快速判断
OpenClaw DeFi 代理的会话状态与控制循环技能，管理持久会话、FarmingContext、事件流快照、心跳及委托等。
适合任务
- 按 SkillHub 收录说明复用成熟任务流程。
- 通过下载包离线阅读完整 Skill 内容。
- 结合热度指标优先评估常用 Skill。
输入与输出
输入：任务目标、上下文材料、文件路径、约束条件或需要处理的内容。
输出：按 Skill 说明生成的文档、代码、检查结果、计划、建议或操作步骤。
示例任务
- 使用 FarmDash Autonomous Operator 帮我处理当前任务，并说明需要准备哪些输入。
- 根据 FarmDash Autonomous Operator 的说明，先列出使用前的安全检查项。
安装方式
- 下载本站提供的 Skill ZIP 并解压。
- 把解压后的 Skill 目录放入当前 AI 工具支持的
skills目录。 - 如需在线查看原始内容，可打开 GitHub 的
SKILL.md。
风险边界
SkillHub 提供了源站安全报告入口，但本站不替代人工审查。使用前仍需检查权限、外部依赖和敏感数据边界。
SKILL.md 文档介绍
FarmDash Autonomous Operator
Autonomous Operator keeps a multi-skill agent coherent across turns. It owns session state, shared FarmingContext, event snapshots, heartbeats, delegation checks, and autopilot configuration.
It does not hold private keys. It does not execute swaps or perps directly.
Tools
create_session
Creates a persistent agent session and returns a one-time sessionToken. Store it securely in the agent runtime. FarmDash stores only a hash.
session_heartbeat
Extends the session expiry. Use it during active autonomous loops.
get_farming_context
Reads shared context for the session:
- objective.
- portfolio scope.
- risk settings.
- workflow state.
- ledger summary.
- data freshness timestamps.
patch_farming_context
Patches shared context. The server controls sessionId, agentAddress, revision, and updatedAt; do not try to override them.
get_event_stream_snapshot
Reads recent agent events as a JSON snapshot. Use this before planning and after execution.
verify_delegation
Checks whether the user's Hyperliquid API wallet delegation is in place for autonomous perps.
configure_autopilot and autopilot_cycle
Configure and run bounded autonomous cycles. These cycles must respect user allowlists, risk limits, and execution confirmations.
Control Loop
1. create_session.
2. get_farming_context.
3. get_event_stream_snapshot.
4. Use Trail Marshal to plan a workflow.
5. Patch objective, risk, workflow, and freshness state with patch_farming_context.
6. Run read-only sense tools.
7. Present execution steps to the user through the owning execution skill.
8. session_heartbeat.
9. After execution, update ledger and freshness state.
Agent Rules
- The session token is a capability. Never display it in normal user-facing prose.
- A context patch is state, not permission.
- A workflow plan is not a user confirmation.
- If risk status is
halted, do not call execution tools. - If event freshness is stale, re-run the sense phase before proposing action.
Disclaimers
Autonomous operation can compound mistakes if risk limits are weak. Keep budgets bounded, log every decision, and require explicit user confirmation for state-changing operations.
Install: Copy this file into your OpenClaw workspace, or fetch https://www.farmdash.one/openclaw-skills/farmdash-autonomous-operator/SKILL.md.
Companion skills: FarmDash Trail Marshal, FarmDash Signal Architect, FarmDash Camp Guard, FarmDash Ledger Keeper.