Documentation menu
Slack-native memory

Install the Anansi Slack app and your team's memory builds itself — no API calls, no glue code. A working showcase of the same engine behind the API.

Anansi reads the public channels you choose and synthesizes two layers automatically:

  • Team memory — a shared, synthesized profile of what your whole team knows, decides, and is working on.
  • Per-person memory — for each teammate, a personal profile and a bi-temporal view of how their context changes over time (roles, projects, tools, who they work with). This is the same engine behind the entity graph and point-in-time queries — populated straight from Slack.

Setup

1

Install the Slack app

Complete the OAuth flow and pick which public channels to index.

2

Backfill runs automatically

Anansi backfills recent history, then keeps memory current as people post.

3

Ask questions

Use /ask for answers from team memory, or /memory to inspect what's stored.

Only channels you select are read, and only the bot's channels. Direct messages are never ingested.

Commands

CommandWhat it does
/ask <question>Answer from team memory, in-channel
/ask quietly <question>Same, but only you see the answer
/memoryMemory status: chunks, last synthesis, channel backfill
/memory meYour synthesized profile — facts, current focus, timeline, connections
/memory about @userA teammate's profile (only you see it)
/memory forget-meOpt out — purge your personal profile, stop building one
/memory remember-meOpt back in
/memory dashboardDM yourself a magic link to the web dashboard
/memory purge [#channel] [confirm]Admin: delete indexed memory

/memory me renders the personal profile Anansi has synthesized for you:

text
Your memory Known facts • Senior backend engineer; owns the payments service • Prefers TypeScript and pnpm Currently • Migrating webhooks to the new retry queue Timeline • Works on Payments (since 2024-03) ✓ • Worked on Onboarding (2023-06 – 2024-03) Connections • you works with Dana • you uses PostgreSQL

Everything here is derived from what you've said in indexed public channels.

Privacy

Per-person memory is built only from public channel messages Anansi is invited to — never DMs, never private channels it isn't in. Still, anyone can opt out of having a personal profile with /memory forget-me. This immediately:

  • deletes your synthesized personal profile and your entries in the entity graph,
  • detaches your past messages from your personal memory (they remain part of the team profile, since they were public), and
  • stops Anansi attributing future messages to you.

Opt back in anytime with /memory remember-me. Admins can purge memory at the workspace or channel level with /memory purge.

How it relates to the API

Slack-native memory uses the same store and synthesis as the developer API. A Slack workspace is provisioned a developer account behind the scenes, so per-person Slack memory is queryable through the same GET /v1/context and GET /v1/entities endpoints (including asOf / asOfKnowledge) once you have an API key — Slack and API memory are one graph, not two.