MCP server for MemoryOps - a governed agent memory system with memory ingestion, retrieval, processing, and tool management capabilities
MemoryOps exposes 14 tools with complete input schemas and descriptions. Naming follows verb_noun convention (create_tool, list_tools, get_tool, etc.), which is strong. However, descriptions are terse (avg ~80 chars, well below the 194-char baseline for A+ tools), lacking context on WHEN to use each tool and what distinguishes similar operations. Parameter descriptions exist but are minimal ('Tool name (required)' vs 'The unique identifier for the tool within the workspace; used in subsequent get_tool and invoke_tool calls'). Output schemas are not documented, callers cannot see what fields to expect from list_tools or get_tool_version. Error handling is absent from the visible code; no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Security concern: get_tool_secret exposes plaintext auth secrets, violating secret-injection pattern. Composition is sound, each tool has one responsibility, but the interface lacks the richness needed for confident LLM planning.
Create a new workspace tool with name, description, endpoint URL, HTTP method, input/output schemas, authentication, and governance settings
Delete a workspace tool
Export all workspace tools in a portable format
Retrieve a specific workspace tool by name
Retrieve the plaintext authentication secret for a workspace tool
Retrieve a specific version of a workspace tool
Output schemas not documented. Callers cannot see what fields list_tools, get_tool, or invoke_tool return. LLMs cannot plan downstream operations or extract required IDs for chaining.
get_tool_secret exposes plaintext authentication secrets as tool output. Secrets in responses enter LLM context and risk logging/echoing to users. Violates secret-injection pattern.
Descriptions are terse (avg 80 chars vs 194-char baseline). Missing context on WHEN to use each tool, prerequisites, and distinctions between similar operations (e.g., test_tool vs invoke_tool).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 64 | 2026-07-28+ | v2 |
Import workspace tools from a portable format with optional overwrite
Invoke a workspace tool with a request body and optional headers
List invocations of a workspace tool with optional limit
List all versions of a workspace tool
List workspace tools with pagination support
Rollback a workspace tool to a previous version
Test a workspace tool by invoking it with a request body and optional headers
Update a workspace tool with optional fields for description, endpoint URL, HTTP method, schemas, authentication, and governance settings
No error handling guidance. Tools lack recovery hints, retryability classification, or actionable error messages. LLMs cannot self-correct on failures.
Destructive operations (delete_tool, import_tools with overwrite=true) lack confirmation or dry-run support. Agents can accidentally delete tools or overwrite configurations without safeguards.