12 Capabilities at a Glance
Everything Claude Code can do for your paper. Click any entry for a 30-second summary and a paper-writing scenario.
Grouped by Type
The Agent LoopBash is All You Need
“The minimal agent kernel is a while loop + one tool”
OpenToolsOne Handler Per Tool
“The loop stays the same; new tools register into the dispatch map”
OpenTodoWritePlan Before You Act
“An agent without a plan drifts; list the steps first, then execute”
OpenSubagentsClean Context Per Subtask
“Subagents use independent messages[], keeping the main conversation clean”
OpenSkillsLoad on Demand
“Inject knowledge via tool_result when needed, not upfront in the system prompt”
OpenCompactThree-Layer Compression
“Context will fill up; three-layer compression strategy enables infinite sessions”
OpenTasksTask Graph + Dependencies
“A file-based task graph with ordering, parallelism, and dependencies -- the coordination backbone for multi-agent work”
OpenBackground TasksBackground Threads + Notifications
“Run slow operations in the background; the agent keeps thinking ahead”
OpenAgent TeamsTeammates + Mailboxes
“When one agent can't finish, delegate to persistent teammates via async mailboxes”
OpenTeam ProtocolsShared Communication Rules
“One request-response pattern drives all team negotiation”
OpenAutonomous AgentsScan Board, Claim Tasks
“Teammates scan the board and claim tasks themselves; no need for the lead to assign each one”
OpenWorktree + Task IsolationIsolate by Directory
“Each works in its own directory; tasks manage goals, worktrees manage directories, bound by ID”
Open