Ambient agents
for your codebase
Define automated agents that continuously maintain and improve your code. Dependency upgrades, feature flag cleanup, dead code removal — running while you sleep.
Trigger-based
Cron live, webhooks coming soon
Sandboxed
Safe execution in isolated environments
MCP-native
MCP, skills, and triggers in progress
Features
Everything an ambient
agent needs
Trigger-driven
Cron schedules are live today. GitHub webhooks, Slack messages, PagerDuty incidents, and manual triggers are in progress.
Natural language instructions
Write plain English prompts with @-mentions for tools and MCPs. The agent resolves references and executes autonomously.
Plug-and-play MCPs
MCP support is in progress for Datadog, Sentry, Slack, Databricks, and custom servers so agents can reach your toolchain.
Sandboxed execution
Every run happens in an isolated environment. OS-level sandboxing for local dev, containers for production. Safe by default.
Memories and skills
Reusable skills and persistent memories are in progress, so automations can carry domain knowledge across runs.
PR-native output
Automations open pull requests, post to Slack, create issues. Review agent work like any other contributor.
Use Cases
Automations that actually
run your playbook
Dependency upgrades
Scans for outdated packages, attempts upgrades, runs your test suite. Opens a PR for each successful upgrade with changelog notes.
Feature flag cleanup
Detects flags that are fully rolled out and stale. Removes the flag, dead code paths, and opens a clean PR.
Incident triage
Will pull relevant logs and metrics from Datadog, trace the error, and post a structured analysis to your Slack on-call channel.
User feedback to code
Will read feature requests from Slack, check for duplicate Linear issues, and either implement or summarize the work needed.
How It Works
Three steps to
autonomous maintenance
Define your automation
Use the visual editor to write natural language instructions, choose a model, and configure cron schedules. More triggers and MCP servers are in progress.
# Instructions Upgrade minor dependencies every Monday. Run the test suite. Open a PR with changelog notes.
Agent runs in a sandbox
When the schedule fires, Caret spins up an isolated execution environment, clones your repo, and runs the agent with repo access.
Trigger fired: cron (Mon 09:00) Sandbox: creating... Cloning repo: acme/backend (main) Agent: loading repo context Agent: executing instructions...
Review the results
The agent opens PRs and records every run with diffs, summaries, and full execution traces you can audit.
Run #42 completed (3m 21s) Files changed: 4 PR opened: #1847 "Upgrade lodash 4.17→4.18" Tests: 142 passed, 0 failed
Stop maintaining.
Start shipping.
Let ambient agents handle the toil so your team can focus on what matters.