> ## Documentation Index
> Fetch the complete documentation index at: https://ade-ac1c6011-ade-im-seeing-error-messag-eright-58df792e.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Templates and outputs

> Start from a built-in automation template, then configure how results are delivered across lanes, PRs, Linear, and ADE's artifact store.

## Templates

Templates are pre-configured automation rules you can activate with a click. Each ships with a tested trigger, execution surface, model config, guardrails, and prompt — every field is editable after creation.

Navigate to **Automations → Templates** to browse them.

<Note>
  Templates whose execution surface is a Worker run show a **"Coming soon"** chip in packaged builds and the Use Template button reads "Coming Soon" with a tooltip explaining why. The card is still visible so you can see the shape of the rule and plan around it.
</Note>

### Built-in templates

<CardGroup cols={2}>
  <Card title="Daily Agent Brief" icon="message">
    **Trigger:** `schedule` — weekdays at 9am.
    **Execution:** Agent session.

    Summarizes repo activity since the last weekday brief — commits, PRs, worker runs, and risk signals — in an automation-only chat thread. Pulls repo state and procedures as context.
  </Card>

  <Card title="PR Review Session" icon="code-pull-request">
    **Trigger:** `git.pr_opened` on `main`.
    **Execution:** Agent session.

    Reviews every new PR for risky changes, missing tests, and release impact. Posts findings citing specific files and checks, with high thinking level for thorough analysis.
  </Card>

  <Card title="Linear Triage Worker run" icon="flag">
    **Trigger:** `linear.issue_created` (team: ENG).
    **Execution:** Worker run (disabled in packaged builds).

    Launches a triage worker run on new Linear issues — owner, severity, next action — using worker run-level tooling. In production builds the card shows "Coming soon" until Workers ships.
  </Card>

  <Card title="Nightly Test Sweep" icon="flask">
    **Trigger:** `schedule` — weekdays at 2am.
    **Execution:** Built-in (`run-tests`).

    Runs your configured test suite without starting a chat or worker run. Results are saved as automation artifacts. Requires a test suite ID in project config.
  </Card>

  <Card title="Push Conflict Scan" icon="bug">
    **Trigger:** `git.push` on `main`.
    **Execution:** Built-in (`predict-conflicts`).

    Runs conflict prediction after every push to the default branch, so at-risk lanes surface before they block work.
  </Card>
</CardGroup>

***

## Using a template

<Steps>
  <Step title="Browse">
    Open **Automations → Templates**. Each card shows trigger, execution kind, and a short summary.
  </Step>

  <Step title="Use Template">
    Click **Use Template**. ADE creates a new rule pre-filled with the template's configuration and opens it in the editor. (For a worker run-based template in a packaged build the button reads **Coming Soon** and is disabled.)
  </Step>

  <Step title="Customize">
    Adjust trigger conditions, prompt, model, guardrails, and outputs. Built-in rules expose a **Target lane** dropdown at the top of the action editor — leave it on the default to auto-select from the trigger or primary lane, or pick a specific lane.
  </Step>

  <Step title="Enable">
    Toggle the rule to **enabled**. It fires on the next matching trigger.
  </Step>
</Steps>

<Tip>
  Templates default to `anthropic/claude-sonnet-4-6` with medium or high thinking level. Switch to any configured provider and model after creation — the AI Providers panel lists what's available.
</Tip>

***

## Template anatomy

Every template produces a rule with these fields:

| Field           | What it does                                                                     |
| --------------- | -------------------------------------------------------------------------------- |
| Name            | Identifier shown in the Automations list                                         |
| Trigger         | Event or schedule that fires the rule                                            |
| Execution kind  | `agent-session`, `worker run`, or `built-in`                                     |
| Executor mode   | Always `automation-bot` for templates                                            |
| Prompt          | Instruction given to the agent (for agent-session and worker run kinds)          |
| Model config    | Model ID and thinking level                                                      |
| Tool palette    | ADE CLI domains the agent can use (repo, git, github, linear, worker run, tests) |
| Context sources | What context is seeded (repo state, procedures, lane packs)                      |
| Guardrails      | Max duration, confidence threshold, review gates                                 |
| Outputs         | Where results land (see below)                                                   |
| Review profile  | `quick` or `full`                                                                |
| Billing code    | Cost tracking tag, prefixed with `auto:`                                         |

***

## Outputs and routing

Automations can produce several output types. Configure routing in the rule's settings to control where results land.

| Output            | What it is                                       | Example                                           |
| ----------------- | ------------------------------------------------ | ------------------------------------------------- |
| Commits           | Changes committed to a branch in a specific lane | Test fixes, dependency updates                    |
| Pull Requests     | New or updated PRs on GitHub                     | Dependency-update PRs, auto-generated feature PRs |
| Linear updates    | Issue state changes, comments, label assignments | Moving a triaged issue to Ready                   |
| Notifications     | In-ADE toast or macOS system notifications       | "Nightly tests passed", "3 lanes have conflicts"  |
| Lane pack updates | Refresh a lane's context pack                    | After an automation modifies files in a lane      |
| Chat messages     | Post to a lane's chat                            | Review findings in the PR's lane chat             |
| Artifacts         | Reports, logs, or summaries stored with the run  | Test sweep results, daily brief summary           |

### Output disposition

* **Comment-only** — results are stored as artifacts and optionally posted as comments (PRs, Linear, lane chats). No code changes.
* **Commit** — the automation can make code changes and commit them. For rules like dependency updates.
* **Draft PR** — the automation creates a draft PR for human review.

<Warning>
  `commit` and `draft PR` dispositions modify your codebase. Always pair them with guardrails (max duration, confidence threshold) and consider starting in review mode so the automation pauses for approval before acting.
</Warning>

***

## Viewing automation runs

The **History** tab lists every run with its status, trigger, duration, and cost. Run rows now support direct navigation — if a run executed as a worker run, the detail panel includes a link that opens the worker run in the Workers view.

Rows are filterable by rule, status, and time window. Errors are surfaced inline; you don't have to open the detail panel to see that something failed.

***

## Creating custom templates

ADE does not yet ship custom template authoring. To replicate a rule with different parameters, right-click an existing rule → **Duplicate**, then modify the copy.

***

## Next steps

<CardGroup cols={2}>
  <Card title="Automation rules guide" icon="scroll" href="/guides/automation-rules">
    Step-by-step walkthrough for authoring custom automation rules.
  </Card>

  <Card title="Guardrails" icon="shield" href="/automations/guardrails">
    Safety limits, review modes, and intervention policies.
  </Card>
</CardGroup>