What can you do with this server?

A capture-and-recall task server for phone-dumped thoughts: log quick notes as open tasks, then list, search, close, or count them over MCP. * Capture a thought as a new open task (capture_task: text, optional source, default mcp) — same path as the Android capture box. * List open tasks newest-first (list_open_tasks: optional limit default 30, optional category like grocery, bills, people, inbox). * Search task text or category (search_tasks: query, limit default 20, include_done default false). * Complete a task by id (complete_task: task_id). * Drop a task without completing it (drop_task: task_id). * Report stats — counts by status and open categories (task_stats, no input). All tools return arbitrary JSON dictionaries (schema-unconstrained output).

Which integrations are available for this server?

Allows capturing thoughts or tasks from an Android device via a floating bubble, widget, or quick capture button, which are then sent to the server for storage and retrieval.

How do I use prospective-memory?

1. Click on "Deploy Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@prospective-memory list open tasks" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

de en es ja ko ru zh

prospective-memory

by Daemon-Killer

Overview Schema Source Related Servers Discussions Admin

TypeScript

Local

Prospective Memory

Phone-first capture. MCP pull. Not Takeout.

Dump a thought in a few seconds on Android. Later ask Grok: “what’s open?”
Phone JARVIS (v0.7): OTPs, WhatsApp previews, mail/SMS — pull from the laptop, home on the phone.

Sibling of google-activity-assistant — same kind of MCP host, different store.

Loop

Android capture  --POST /v1/capture-------->  FastAPI (Railway or PC)
Android JARVIS   --POST /v1/jarvis/events--/         │
  (OTP / WhatsApp / mail / SMS)                      │
Grok  --MCP list_open_tasks / latest_otp / last_whatsapp / missed_summary

The phone owns personal life (OTP, WhatsApp, mail). The laptop asks the phone's inbox. OTPs expire in 3 minutes and are not stored as tasks.

Related MCP server: genmem

Server (Windows)

cd $env:USERPROFILE\Desktop\prospective-memory
uv sync
uv run pmem token          # print X-PMEM-TOKEN for the phone
uv run pmem serve          # http://0.0.0.0:8790

Leave serve running. In another terminal / Grok, MCP serve-mcp is already wired if you add the config below.

Phone must reach this machine:

Same Wi‑Fi: http://YOUR_LAN_IP:8790
Out of house: Cloudflare Tunnel / Tailscale (see docs/V1.md)

Android

Open android/ in Android Studio (or gradlew.bat assembleDebug).
Set server URL + token on first launch (same token as pmem token).
Save URL + token. Add lingo lines (d=dahi lena).
Enable floating + tile (overlay permission) — tap + → type → Enter. Drag the + onto the bottom ✕ to hide it; Enable turns it back on.
Home screen widget (stretch wide for a type-here bar) or long-press app icon → Capture. Keyboard opens immediately.
Empty Enter repeats the last lingo chip. Toast shows category · text. Offline posts queue and flush when the network is back.
Optional Write (handwriting test) is in android/.../ink/. First open downloads an on-device model. Not on the bubble. Drop it via ink/REMOVE.txt.
Phone JARVIS: tap Turn on phone JARVIS → enable Capture JARVIS in notification access. If the toggle is greyed out (sideload), App info → three dots → Allow restricted settings, then try again. A heads-up with Copy appears when an OTP is seen.

Allow HTTP cleartext for LAN IPs (already in network security config).

MCP (Grok)

~/.grok/config.toml:

[mcp_servers.prospective_memory]
command = "C:\\Users\\bda99\\.local\\bin\\uv.exe"
args = [
  "run",
  "--directory",
  "C:\\Users\\bda99\\Desktop\\prospective-memory",
  "pmem",
  "serve-mcp",
]
enabled = true
startup_timeout_sec = 60

Ask: “list open tasks” / “anything grocery?” / “what’s the OTP?” / “last WhatsApp from Mom” / “what did I miss?”

CLI

uv run pmem capture "dahi lena"
uv run pmem list
uv run pmem stats
uv run pmem otp
uv run pmem whatsapp
uv run pmem missed

Non-goals

Unofficial WhatsApp protocol / full chat history (notification previews only)
UPI send, unsupervised mail send
OS reminders, geofence
Merging with Google Takeout activity
Multi-user cloud

Available Tools

17 tools

capture_cultural_itemB

Capture a cultural or leisure recommendation (movie, show, documentary, book) into the ledger.

Inputs: title (required), notes, genres, creator, +5 more. Returns: object

capture_taskA

Save a thought as an open task (same path as the Android capture box).

Inputs: text (required), source. Returns: object

complete_reminderC

Mark a reminder completed in the unified ledger.

Inputs: reminder_id (required). Returns: object

complete_taskB

Mark a task done.

Inputs: task_id (required). Returns: object

create_reminderC

Create a new reminder in the unified ledger. If armed=false, creates an unarmed inbox item.

Inputs: title (required), armed, notes, due_date, +1 more. Returns: object

drop_taskB

Drop a task without completing it.

