memory-garden
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MG_PUBLIC_DEMO_MODE | No | Set to true to run in public demo mode using the bundled synthetic vault (evals/cognitive_mvp_vault), enabling an offline demo without requiring a personal Obsidian Vault or API key. | false |
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
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_sourcesA | 按关键词与时间窗检索用户本人的记录候选，返回 [A{id}] 编号。 |
| read_sourceA | 读取本轮已发现来源的原文与定位信息（引用前必须先读）。 |
| get_topic_timelineB | 按时间整理同一主题的历次表达；最近一条只是 recent 候选。 |
| find_change_candidatesA | 围绕明确主题提出变化候选端点对（最早端点 × 最近端点）；措辞差异不等于变化。 |
| discover_cognitive_shiftsC | 全库发现可能被忽视的变化候选；较近记录只是 recent 候选，需用户确认。 |
| find_interval_eventsB | 在两个日期端点构成的区间内搜索经历/决定/事件候选。时间相邻不等于因果。 |
| search_hypothesis_evidenceA | 对原因假设分侧检索：stance=support 或 challenge；两侧都必须调用。 |
| get_user_verdictsA | 读取用户对该主题此前的确认/否认判定；被否认的解释不得复用。 |
| ask_gardenB | 运行完整认知回溯 Agent（本地确定性路径），返回带引用的结构化结论。 |
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 9 tools
Most tools have clear roles, but discover_cognitive_shifts and find_change_candidates both surface 'change candidates' and could be confused; search_sources and find_interval_events also overlap in time-based retrieval. The descriptions help clarify scope, but the boundaries are not always immediately obvious.
All names use lowercase snake_case and mostly follow a verb_noun pattern such as search_sources, get_topic_timeline, and find_interval_events. Minor inconsistency exists between verbs like discover, find, and search for similar retrieval actions, but the overall pattern remains predictable.
Nine tools is a well-scoped size for a cognitive retrospection domain. Each tool contributes to a distinct stage of the workflow, from source discovery through evidence retrieval to the final ask_garden synthesis, with no obvious bloat.
The toolset covers the main cognitive backtracking pipeline: search sources, read them, build timelines, identify change candidates, find interval events, gather hypothesis evidence, and check user verdicts. A minor gap is that there is no explicit tool to record or update user confirmations/verdicts, but this may be handled outside the server.