Your first agent session
tl;dr
Get crosslink running and watch your agent handle a task with persistent memory — in under a minute.
1. Install and initialize
cargo install crosslink
cd your-project
crosslink initThis creates .crosslink/ (config, rules, database), .claude/hooks/ (behavioral hooks), and .claude/settings.json (hook wiring). See Installation for more install options.
2. Start a session
You say:
“Let’s get started”
Every conversation begins with a session. If a previous session left handoff notes, your agent sees them immediately.
→
Your agent runs:
crosslink session startOn the first session this is a no-op. On subsequent sessions, handoff notes from the previous agent (or your past self) appear automatically.
3. Create an issue
You say:
“Fix the login bug — it’s high priority”
→
Your agent runs:
crosslink quick "Fix login bug" \
-p high -l bugOne command: creates the issue, labels it, and marks it as the active work item.
5. Commit and close
You say:
“Commit that and close the issue”
The agent uses the built-in /commit skill, which creates the commit and records a result comment on the active issue. Closing auto-appends a changelog entry.
→
Your agent runs:
# /commit stages, writes the message,
# and records the result on the issue
/commit
crosslink close 1
6. End the session
You say:
“Wrap up and leave notes for next time”
→
Your agent runs:
crosslink session end \
--notes "Fixed auth bug #1. Dark mode is next."The next session — whether it is you, another agent, or the same agent after a restart — sees these handoff notes automatically.
What’s next?
- Session Workflow — deep dive into session management
- Design Workflow — iterate on design documents before implementation
- Kickoff — launch autonomous background agents
- Swarm Orchestration — multi-agent phased builds
- Knowledge Management — shared research across agents
- Web Dashboard — browser-based project oversight
- Terminal Dashboard — interactive TUI for browsing issues
- Maintenance — pruning and health checks
- Multi-Agent — coordinate multiple AI agents
- Hooks Guide — behavioral guardrails for AI coding
- CLI Reference — full command reference