Model Context Protocol server that exposes Gamma Generations API tools.
The server defines 4 tools with complete input schemas and clear descriptions. Naming follows verb_noun convention (gamma_*). However, there are significant gaps in output schema documentation, parameter descriptions lack critical constraints, and error handling is minimal. The tool descriptions are adequate but could be more prescriptive about when to use each tool and what the agent should expect.
Create a new Gamma deck generation from a prompt, template, or structured payload.
Download an asset (such as a PDF or image) associated with a completed Gamma generation.
Retrieve the current state of a Gamma generation by identifier.
List recent Gamma generations with optional status filtering.
No output schema documentation. Tools return raw API responses or JSON blobs without documenting what fields the LLM should expect. gamma_create_generation returns JSON.stringify(response) with no structured schema definition. This forces LLMs to guess what fields exist and breaks downstream tool chaining.
gamma_list_generations lacks pagination documentation and limits. The 'limit' parameter has no min/max constraints. No indication of what happens when results exceed limit or how to fetch next page. This violates paginated-result pattern which requires explicit offset/cursor guidance.
Parameter descriptions lack critical constraints. Example: 'numCards' states 'Pro users: 1-50, Ultra users: 1-75' but does not explain how the server determines user tier or what happens if an invalid value is passed. 'additionalInstructions' mentions 'Character limit is 1-500' but the schema has no maxLength constraint. LLMs cannot enforce these rules client-side.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 51 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 51 | - | v1 |
Error handling is minimal. The callGamma function returns generic 'Gamma API request failed' messages with status code and optional JSON detail. There is no categorization of errors as retryable vs user-fixable. No recovery guidance. An LLM cannot determine whether to retry, ask the user, or abort.
gamma_create_generation accepts 10+ optional parameters with no documentation of mutual exclusivity or dependency rules. For example, 'format' enum and 'exportAs' enum appear independent but the description does not clarify if certain format+exportAs combinations are invalid. Undocumented dependencies cause silent misuse.
No dry-run or confirmation pattern for gamma_create_generation. This is a WRITE tool that generates content (risk:WRITE). The description does not mention whether the generation is immediate, queued, or requires confirmation. Agents cannot assess whether this is safe to call without verification.
gamma_create_generation error handling does not validate 'inputText' until runtime. The code checks 'if (!inputText)' and returns error, but the parameter is marked optional in the schema. This creates an inconsistency: the schema says optional, but the tool requires it. LLMs will omit the parameter because the schema permits it.
No tool annotations for read-only vs destructive behavior. The MCP spec (2026-07-28) supports readOnlyHint, destructiveHint, and idempotentHint. gamma_create_generation is clearly destructive (creates content, marked risk:WRITE) and gamma_get_generation is read-only, but these are not declared in the schema.