Coordinate Claude Code and Codex on the same PR

Recipe for running two agents — say, Claude Code and Codex — on the same PR without collisions.

Step 1 — Open a session

From any shell (yours or an agent's):

sl session start --project my-repo --ttl 14400 --json

# > session_id: sess_abc123

TTL is in seconds (14400 = 4 hours). Share the `session_id` with the other agent.

Step 2 — Agent A joins

In Claude Code's terminal:

sl session join sess_abc123 --name claude-a1b2

sl session say sess_abc123 "I will take the auth refactor"

Step 3 — Agent B joins

In Codex's terminal:

sl session join sess_abc123 --name codex-c3d4

sl session say sess_abc123 "Picking up the database migration"

Both agents now see each other on the session stream.

Step 4 — Lock files before editing

Each agent should lock a file before editing it:

# Claude claims src/auth/

sl session lock sess_abc123 --path src/auth/



# Codex claims src/db/

sl session lock sess_abc123 --path src/db/migrations/

If an agent tries to lock a path another agent holds, the command blocks until release or a stale timeout.

Step 5 — Claim a task

If there is a shared task queue (e.g. migration PR with 5 subtasks), an agent claims one:

sl session claim sess_abc123 --work migration-step-3 --lease 1800

# heartbeat every 60s to keep the lease alive

sl session heartbeat sess_abc123 --work migration-step-3

# release when done

sl session release sess_abc123 --work migration-step-3

If the agent crashes, the 1800s lease expires and another agent can pick up the task automatically.

Step 6 — Monitor from the dashboard

Open [sentinelayer.com/admin/sessions](https://sentinelayer.com/admin/sessions) to see both agents live: roster, per-agent cost, leases, locks, recent events.

Step 7 — Kill a runaway agent

If one agent goes into a loop or stalls:

sl session kill --session sess_abc123 --agent codex-c3d4 --reason "loop_detected"

The agent's leases are revoked and an `agent_killed` event is posted to the stream so the other agent (and you) see it.

Step 8 — Archive when done

sl session archive sess_abc123

Session ships to S3 (if configured) with `analytics.json`, `artifact-chain.json`, and full `timeline.ndjson` + SHA-256 chain.

Common pitfalls

• **Forgetting to lock** — two agents edit the same file; second save overwrites first. Always `sl session lock` before `Edit`/`Write` tool calls.
• **Missing heartbeats** — if your loop does not heartbeat within the TTL, lease expires and another agent takes over mid-task. Heartbeat every ~30% of lease duration.
• **Dead API sync** — CLI streams locally to NDJSON always. Dashboard sync is best-effort. Local state is ground truth.

Related

• [Senti overview](/docs/senti/overview) — all session commands
• [CLI v0.8 Reference](/docs/cli/v0-8-reference#senti-sessions-new-in-08)

Structured Answers