Everything
on each ticket
Context, checks, learning and controls come with every ticket Pilot takes.
Context
Pilot knows your project before it edits a file.
Project docs
Pilot reads the .agent/ folder in your repo: architecture, patterns and how-to guides. It creates the folder on the first run if the repo has none.
Docs →Knowledge graph
Tasks, files and patterns are linked, so a later ticket finds what an earlier one learned.
Docs →Decisions on record
After each run, Pilot records non-obvious choices and updates the docs index.
Docs →Execution
Many tickets at once, without collisions.
Parallel when safe
Tickets that touch different folders run at the same time. Tickets that share folders run one after another, oldest first.
Docs →Many repos, one Pilot
One Pilot watches several repositories. Each repo has its own queue and merge rules.
Docs →Chat to a ticket
Ask about your code in Slack or Telegram, or turn one sentence into a tracked issue.
Docs →Quality
Proof in each PR, and a trend over time.
Learning
Each run makes the next one better.
Control
You set the limits. Pilot works inside them.
Approvals
Approve before the run, before the merge or after a failure, in Telegram, Slack or a PR review.
Docs →Budget caps
Daily, monthly and per-ticket limits. A cap pauses work, and an alert comes at 80% of the daily limit.
Docs →Teams and roles
Owner, Admin, Developer and Viewer roles, with an audit log of team and task actions.
Docs →Visibility
See each ticket, each run and each dollar.
Dashboard
Tickets, queue, tokens and cost in your terminal, or the same panels in a desktop app for macOS, Windows and Linux.
Docs →Replay
Each run is recorded: tool calls, token use and file changes. Replay a failed run step by step.
Docs →Alerts and metrics
Alerts on stuck runs, spend and security to Slack, Telegram, email, webhooks or PagerDuty. Metrics for Prometheus and Grafana.
Docs →Start
Run Pilot free in your terminal, or move the team to Pilot Cloud.