# Agent orchestration (https://jackin.tailrocks.com/research/agents/orchestration/)



This category organizes the jackin❯ research on agent orchestration. Choose a page by the question you need to answer.

## Browse [#browse]

<Card title="Agent Orchestration Program" href="/research/agents/orchestration/program-research/">
  Synthesizes external orchestrator research into a phased direction for terminal-first, isolated agent operations.
</Card>

<Card title="Orca ADE research" href="/research/agents/orchestration/orca-ade/">
  Compares Orca's worktree-native agent environment with jackin❯ to identify reusable fleet and desktop interaction patterns.
</Card>

<Card title="Agent Attention Prompts: Design Exploration" href="/research/agents/orchestration/agent-attention-prompts-design/">
  Designs host notifications for agents that finish, block, or await input without disrupting operator focus.
</Card>

<Card title="Idle Runtime Cleanup: Design Notes" href="/research/agents/orchestration/idle-runtime-cleanup-design/">
  Defines safe idle hooks for reclaiming runtime resources and preparing warm agents for repeated autonomous work.
</Card>

<Card title="jackin❯ Capsule multiplexer design rationale" href="/research/agents/orchestration/jackin-capsule-multiplexer-rewrite/">
  Explains the terminal, protocol, and compatibility decisions behind the current jackin❯ Capsule multiplexer architecture.
</Card>

<Card title="Stack Integration Hints: Design Rationale" href="/research/agents/orchestration/stack-integration-hints/">
  Examines how role and workspace metadata can expose stack-aware commands without hard-coding framework behavior.
</Card>

<Card title="Workspace description field: prior art and shape" href="/research/agents/orchestration/workspace-description-field/">
  Defines workspace descriptions as shared operator and agent context, including scope, storage, and presentation.
</Card>

<Card title="Déjà Vu conversation-capture design" href="/research/agents/orchestration/conversation-capture/">
  Designs durable, searchable conversation capture across supported agent runtimes.
</Card>

<Card title="Fleet operations" href="/research/agents/orchestration/fleet-operations/">
  Research on fleet operations.
</Card>

<Card title="Memory" href="/research/agents/orchestration/memory/">
  Research on memory.
</Card>

<Card title="Multi-agent collaboration" href="/research/agents/orchestration/multi-agent-collaboration/">
  Research on multi-agent collaboration.
</Card>

<Card title="Recovery" href="/research/agents/orchestration/recovery/">
  Guide to preserving, resuming, snapshotting, and rolling back isolated agent sessions.
</Card>

<Card title="Remote execution" href="/research/agents/orchestration/remote-execution/">
  Research on remote execution.
</Card>

<Card title="Resource limits" href="/research/agents/orchestration/resource-limits/">
  Guide to declarative runtime ceilings and operator-facing resource visibility research.
</Card>

<Card title="Terminal observation" href="/research/agents/orchestration/terminal-observation/">
  Guide to terminal-state observation, runtime-status arbitration, and automation research.
</Card>

<Card title="Workflow systems" href="/research/agents/orchestration/workflow-systems/">
  Guide to task-source abstraction and durable agent-workflow orchestration research.
</Card>

## How to read [#how-to-read]

Start with the page closest to your current question. Follow its related-work links for evidence in adjacent categories and its Roadmap links for implementation commitments.