Agent Memory Engine
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 | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| memory_batch_inspectB | Fetch complete High-Density Identity Cards for multiple folders and files in a single call (Multi-Query). |
| memory_find_by_intentA | Search symbols, files, or exact code blocks using natural human language, Indonesian colloquial slang, command habits, or typos. |
| memory_get_phase_mapB | Retrieve detailed internal logic execution phase map for a script from start to finish without reading raw code. |
| memory_enrich_knowledgeC | Save deep understanding of a file/folder or function/symbol. MANDATORY: Write technical specs (purpose, logic_phases, upstream, downstream, guardrails, side_effects, error_handling) in High-Density Technical English (saves 40% tokens). MANDATORY: Write colloquial_aliases in natural Indonesian user habits/slang for instant FTS5 matching. |
| memory_sync_codebaseB | Automatically sync codebase via AST parsing and SHA-256 hashing to detect exact line and symbol changes. |
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 5 tools
Each tool has a fairly distinct role: bulk read (batch_inspect), search (find_by_intent), deep read of one script's logic (get_phase_map), sync/index (sync_codebase), and write (enrich_knowledge). The only mild overlap is between batch_inspect and get_phase_map, since both return stored knowledge about code, but the granularity difference (identity cards vs. execution phases) is stated clearly enough to choose correctly.
All five tools share a clear `memory_` prefix and use snake_case, giving a predictable namespace. The verb ordering varies slightly (adjective+verb in batch_inspect, verb+preposition in find_by_intent, verb+noun elsewhere), but each name remains self-descriptive and readable.
Five tools is lean but each covers a distinct stage of the memory lifecycle (sync, search, bulk read, deep read, write). It stays well under the heavy-tool threshold and no tool feels redundant, though the surface is minimal for a full memory engine.
The read/create/sync path is covered, but there is no delete or invalidate operation for stale knowledge, no explicit update/amend for existing entries, and no listing/index tool to discover what folders are stored. Agents can work around some of this via search and re-enrich, but lifecycle coverage has notable holes.