lifedb-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| LIFEDB_DB | No | Path to your LifeDB SQLite timeline database, e.g. /path/to/timeline.db |
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 |
|---|---|
| lifedb_statsA | 库自检: schema 版本、各模态/来源计数、时间覆盖。应用接入后先调这个。 |
| lifedb_query_monthC | 某月时间线。month 形如 '2026-10'。返回统一 schema 事件行 (ts_start 升序)。 |
| lifedb_aroundB | 跨模态事件窗: ts 前后 window (如 '30m'/'2h'/'1d') 内发生的所有事件。 三分支 UNION, 全索引 seek。用于'那一刻发生了什么'。 |
| lifedb_geo_trackA | 原始 GPS 点序列 [ts(epoch秒), lat, lon, alt, speed]。 start/end 为 ISO 时间或 epoch 秒。轨迹渲染 (deck.gl TripsLayer / Manim) 喂这个。 limit 截断时按均匀抽稀前置 (取首尾), 避免只截头部。 |
| lifedb_export_atlasB | 一站式打包某时间窗的 atlas 数据: {track(抽稀GPS), events(逗号分隔的模态 名可选过滤), photos(带坐标照片)}。地图类应用 (Life Atlas/vlog) 的数据源。 |
| lifedb_ingest_jsonlA | 导入统一 JSONL 事件文件 (每行一个 JSON, ts_start/modality/source 必填)。 幂等: payload_hash=sha256(source|payload_ref|ts_start)[:16] 去重, 重复跑零副作用。 默认 dry_run=True 只报预览; 明确传 dry_run=false 才写入。 |
| lifedb_adaptersA | 列出全部数据源 adapter (adapters/*.manifest.json): 名称/模态/模式/幂等机制。 manifest 即插件契约, 见 docs/PRODUCT.md §3。 |
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 7 tools
Most tools have a clear distinct intent: stats (metadata self-check), query_month (monthly timeline), around (ts-window union), geo_track (GPS-only points), export_atlas (bundled atlas for an app), ingest_jsonl (write), adapters (plugin listing). There is mild overlap between around, geo_track, and export_atlas since all read time-windowed event/GPS data, but the granularity (point vs month vs track vs bundle) keeps them separable.
All names share the lifedb_ prefix, giving a strong consistent identity. However the verb_noun pattern is not uniform: query_month, geo_track, export_atlas, ingest_jsonl follow it, while stats, around, and adapters are bare noun/adverb forms.
Seven tools is well-scoped for a personal life-database server: four read/query tools, one export, one ingest, and one metadata/introspection tool. Each tool appears to earn its place without redundancy.
The read surface covers timelines, point windows, GPS tracks, and atlas bundles, and ingest is idempotent, but there is no update/delete or prune operation for events, no fetch-by-id, and no free-form range or modality/text search beyond month windows and the atlas export. These gaps force workarounds for correction and targeted retrieval.