Loading ToolBench
MCP server for long-form storytelling on top of OpenChronicle memory
Mnemosyne has 10 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, mostly short), and most have input schemas with type definitions. However, descriptions are sparse and lack LLM-optimization guidance (e.g., 'List all stories' vs 'Retrieve all stories in the catalog with metadata for selection'). Parameter descriptions are minimal or absent. Output schemas are not documented. Error handling is not visible in tool definitions. The server is STDIO-only, which is a hard transport limitation.
Continue a story by generating the next scene
List all entities (characters, locations, objects) in a story
Export a story or scene to a file
Import a story from a file
Get the current position in a story
Revalidate scenes in a story
Manage story session breaks and continuity
Get the readiness status of the server and its dependencies
Descriptions are too short (10-50 chars) and lack LLM-optimization guidance. E.g., 'List all stories' should be 'Retrieve all stories in the catalog with metadata (title, status, scene count) to help select which story to work with.'
Parameter descriptions are missing or minimal. E.g., mnemo_position has 'storyId' with description 'The ID of the story', lacks format guidance (UUID? string? length?).
Output schemas are not documented in tool definitions. LLMs cannot plan downstream calls or extract data without knowing what fields to expect. E.g., mnemo_stories should document: returns array of {id, title, sceneCount, status, createdAt}.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | 2026-07-28+ | v2 |
List all stories in the catalog
Validate a story for consistency and quality
No error handling guidance visible in tool definitions. Per pattern:recovery-guide, error responses must tell the LLM what to do next. E.g., 'Story not found. Try mnemo_stories() to list available stories.'
Destructive tools (mnemo_export, mnemo_import, mnemo_continue, mnemo_session) lack confirmation or dry-run support. Per pattern:confirmation-request, irreversible operations should support a confirm_before_execute pattern to prevent catastrophic errors.