Oversized MCP result: Cline cache URI vs Sume 256 KiB limit
Cline caches oversized MCP output behind a cline://cache URI. Sume instead refuses at 256 KiB with mcp_output_too_large. Re-read narrower; never resubmit.

They are two different mechanisms. Cline SDK v0.0.89 keeps an oversized MCP result in a per-session cache and hands the model a preview plus a cline://cache/... URI. Sume's hosted MCP server describes no such cache: a result over 256 KiB comes back as an explicit mcp_output_too_large error, and the fix is a narrower or paginated read, never a resubmitted create.
Cline facts are from its releases page; Sume facts from MCP and the server source, read 2026-10-01.
What does Cline do with an oversized MCP result?
The v0.0.89 note says oversized MCP and Composio results are cached in a per-session in-memory cache. The model receives a bounded preview and a cline://cache/... URI that read_files can page through by line range. Entries expire after five model iterations without a read, the cache is capped at 16 MiB per session, and the original output stays in history and tool events.
What does Sume do with an oversized result?
The hosted server defines one cap, MCP_OUTPUT_MAX_BYTES = 256 * 1024. Above it the tool returns a JSON result with code mcp_output_too_large, the max_bytes value, and this message: use a narrower or paginated read; this is an output limit, not a job failure; never resubmit a paid create.
| Cline v0.0.89 | Sume hosted MCP | |
|---|---|---|
| Trigger | Result judged oversized | Result over 256 KiB |
| What the model gets | Bounded preview plus a cline://cache/... URI | mcp_output_too_large with max_bytes |
| Full output | Cached per session, up to 16 MiB | Not returned |
| Next step | Page with read_files by line range | Narrower or paginated read |
Which one applies when Cline calls Sume?
Both can be in play, in order. Sume's cap applies first, on the server, and the Cline cache only sees what Sume returned. So an over-cap Sume result reaches Cline as the small error above, and there is no full output to page through. Point Cline at the hosted URL, https://mcp.sume.com/mcp, as the docs describe for remote MCP clients; see also Cline setup for Sume.
What should I re-read after the error?
Ask for less: a smaller page, fewer ids, or a single job. For job output, read by id with jobs_status or jobs_result instead of listing everything. The error is not a failed generation, so the paid job that produced the data is untouched. Resubmitting the create would run, and bill, a second job. The tools and gates page lists the read tools.
Sources
Related posts
More in Integrations
- Shopify events: event-id header gone, dedupe on Sume job_id
Shopify events no longer send shopify-event-id. On the Sume side, dedupe job webhooks on job_id and send an Idempotency-Key on every paid submit that may retry.
- Shopify product.variants.* trigger: one Sume image per variant
A product.variants.* Shopify events trigger can fire many times at once. Fan each event out to one Sume image job, and queue bursts on your side.
- Slack incoming webhook 1/sec: when 100 Sume jobs finish together
Slack incoming webhooks allow about 1 message per second. When 100 Sume jobs finish at once, store each webhook, then post to Slack from a paced queue.
- Snapchat Ads MCP and hosted Sume tools in one agent
Snap's Ads MCP server answers campaign questions and is read-only at launch. Add Sume's hosted MCP in the same agent to turn the findings into creative.
Written by Sume