Codepanion documentation

Create an API key in the app, store it in your CI system, and rotate it with zero downtime. Then invite your team and set what each person can reach.

Tar what git tracks and POST it with plain curl, with GitHub Actions examples for both an API key and keyless OIDC.

Read-only access to SQL Server or PostgreSQL: connection strings, dedicated users, and query plans.

Label database connections, code snapshots, and log sources per environment, pin an investigation to one, and every tool defaults to it.

Connect Seq directly, or Seq, Azure Monitor and CloudWatch Logs through the connector, so the agent can search your logs and read the entries around a moment in time.

How an investigation runs: the agent's tools, the turn streaming live, and findings that cite the exact files and lines they came from.

Filing an issue and proposing a code change are the only things the agent does that write anywhere. Both stop and wait for a person.

Query results can be piped into a sandbox in your own browser, so the agent computes over production rows that never enter the transcript.

Ask for a trend or a breakdown and the agent draws it. Save the query and the chart as a report and run it again next week.

Every tool the agent runs, every credential change, and every time a Codepanion employee opens your account. What each row holds, and how to get a copy.

Connect GitHub Issues, Jira, Linear, or Salesforce, plus Confluence as a docs source the agent can search and read.

For a database or log store that must not be reachable from our cloud: an outbound-only worker inside your network runs the agent's read-only queries locally, so no credential ever leaves it.