Old prompt says video-router_create? Sume now takes generate_video
Sume retired the hosted MCP tool name video-router_create for generate_video and maps dots to underscores. Update the prompt, then check tools_list.

Replace video-router_create with generate_video in the prompt. Sume's tools-and-gates page (read 2026-10-08) lists two retired aliases: image-generations_create became generate_image, and video-router_create became generate_video. The live tool ids are the underscore names in tools_list, so confirm what your session can see before you rewrite a long prompt.
Old names and live names
Three things happen on every call, according to Sume's MCP docs: the server canonicalizes dots to underscores, it takes dotted aliases such as tools.list, and it uses the names in the live registry. Retired aliases are different: they are not just spelled differently, they were replaced.
| Old or alias form | Live tool id | Notes |
|---|---|---|
video-router_create | generate_video | Retired alias; omit payload.model to route to sume/auto |
image-generations_create | generate_image | Retired alias |
tools.list | tools_list | Dotted aliases still work |
jobs.wait | jobs_wait | Dotted aliases still work |
Where stale names come from
Old names live in saved system prompts, notes files, rule files in an editor and agent memory. A model that reads one of them will try the old id. Search those places for video-router_create and image-generations_create, and replace them, rather than hoping the model self-corrects.
- Project rule files and saved prompts for your editor or chat client.
- Scripts that use
script_runand call tools by name throughsume.call(name, arguments). - Runbooks and wiki pages that list tool ids.
What to ask the agent after the fix
Start with discovery, not a paid call. Ask it to run tools_list, then tools_schema with name: "generate_video" and to explain idempotency_key and dry_run before submitting. Catalog ids for the model field come from video-router_models; that tool name is a read-only discovery tool and is still a live id.
Then submit with a new idempotency_key. Reusing the key from an old failed attempt with a changed payload would return 409 idempotency_conflict, so derive the key from the work item and the payload, not from a timestamp you might forget.
A session without the tool
If generate_video does not appear in tools_list at all, the session is probably read-only: OAuth with Write off hides paid tools. Reconnect with Write on, or use an API key. Remember that Sume Video 1.0 (videos_create) is REST-only and never appears on hosted MCP, so an old prompt that mentions it needs the Developer API, not a rename.
A short migration checklist
Work through these in order; each takes a minute:
- Search saved prompts and rule files for
video-router_createandimage-generations_create. - Replace them with
generate_videoandgenerate_image. - Run
tools_listin a fresh session and confirm both names appear (they need Write or a key). - Call
tools_schemafor each and compare the fields with the old prompt: field names can differ from older examples, and the schema is the contract. - Submit one
dry_runcall per tool before any paid run.
Find the old name
Search your prompts, skills and saved agent instructions for video-router_create and image-generations_create, and replace them with generate_video and generate_image. Then run tools_list once to confirm what the session exposes.
Sources
Related posts
More in Agents
- What an agent should log when it calls Sume: ids yes, signed URLs no
Safe log fields for agents using the Sume API, CLI and MCP: request ids, job ids, status, sanitized media metadata. Never keys, signed URLs or transcripts.
- Run the Sume video agent from your backend with Agent Completions
POST /v1/agent/completions runs the same agent as the Sume Agents chat, with tools and media generation, and returns an async run receipt you poll or webhook.
- Safe automation for AI agents that call paid APIs
Keep agents read-only by default, keep secrets out of logs, and on hosted MCP send an idempotency_key, preview with dry_run, and cap with max_spend_usd.
- Scheduled AI video agent runs: cron, API triggers, and receipts
A Sume schedule is a saved Agents automation that runs on a cron cadence and returns a run receipt. Author it in the dashboard; start and monitor runs by API.
Written by Sume