Configure Superlog AI Agent Automation and PR Settings

When the Superlog AI agent finishes investigating an incident and produces a fix, it opens a pull request on your behalf. Agent automation settings give you precise control over what happens next — from leaving the PR for your team to review, all the way to merging it automatically the moment CI passes. All settings in this section are per-project. Changes take effect on the next agent run; they do not retroactively affect open PRs.

Configuring settings

You can adjust agent automation settings in two ways:

Dashboard: Navigate to Settings → Automation inside the project.
Management API: Send a PATCH request to /api/v1/projects/<projectId> with the fields you want to change. Only the fields you include are written — absent fields are left unchanged.

curl -X PATCH https://<host>/api/v1/projects/<projectId> \
  -H "Authorization: Bearer sl_management_xxxx" \
  -H "Content-Type: application/json" \
  -d '{"automerge_fix_prs": "when_checks_pass", "automerge_method": "squash"}'

Auto-merge policy (`automerge_fix_prs`)

This setting controls whether the agent merges the PR it opens, and when.

Value	Behaviour
`never` (default)	The agent opens the PR but never merges it. Your team reviews and merges manually.
`when_checks_pass`	Uses GitHub’s native auto-merge feature. The PR merges automatically once all required status checks pass.
`immediately`	Calls the GitHub merge API right after opening the PR, without waiting for any CI checks.

immediately is a powerful setting. If your CI is flaky or your repo has no required status checks, Superlog will merge the fix PR before any tests have run. Enable it only when you have high confidence in your pipeline’s reliability.

when_checks_pass delegates the merge decision to GitHub’s auto-merge mechanism. Make sure your repository has at least one required status check configured, or the PR may merge as soon as it is opened.

Merge strategy (`automerge_method`)

When automerge_fix_prs is set to when_checks_pass or immediately, this setting controls the Git merge strategy used.

Value	Behaviour
`squash` (default)	Squashes all commits on the branch into a single commit on the base branch. Keeps history clean.
`merge`	Creates a merge commit. Preserves the full branch history.
`rebase`	Rebases the branch commits onto the base branch tip. Produces a linear history without a merge commit.

If automerge_fix_prs is never, this field has no effect but is still stored and will apply if you later enable auto-merge.

Target branch (`pr_base_branch`)

By default, the agent targets the repository’s default branch (usually main or master). Set pr_base_branch to override this per project.

curl -X PATCH https://<host>/api/v1/projects/<projectId> \
  -H "Authorization: Bearer sl_management_xxxx" \
  -H "Content-Type: application/json" \
  -d '{"pr_base_branch": "develop"}'

Set pr_base_branch to null or an empty string to revert to the repository’s default branch. The value must be 200 characters or fewer.

If your team uses a develop or staging branch as the integration target, set pr_base_branch to that branch name. This prevents the agent from opening PRs that bypass your normal review flow.

Project context

Project context is a free-text description of your architecture that the agent reads before every investigation. Use it to tell the agent about your stack, key services, inter-service dependencies, coding conventions, or anything else that would help it understand a failure faster.

Maximum length: 8,000 characters
Set from the dashboard: Settings → Project Context
Set via the MCP tool: use the set_project_context tool in your MCP-enabled environment
Set via the API: include project_context in a PATCH /api/v1/projects/<projectId> request

curl -X PATCH https://<host>/api/v1/projects/<projectId> \
  -H "Authorization: Bearer sl_management_xxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "project_context": "Monorepo. Main services: orders (Node/Express), payments (Go), fulfillment (Python). All communicate via gRPC. DB is Postgres 15 with read replicas in us-east-1 and eu-west-1."
  }'

Keep project context focused on facts that stay stable over weeks — service names, languages, key libraries, database layout, and on-call conventions. Avoid dynamic information like current incidents or recent deploys, which go stale quickly.

For advanced guidance on shaping agent behaviour with domain knowledge, see Agent Memory.

Full example: enabling squash auto-merge

The following call sets when_checks_pass auto-merge with a squash strategy on a project, targeting the main branch:

curl -X PATCH https://<host>/api/v1/projects/<projectId> \
  -H "Authorization: Bearer sl_management_xxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "automerge_fix_prs": "when_checks_pass",
    "automerge_method": "squash",
    "pr_base_branch": "main"
  }'

The response returns the full project object with all current settings reflected.

Connect GitHub

Set up the GitHub integration so the agent can open and merge PRs on your repositories.

Agent Memory

Add long-lived domain knowledge to guide the agent’s investigations.

Get Started

Core Concepts

Integrations

Features

Configuration

Configure Superlog AI Agent Automation and PR Settings

Configuring settings

Auto-merge policy (`automerge_fix_prs`)

Merge strategy (`automerge_method`)

Target branch (`pr_base_branch`)

Project context

Full example: enabling squash auto-merge

Connect GitHub

Agent Memory

​Configuring settings

​Auto-merge policy (automerge_fix_prs)

​Merge strategy (automerge_method)

​Target branch (pr_base_branch)

​Project context

​Full example: enabling squash auto-merge

​Related

Connect GitHub

Agent Memory

Configuring settings

Auto-merge policy (`automerge_fix_prs`)

Merge strategy (`automerge_method`)

Target branch (`pr_base_branch`)

Project context

Full example: enabling squash auto-merge

Related