MIHAD
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| memory_recallA | Search the project's persistent memory before starting work. Returns verified skills, facts and lessons from earlier sessions. Items marked provisional or unverified have NOT passed independent checks - treat them as hints only. 'preferences' are the user's standing working preferences: follow them. |
| memory_proposeA | Propose something worth remembering for future sessions. kind=skill: a reusable technique or code pattern, verified by running existing project tests. kind=fact: a fact about this codebase, verified by quoting a project file. kind=preference: how the user wants work done in future sessions; verify.quote must be the user's exact words. kind=lesson: a general lesson; stays provisional until a human approves it. Only adopted items count as verified knowledge. |
| memory_correctB | Mark a memory item as wrong. It and everything derived from it is suspended. |
| memory_report_outcomeB | At the end of a task, report which memory items you used and whether they helped or caused problems. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool targets a distinct action on memory items: recall (read), propose (create), correct (invalidate), and report_outcome (feedback). Their purposes do not overlap, and the descriptions clearly delineate when to use each.
All tools follow a consistent memory_<verb> snake_case pattern: memory_recall, memory_propose, memory_correct, memory_report_outcome. The convention is predictable and readable.
Four tools is well-scoped for a persistent memory interface, covering the essential agent interactions without redundancy. Each tool earns its place in the memory lifecycle.
The set covers read, create, invalidate, and feedback, but lacks explicit update or delete operations. Agents can work around this by proposing new items and marking old ones as wrong, so the gap is minor.