Does the CLI replace Git?

No. It uses your existing repository and system Git for git operations, then adds Trylle context for pull requests, issues, CI, automations, stacks, and review. The interactive diff can also detect Git or Jujutsu working copies.

Do I have to configure an AI provider?

No. Platform commands such as status, repo, pr, issue, stack, and ci work without the optional local AI helpers. Draft, explain, ask, and agent use the provider or local agent CLI you configure.

Which coding agents can it run?

The agent command supports Codex CLI, Claude CLI, and OpenCode CLI. The diff review handoff also works with any agent that can run a shell command and consume stdout.

What should I know before running try agent?

Agent mode is intentionally unattended and starts each provider with broad non-interactive permissions so it can edit Git state and reach remote services. Run it only in repositories and environments you trust, then review the resulting diff before shipping.

Can agents and scripts avoid the formatted terminal output?

Yes. Add --json to platform and supported local commands for machine-readable output. The raw api command is limited to the configured public Trylle API; it is not a private service or admin passthrough.

A daily driver, not a command catalog.

Repository context, pull requests, CI, and review in one command line. The interactive diff sends your notes straight back to the agent.

Get started

npm i -g @trylle/trylle-cli

Extend the Magic to your terminal

Each command starts from the repository you are already in. See what needs attention, act on it, and return to the code without rebuilding context in another tab.

Start with the work queue

try status combines your identity, current repository, unread inbox, review requests, open pull requests, recent issues, CI, automations, and invitations.

Review at terminal speed

try diff opens a side-by-side TUI for working changes, commits, ranges, stacks, or pull requests—with file progress, themes, watch mode, and inline annotations.

Close the agent loop

Annotate a selection, hunk, or file, then press s. The CLI sends clean review notes through stdout so Codex, Claude, or any shell-capable agent can take the next turn.

High-performance TUI for your diffs and reviews

Working changes, a commit range, a stack, or a whole pull request open in the same side-by-side reviewer. Annotate a line, a hunk, or a file, then send the notes to the pull request or straight to your agent.

[2]Modified File

@@ -42,9 +42,17 @@ async function retry(job: Job)

4242 const lease = await claim(job);

4343 if (!lease) return null;

4444

45- await run(job);

46- await lease.release();

45+ try {

46+ await run(job, { signal: lease.signal });

47+ } finally {

48+ await lease.release();

49+ }

4750

4851 metrics.record('queue.retry', lease.id);

4952 return lease.id;

5053}

!Does the release still need a timeout when the signal aborts?

fix/auth-refreshsrc/queue/retry.ts+12-4

Stacked pull requests, first-class support

Parent branches are first-class in the platform, so the CLI can build and rebuild a chain without rewriting your remotes or installing a second tool.

try stack create — Branch off the pull you are already working on. The parent link is recorded server-side, not inferred from branch names.

The rest of the workflow is one command away.

The CLI follows a predictable noun-first shape and exposes JSON output for scripts and agents.

Pull requests and stacks

Create, inspect, check out, review, merge, and manage review threads. Build dependent branches with stack create, restack, sync, and submit.

try pr current
try pr review
try stack submit

CI and automations

List runs, stream logs, watch completion, cancel jobs, inspect usage, and enable or disable repository automations.

try ci watch
try ci logs --follow
try automation runs

Repositories and issues

Clone with system Git, inspect repository context, triage issues, manage labels and comments, and jump back to the web when you need it.

try clone owner/repo
try repo context
try issue timeline

Agents and local AI

Run Codex, Claude, or OpenCode against the current repository. Draft commits and branches, explain changes, or generate guarded Git commands with your chosen provider.

try agent
try draft branch
try explain

Ship AI work your team can actually trust

Agent execution, review, permissions, and CI in one system. No credit card and no trial time. Just free.

Get started free — no card Explore repos