PocketContext
Get PocketContext

WHAT YOU CAN DO / POCKETCONTEXT 0.23.1

Browse once.
Reuse the context.

Stop searching across tabs. Let your agent gather the context, keep it in PocketContext, and help you decide what to do next.

Find your starting point →

FROM TABS TO CONTEXT

Your agent can use the browser
you’ve signed into.

With Grok Bot or a similar computer-use agent, sign in to LinkedIn, WhatsApp Web or Luma in the agent’s browser. Choose the profiles, chats or calendars it should read. The agent gathers accessible material; PocketContext keeps the useful results.

Less manual searching

Instead of opening every chat, scrolling back and copying details between tabs, ask for a catch-up brief, a person’s background or a shortlist of events.

No suitable API? A browser may be enough.

The website does not need to offer an API, an MCP connector or support for agent skills. A capable agent can work through the interface. Its access still depends on valid sessions, site restrictions and readable content.

START WITH AN OUTCOME

What would help you today?

01 / WhatsApp Web

Catch up on conversations

Turn selected group conversations into a useful brief of decisions, requests and unresolved questions.

See the workflow ↗02 / LinkedIn + captured conversations

Prepare a follow-up

Connect a person’s profile with captured conversations so you can prepare a relevant follow-up.

See the workflow ↗03 / Luma

Choose events for your week

Turn the calendars you follow into a dated shortlist with the details you need to choose.

See the workflow ↗

MAKE THE BROWSER WORK COUNT

Spend time finding changes.
Keep the context ready.

Computer use can be slow and token-intensive: your agent opens pages, scrolls, expands messages and reads screens. Repeating that journey for every question adds latency and model usage.

  1. 01 / CAPTURE

    Browse the source

    Your agent reads selected material through an authenticated browser and records what it could capture.

  2. 02 / PRESERVE

    Keep useful knowledge

    Store captured sources. Have your agent publish Markdown pages with properties, links and citations.

  3. 03 / REUSE

    Retrieve directly

    Use SQL, full-text search and authenticated REST APIs to fetch relevant stored context without navigating those screens again.

Reuse can reduce repeated browsing and token use. Capture, synthesis and refreshes also have a cost; actual savings depend on the workflow. No speed or token-saving benchmark is claimed.

READ IT LIKE A NOTE. QUERY IT LIKE DATA.

One page per person,
calendar or conversation.

Markdown holds the readable detail. Properties hold fields your agent can filter. Links connect related pages, and citations lead back to stored evidence.

ILLUSTRATIVE PAGEtranscript

Builders Circle · group conversation

source_app
whatsapp
community
Builders Circle
captured_from
2026-11-01
captured_through
2026-11-03
See the Markdown body
## Captured messages
Dated messages, with speaker labels.

## Capture gaps
Attachments and unavailable history are listed here.

Properties for querying. Markdown for the detail. Citations and links connect the evidence.

ILLUSTRATIVE PAGEentity

Alex · Example Studio

source_app
linkedin
organization
Example Studio
role
Operations
captured_at
2026-11-03
See the Markdown body
## Professional background
A readable profile based on captured material.

## Conversation context
Relevant discussions, with evidence and links.

Properties for querying. Markdown for the detail. Citations and links connect the evidence.

ILLUSTRATIVE PAGEentity

City Builders · followed calendar

source_app
luma
city
Berlin
followed
true
captured_at
2026-11-03
See the Markdown body
## About this calendar
Community description and source link.

## Upcoming events
Captured listings, dates and availability, with evidence.

Properties for querying. Markdown for the detail. Citations and links connect the evidence.

Fictional examples with suggested property conventions, not built-in service schemas. Pages, properties and history live in the database; they are not standalone Markdown files by default. Your agent creates and verifies relationships.

FOR YOUR AGENT. FOR YOU.

Fast retrieval for the next question.
A reader for the full picture.

Your agent can query properties with read-only SQL, search pages and source passages, and use authenticated APIs to retrieve evidence and stage updates.

You can open the included browser reader to inspect published Markdown pages, properties, citations, original sources and revision history. The stored result is something you can return to, beyond the latest chat response.

Explore the reader and setup guide →

ILLUSTRATIVE READER CONTENT

A follow-up worth preparing

A captured conversation mentions interest in a product demo. The person brief preserves the context; the next step is to check whether that interest is still current.

Show the supporting excerpt ↗

Fictional conversation: “I’d like to see how this could help our weekly reporting.”

This supports interest at capture time, not a current commitment.

Published page · Source passage · Earlier revisions

KEEP IMPORTANT CONTEXT FRESH

Let a routine do the next round.

An agent scheduler, such as Grok Bot routines ↗, can run a workflow on a schedule. Ask your agent to revisit selected pages each night and save accessible changes, so the next question can start with the stored context.

“Each night, check these selected WhatsApp chats, LinkedIn conversations and Luma calendars. Capture new material, update the related PocketContext pages, and summarize changes that need my attention. Record when you checked and report anything you couldn’t read.”

Start small, then schedule

Choose a few important sources. Run the capture once, inspect its coverage and cost, then ask your agent to schedule the proven workflow. Revisit the source when freshness matters.

Know who does what

The agent provides browsing and scheduling. PocketContext stores evidence, publishes revisions and serves queries. It has no built-in website connectors or scheduler. Expired sessions and failed reads need to be reported.

This nightly capture is a proposed workflow, not an end-to-end tested integration. External AI services receive the context the agent sends.

CONTEXT THAT CARRIES FORWARD

Your next agent can start here.

Configure your agents to use the same PocketContext project. One can collect selected conversations; another can retrieve the published knowledge and its evidence. They do not need to rebuild separate collections.

Read and ask from your phone

Ask your setup agent for a domain with built-in HTTPS to open the PocketContext reader directly in your phone browser. Set up the reader →

For chat, ask the setup agent to enable MCP on that domain. Add the endpoint in Gemini web, approve access, then test the connection on your phone. Coding agents can also use the CLI and installed skill.

See the mobile workflows →

Remote chat comes from the agent. The PocketContext host and the required connection must remain available.

A SMALL FIRST STEP

One conversation.
A useful answer you can keep.

Set up PocketContext, bring a selected note or transcript, and ask for a brief with evidence and open questions.

Set up PocketContext →Compare with your current tools →

Examples are fictional, informed by exploratory workflows. Product scope: PocketContext 0.23.1. Read the data model ↗ · Grok Bot browser workflows ↗