Providers are the external services AgentRail connects to in order to receive issues, submit pull requests, track CI, and observe code review. In local OSS mode, you supply your own credentials and AgentRail calls provider APIs from your machine. Provider connections are optional — you can run AgentRail without any providers for local evaluation — but you need at least one configured to handle real issue intake and PR submission.Documentation Index
Fetch the complete documentation index at: https://agentrail.app/docs/llms.txt
Use this file to discover all available pages before exploring further.
Connecting a provider
Use the interactive setup command for any provider:provider connect command:
- Prompts for credentials in a hidden input (your secrets are not echoed to the terminal).
- Writes the credentials to a local env file under
~/.agentrail. - Runs a full provider readiness check — the same check used by
agentrail provider doctor. - Applies safe local fixes where it can.
- Reports exact next steps for any remaining blockers that require changes in the provider account or repo settings (things AgentRail cannot change for you).
GitHub
GitHub is the primary provider for PR submission, CI observation (GitHub Actions), and code review feedback. What GitHub enables:adapter_managedPR submission — AgentRail creates or reuses pull requests from the task’s persisted source metadata and returnsprUrlandprNumber.- GitHub Actions CI status — AgentRail surfaces a compact CI summary at
/tasks/{id}/ci-status. - Code review feedback — AgentRail observes review decisions and requested changes at
/tasks/{id}/review-feedback.
provider connect github (and during agentrail init), you choose how AgentRail handles code review before shipping:
| Policy | Behavior |
|---|---|
github_rules | Follow GitHub’s own branch protection and review signals. This is the default. |
always_require | Always require an approval for AgentRail-created PRs before shipping, even if the repo does not require it. |
never_require | Skip approval waits once CI is green. |
CircleCI
CircleCI provides CI status for tasks that use CircleCI pipelines instead of GitHub Actions. What CircleCI enables:- CI status observation for tasks using CircleCI pipelines.
- Pipeline-run API access when automatic branch builds are not available.
- Inbound webhook verification using
CIRCLECI_WEBHOOK_SECRET— configure your CircleCI webhook to point to your AgentRail server’s/providers/circleci/webhookspath.
provider connect circleci:
- AgentRail prompts for your full CircleCI project slug in the format
circleci/<org-id>/<project-id>. - It verifies project access.
- If
.circleci/config.ymlis missing and AgentRail can infer a safe starter Node.js workflow, the connect command creates it. - If automatic CircleCI branch builds are not available, AgentRail configures the pipeline-run API when your token can access a pipeline definition.
- If any settings cannot be fixed locally, setup reports the exact blocker and what you need to change in your CircleCI account.
Linear
Linear provides issue intake. After connecting, you can import individual issues into AgentRail’s task queue. Import an issue:Checking provider health
Two commands give you visibility into provider status at any time:agentrail provider test <provider> runs the same readiness engine as agentrail provider doctor <provider>. A passing test means the provider is ready for the full AgentRail lifecycle — not merely that a token exists and the API is reachable..png?fit=max&auto=format&n=TNCV8lxzL1ij3U4N&q=85&s=7345b71aa1c450c74f35b67c2655fc43)