Inputs: task_id (required). Returns: object

last_whatsappA

Recent WhatsApp notification previews from the phone. Optional sender name filter (e.g. Mom). This is the notification shade, not full chat history.

Inputs: limit, sender. Returns: object

latest_otpA

Newest OTP from the phone (SMS/bank/mail/WhatsApp notification). Expires in ~3 minutes. If redacted=true, Android hid the digits — tell the user to look at the phone. Never invent a code.

Inputs: no named parameters. Returns: object

list_cultural_itemsC

List movie and cultural recommendations queued in the leisure ledger.

Inputs: genre, limit, query, status, +2 more. Returns: object

list_open_tasksB

List open tasks, newest first. Optional category: grocery, bills, people, inbox, …

Inputs: limit, category. Returns: object

list_remindersB

List reminders from the unified Remy Reminders ledger (pending, snoozed, completed).

Inputs: limit, status. Returns: object

missed_summaryA

What arrived on the phone recently: WhatsApp, mail, SMS, OTPs (codes not included). Default last 60 minutes.

Inputs: minutes. Returns: object

play_musicB

Start zero context-switch background music playback for a query (e.g. 'play spb songs hindi', 'play lofi').

Inputs: query (required). Returns: object

resolve_music_queryA

Resolve a natural language music query (e.g. 'play spb songs hindi', 'play lofi') to stream metadata.

Inputs: query (required). Returns: object

search_tasksA

Search task text/category. By default only open items.

Inputs: query (required), limit, include_done. Returns: object

snooze_reminderC

Snooze a reminder to a target ISO date/time or by N minutes.

Inputs: reminder_id (required), minutes, target_iso. Returns: object

task_statsB

Counts by status and open categories. Includes live JARVIS notification counts.

Inputs: no named parameters. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

Deploy Server

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.3/5.0

Scored 2026-10-30 15:35 across 17 tools

Disambiguation4/5

Tools are mostly cleanly separated by domain: reminders, tasks, phone notifications, cultural items, and music. There is some conceptual overlap between reminders and tasks, and between resolve_music_query and play_music, but descriptions make the intended boundaries workable.

Naming Consistency4/5

Most action tools follow a snake_case verb_noun pattern (create_reminder, list_open_tasks, complete_task, play_music). A few read tools use noun-first labels such as task_stats, latest_otp, last_whatsapp, and missed_summary, and create/capture are used inconsistently for similar creation actions, but naming remains readable.

Tool Count4/5

At 17 tools the surface is slightly above the ideal 3-15 range, but it covers five distinct subdomains: tasks, reminders, notifications, cultural items, and music. Each tool appears to earn its place rather than being redundant bulk.

Completeness3/5

Core flows for tasks and reminders are present: create, list, complete, and for tasks also search/drop/stats. However, there are notable lifecycle gaps: reminders and tasks lack explicit update/edit and delete operations, cultural items can be captured/listed but not marked consumed, and music playback has no stop/control counterpart.

Maintenance

ActivityMaintained

ResponsivenessNo recent community issues

Resources

Related MCP Connectors

Motecloud Memory
io.github.motecloud
Cloud-hosted MCP server for durable AI memory
Knowledge & Memory Cloud Platforms
2
https://motecloud.io/mcp
hjarniA
com.hjarni
Markdown-based note-taking with a hosted MCP server. Your notes serve you and your AI.
Knowledge & Memory Note Taking Workplace & Productivity
4131
https://hjarni.com/mcp
cortexpadA
com.cortexpad
Google Keep-style notes app with an MCP server for AI agents to read/write notes.
Note Taking Knowledge & Memory
162
https://api.cortexpad.com/mcp
Engram
app.getengram
Persistent, verbatim, searchable memory for AI assistants — one memory across every MCP client.
Knowledge & Memory AI & Machine Learning
122
https://mcp.getengram.app/mcp

Related MCP Servers

nudge
Project Management Workplace & Productivity
davesleal
A
license
A
quality
D
maintenance
An MCP server that connects AI assistants like Claude to your todo app, enabling natural, friendly reminders and task management without a server to run.
Updated 8 months ago (2026-03-03 21:24 UTC)
8
12 npm
MIT
genmem
Knowledge & Memory Note Taking
helloSobuj
A
license
Not graded
quality
C
maintenance
A local-first MCP server that gives AI assistants long-term memory by storing, searching, and recalling notes as Markdown files on your machine.
Updated 5 months ago (2026-06-11 09:08 UTC)
14 npm
MIT
cortex
Knowledge & Memory Note Taking Search
moguls753
A
license
Not graded
quality
B
maintenance
A self-hosted personal memory assistant that captures and retrieves thoughts via voice or text, with MCP tools for querying and managing entries.
Updated 2 months ago (2026-08-30 21:02 UTC)
98 npm
AGPL 3.0
open-brain-mcp
Knowledge & Memory RAG Systems Search
uwek
A
license
Not graded
quality
D
maintenance
An MCP server for storing and retrieving thoughts, notes, and ideas with semantic vector search, enabling integration with Claude Desktop and other MCP clients.
Updated 8 months ago (2026-03-05 22:17 UTC)
MIT