 2025-11-25
URL
ChatGPT
–
Claude
–

TDQS

A3.8/5.0

Scored across 18 tools

Disambiguation4/5

Most tools map to a distinct resource+action pair, and the run-control trio (start/pause/cancel/resume) is explicitly differentiated in the descriptions (terminal vs resumable, checkpoint semantics). The only soft overlaps are draft_node vs create_graph/update_graph_nodes (mitigated by draft_node's explicit 'WITHOUT creating anything') and get_graph vs get_canvas_url, both of which are clarified in-line.

Naming Consistency5/5

Every tool is snake_case verb_noun (get_graph, create_graph, list_templates, instantiate_template, provision_credential, search_runs, update_graph_nodes), including the less conventional get_started and get_canvas_url which still follow the verb+noun shape. No camelCase mixing or vague bare verbs anywhere in the set.

Tool Count4/5

18 tools is at the upper end, but the surface genuinely spans several sub-domains: graph authoring, run lifecycle, run recall/sharing, templates, credentials, billing, and onboarding. Each tool earns its place in one of those clusters rather than being redundant, so the count is slightly heavy but not padded.

Completeness3/5

Graph CRUD is create/get/list/update but has no delete_graph, and run listing is only indirectly covered via search_runs (no list_runs for a graph). The credential vault is write-only from the agent's perspective — provision_credential exists but there is no list_credentials, revoke, or rotate, leaving a real lifecycle gap.

Available Tools

18 tools
cancel_runAInspect

Cancel a run between hops — in-flight hops finish, queued hops no-op. Requires editor access on the run's graph; a run already in a terminal status errors.

Inputs: runId (required), reason.

create_graphAInspect

Materialize a validated graph spec into a real graph on the caller's account — nodes reference each other by caller-chosen string refs, edges wire refs, entry names the entry node. Rejects invalid topology with the schema's own errors. Returns the new graph id plus its canvasUrl.

Inputs: name (required), nodes (required), edges, entry, +1 more.

draft_nodeAInspect

Turn a plain-English brief into a structured node spec (name, role, systemPrompt, description, provider, model) WITHOUT creating anything — the agent reviews/edits the draft, then materializes it with create_graph or update_graph_nodes. The same extraction the canvas quick-add runs.

Inputs: description (required), graphId.

get_canvas_urlBInspect

Deep link into the web canvas for a graph — open it to watch or steer runs live. Requires viewer access.

Inputs: graphId (required).

get_graphAInspect

Fetch one graph's full live spec — nodes, edges, and settings — as the canvas model. Requires viewer access on the graph.

Inputs: graphId (required).

get_runAInspect

Read one run's status, output, and any parked tool-call approvals. Requires viewer access on the run's graph.

Inputs: runId (required).

get_startedBInspect

Onboard the workspace end-to-end: seeds the caller's assistant graph (name + optional tone/focus) and optionally applies an assistant-shape template. Idempotent — calling again returns the existing assistant, never a duplicate.

Inputs: name (required), spec, tone, focus, +1 more.

get_usageAInspect

The caller's plan and meter: free-tier included-model calls used today vs the daily cap, hosted-key spend this month, and the upgrade URL — read-only. Scoped keys are refused (billing is account-level, not graph-level).

Inputs: no named parameters.

instantiate_templateAInspect

Instantiate an automation template (from list_templates) — validates the spec against the template's own schema, re-runs credential/origin preflight, builds the graphs + schedules on the caller's account, and returns nextSteps. Failures arrive as a structured error result naming the missing credentials/origins.

Inputs: spec (required), templateId (required).

list_graphsAInspect

List the graphs this credential can see — graphs the caller owns plus graphs shared with them, each tagged with the caller's access role and MCP slug when published.

Inputs: no named parameters.

list_templatesAInspect

The automation catalog an agent can instantiate — each entry's fields list names the spec keys instantiate_template's spec object takes, with per-field question text for conversational collection.

Inputs: no named parameters.

pause_runAInspect

Pause a run at the next step/hop boundary — resumable via resume_run, unlike cancel_run which is terminal. The in-flight model call finishes; a mid-tool-loop pause parks a checkpoint the resume continues in place, a between-hop pause parks the next hop before it starts. Requires editor access on the run's graph; already-parked or terminal runs error.

Inputs: runId (required).

provision_credentialAInspect

Store a provider API key in the caller's encrypted credential vault — the result of a conversational 'paste your key' step. Returns the stored credential's summary; the key itself is never echoed back.

Inputs: apiKey (required), provider (required), label, baseUrl.

resume_runAInspect

Resume a paused or errored run — a paused run continues its parked hop from its checkpoint, an errored one re-executes the failed node. A budget_exceeded run requires a new maxCostUsd in the arguments. Requires editor access on the run's graph.

Inputs: runId (required), maxCostUsd.

search_runsAInspect

Full-text search across previous runs' hop logs (inputs and outputs) on graphs the caller owns — the same recall the in-run search_runs tool gives agents. Returns content-ranked hits plus laneHits (recent hops of graphs whose name matches the query), with run/node/hop provenance. Pass days to bound to recent runs.

Inputs: query (required), days, limit, graphId.

share_runAInspect

Mint a read-only share/embed token for one run — the same token POST /runs/:id/share creates, embedding topology and hop status (never hop content). Requires owner or editor access.

Inputs: runId (required), ttlDays, ttlSeconds.

start_runAInspect

Launch a run on a graph — pinned mode (the saved snapshot), attributed to the caller. Returns immediately; poll get_run for status. Requires an explicit graph membership (owner or member row) and stays under the caller's monthly spend cap on that graph.

Inputs: input (required), graphId (required), context.

update_graph_nodesAInspect

Edit a graph's topology — add/remove/rename nodes and rewire edges in one call. Applies ops in order (remove edges → remove nodes → update nodes → add nodes → add edges) and reports per-op results. Refuses on the workspace's assistant graph unless force is set. Editor role required (the canvas's own bar); scoped API keys cannot call it (their graphIds can't cover topology writes).

Inputs: graphId (required), force, addEdges, addNodes, +3 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      C
      quality
      A
      maintenance
      Self-hosted, source-available AI workflow automation platform. Build multi-agent, RAG, and tool-using pipelines on a visual canvas and publish any workflow as an MCP server (stdio/SSE/Streamable HTTP). Also an MCP client via the agent node.
      2
      1,820
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI coding agents to visually build and run AI generation pipelines on a canvas by adding, wiring, and executing nodes via MCP tools.
      Apache 2.0
    • A
      license
      Not graded
      quality
      B
      maintenance
      Gives AI agents a live, persistent diagram canvas they can draw on, inspect via screenshots, and refine element by element, then export as editable .excalidraw files to commit alongside code. Supports architecture diagrams and flowcharts through a CLI, MCP tools, or REST API.
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Self-hosted visual workflow builder and MCP-native agent runtime. Connects agents to MCP servers over stdio and HTTP, and serves workflows and registered tools to other MCP clients through a gateway with per-client tokens, tool policy and approvals.
      MIT
    Try in Browser

    Glama MCP Gateway

    Add one secure layer between your agents and this server.

    Resources