Skip to main content

Documentation Index

Fetch the complete documentation index at: https://datost.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

A Log Connection is a configured link between Datost and an external operational data source — typically an observability, product analytics, or CRM API. Unlike warehouse connections, which let the agent write SQL, Log Connections give Datost a typed set of tools to query services like Datadog, Sentry, PostHog, Axiom, HubSpot, Monday.com, Salesforce, QuickBooks, Zoho, and Granola directly through their APIs.

​
What Datost uses them for

Log Connections let the agent answer questions that warehouse data alone cannot:
  • Incident and error investigation. Pull recent Sentry issues, Datadog logs, or Axiom events when a metric spikes or a user reports a bug.
  • Product analytics lookups. Query PostHog funnels, events, and person properties alongside warehouse revenue data.
  • CRM context for revenue questions. Cross-reference HubSpot, Salesforce, or Zoho records when someone asks about a specific deal, contact, or pipeline stage.
  • Ops and finance context. Read QuickBooks invoices, Monday boards, or Granola meeting notes to enrich answers with operational ground truth.
Log Connections are read-only tools. Datost queries the external API to answer a question, then formats the results inline in Slack — it does not sync or copy data into your warehouse.

​
Supported connection types

Each type has its own auth model, configured once per connection:
  • API key: Datadog (site + API key + App key), Sentry (org slug + auth token), Axiom (API token + optional dataset), Granola, Monday.com, PostHog (personal API key).
  • OAuth: HubSpot, Salesforce, QuickBooks, Zoho, PostHog (OAuth flow).

​
Connect a source

1

Open Log Connections

In the Datost web app, go to Settings → Log Connections and click Add connection. Pick the type you want to connect.
2

Provide credentials

For API-key types, paste the required fields (for example, Datadog site datadoghq.com, API key, and App key). For OAuth types, click Connect and complete the provider’s consent screen — Datost stores refresh tokens encrypted and rotates access tokens automatically.
3

Add a description

Write a short description of what this connection covers (e.g. “Production Datadog — API and worker services”). The agent reads this to decide when to reach for the tool.
4

Scope access with roles

Optionally restrict the connection to specific custom roles so only certain Slack channels or users can trigger queries against it.

​
Query format

You do not write queries yourself. Datost exposes each connection as a set of typed tools — for example, datadog.search_logs, sentry.list_issues, posthog.query_events, hubspot.search_contacts. The agent picks the right tool and parameters based on the Slack question, then summarizes results in thread.
Secrets never leave Datost’s encrypted secret store. Rotate the underlying API key or OAuth grant in the source system if a credential is compromised — Datost will surface refresh errors in the connection list.

​
Limitations

  • Rate limits are the provider’s. A very chatty question may hit Datadog or HubSpot quotas; Datost surfaces the error rather than retrying aggressively.
  • Historical depth is whatever the provider exposes via API (for example, Datadog log retention on your plan).
  • Only the connection types listed above are supported today. Arbitrary custom HTTP endpoints are not yet available — reach out if you need one.