Integrations

Connect external tools

Integrations let ContextStream pull context from your team's tools and turn it into searchable memory events for your workspace and projects.

Integrations are per-workspace

Each ContextStream workspace can connect to different external accounts (e.g., different Slack workspaces). Configure integrations from your workspace settings.

Why Connect Integrations?

Integrations automatically enrich your AI context. Once connected, your existing prompts become smarter without any extra effort.

Automatic Enrichment

context_smart and session(action="smart_search") automatically include relevant GitHub issues, Slack threads, and team decisions.

Knowledge Extraction

Decisions, lessons, and insights are automatically distilled from high-value discussions and made searchable.

Cross-Source Search

Search by meaning across GitHub repos, Slack channels, and your captured decisions—all in one query.

Example: Ask your AI "What did we decide about authentication?"

Your AI will automatically surface: decisions from GitHub issue discussions, relevant Slack threads, and your manually captured memory—all without calling any integration-specific tools.

How Context Enrichment Works

When you connect integrations, your AI tools become smarter automatically. Here's the flow:

1

Integration Syncs Data

GitHub issues, PRs, releases, and comments (or Slack messages and threads) are synced to your workspace as memory events.

2

Semantic Indexing

Each memory event is indexed with embeddings for semantic search. You can search by meaning, not just keywords.

3

Knowledge Extraction

High-engagement content (threads with many reactions, issues with lots of comments) is analyzed to extract decisions, lessons, and insights.

4

Automatic Context

When you call context_smart, it searches all memory including integration data. Relevant GitHub issues and Slack discussions surface automatically in AI responses.

No extra tools needed: Once connected, session_init reports which integrations are available, and context_smart includes relevant integration data in every response. Use integration(provider="github|slack|all", action="...") only when you need specific queries (like "show all open issues" or "search Slack for deployment").

What Integration Data Looks Like

When integrations are connected, session_init tells your AI what sources are available:

{
  "connected_integrations": ["github", "slack"],
  "recent_memory": [
    { "source_type": "github", "title": "Auth flow redesign", ... },
    { "source_type": "slack", "title": "Deployment discussion", ... },
    { "source_type": "manual", "title": "User's captured note", ... }
  ]
}

Each memory item includes a source_type field (github, slack, or manual) so your AI can reference where information came from.

When you search with memory(action="search"), you can filter by source to get integration-specific results, or leave it unfiltered to search across all sources at once.

Integration MCP Tools (v0.4.x)

Beyond automatic enrichment, you can directly query integration data using the consolidated integration tool:

v0.4.x Consolidated Syntax:

integration(provider="github|slack|all", action="...", ...)

GitHub Actions

  • integration(provider="github", action="stats")
  • integration(provider="github", action="search", query="...")
  • integration(provider="github", action="activity", days=7)
  • integration(provider="github", action="knowledge")
  • integration(provider="github", action="summary", days=7)

Slack Actions

  • integration(provider="slack", action="stats")
  • integration(provider="slack", action="search", query="...")
  • integration(provider="slack", action="discussions", days=7)
  • integration(provider="slack", action="knowledge")
  • integration(provider="slack", action="summary", days=7)

Integration Status

Use integration(provider="all", action="status") to check sync status, last sync time, and health of all connected integrations.

Cross-Source Search

Use integration(provider="all", action="search", query="...") to search across all connected integrations in a single query. Results are ranked by relevance across GitHub and Slack.

Unified Summary

Use integration(provider="all", action="summary", days=7) to get a combined summary across GitHub and Slack with activity stats, key highlights, and trending topics.

Unified Knowledge

Use integration(provider="all", action="knowledge") to get decisions, lessons, facts, and insights extracted from all connected integrations in a single query.

Where integrations show up

Integrated data becomes Memory Events (with a source like github or slack). Your AI tools can then retrieve it via MCP and search.

ChatGPT and GPT are trademarks of OpenAI, Inc. GitHub is a trademark of GitHub, Inc. Slack is a trademark of Slack Technologies, LLC, a Salesforce company. ContextStream is not affiliated with, endorsed by, or sponsored by OpenAI, Inc., GitHub, Inc., Slack Technologies, LLC, or Salesforce, Inc.