Skip to content
BoldCreate

Cloud Agent Automations

Set up Cloud Agent automations with schedules, code events, or webhooks, and use built-in recipes to run tasks automatically.

Overview

Automations are Cloud Agent tasks that start on their own, without manual initiation. You set when it runs, where it works, and what it may do. Each automated run becomes a task in your sidebar with a pull request, a review comment, or findings for your team to review and merge.

Core Value Proposition

  • Cloud Agent tasks triggered by schedules, code events, or webhooks
  • 10 built-in recipes for common automation scenarios
  • Every run works in its own isolated sandbox
  • Changes arrive as pull requests for team review
  • Memory carries across runs (e.g., skips already-handled bugs)
  • All spending and audit controls apply

Key Capabilities

FeatureDescription
Three Trigger TypesSchedule (cron/weekday/daily/hourly), Code events (PR/comment/review/check), Webhook (CI/monitoring/tracker)
10 Built-in RecipesPre-configured automations for common tasks (find bugs, add tests, generate docs, update deps, review PRs, fix CI, etc.)
GuardrailsSame sandbox, spending cap, and audit log as manual tasks; changes reviewed before merge
Code Event FiltersRun only on specific: base branch, head branch, author, title, labels, or draft status
Memory Across RunsAutomation remembers previously handled bugs, alerts, and tickets; skips duplicates
Webhook IntegrationAccept POST requests with Bearer token auth; up to 10 runs/hour, deduplication within 10 minutes
Test RunsTry one automation run before turning it on; review the task it produces
NLP SetupDescribe automation in one sentence; BoldCreate drafts the trigger, repo, and instructions for review

Setup & Configuration Steps

Creating an Automation

  1. Describe the automation in a sentence.
  2. Example: “Find critical bugs in my commits every morning”.
  3. BoldCreate drafts: trigger, trigger conditions, recipe/instructions.
  4. Review and adjust.
  5. Select repositories and environment.
  6. Set run limits and filters.
  7. Try a test run.
  8. Turn on when ready.

Option 2: Start from Recipe

  1. Browse 10 built-in recipes.
  2. Pick one (e.g., “Review a pull request”, “Find vulnerabilities”).
  3. Customize trigger and conditions.
  4. Select repositories.
  5. Test and enable.

Option 3: Manual Configuration

  1. Pick trigger type (schedule, code event, webhook).
  2. Configure trigger-specific settings.
  3. Write or select automation instructions.
  4. Set what changes are allowed.
  5. Add filtering conditions if needed.
  6. Test and enable.

Trigger Types & Use Cases

Schedule Triggers

When to Use: Routine work that recurs at fixed intervals

FrequencyExampleUse Case
HourlyEvery hourMonitor logs, fetch alerts, triage failures
DailyEvery day at 9 a.m.Find critical bugs, update docs, refresh data
WeekdaysMonday–Friday at 9 a.m.Weekly test coverage, review old issues
WeeklyEvery Monday 9 a.m.Dependency updates, security scans
CustomCron: 0 9 * * 1-5Precise scheduling by hour, day, month, weekday

Built-in Recipes (10 Templates)

Each recipe comes pre-configured with trigger, instructions, and output type. Customize and save.

Daily Schedule Recipes

RecipeTriggerDoesOutput
Find critical bugsEvery dayScans last day’s commits for high-severity bugs; fixes each with a failing test firstPull request
Add test coverageEvery dayFinds recently changed code without tests; writes focused tests in repo’s stylePull request
Generate docsEvery weekUpdates READMEs, doc comments, docs pages for modules changed in last weekPull request
Weekly dependency updateEvery MondayUpdates dependencies to latest patch/minor releases; builds and runs testsPull request

Code Event Recipes

RecipeTriggerDoesOutput
Review a pull requestPR openedReads diff and context; finds bugs, missing error handling, missing testsReview comment
Find vulnerabilitiesPR opened/updatedValidates security findings by tracing code paths; reports severity and fixReview comment
Fix failing CICheck failedReproduces failure; fixes with smallest change; runs check againPull request

Webhook Recipes

RecipeTriggerDoesOutput
Respond to an alertSentry, Datadog, PagerDutyFinds code an alert points to; explains likely cause with fixFindings or pull request
Triage a CI failurePipeline webhookDecides if failure is flaky, infrastructure, or regression; names commitVerdict + evidence
Investigate a new ticketLinear, Jira trackerFinds files involved; checks for duplicates; estimates change scopeFindings or pull request

Quick Start Guide

  • Pick a recipe → E.g., “Review a pull request”
  • Or describe it → “Find bugs in new pull requests every day”
  • Set trigger → Schedule, code event, or webhook
  • Select repos → Where should this automation work?
  • Test run → Try once before enabling
  • Turn on → Automation now runs unattended
Edit on Gitea