management-apilisted

# Management API Skill ## Purpose Provide a stable local JSON query contract over `.agent/` runtime state. Dashboard, briefing, worktree, mission, and future CLI commands should read this layer instead of each parsing the same Markdown and JSON files differently. Runtime writers are intentionally narrow: write run state and run events through this skill so dashboard status reflects what agents are actually doing. ## Commands ```bash node .agent/skills/management-api/scripts/index.js query dashboard-state node .agent/skills/management-api/scripts/index.js query runs node .agent/skills/management-api/scripts/index.js query queues node .agent/skills/management-api/scripts/index.js query sessions node .agent/skills/management-api/scripts/index.js query inbox node .agent/skills/management-api/scripts/index.js query decisions node .agent/skills/management-api/scripts/index.js query waitpoints node .agent/skills/management-api/scripts/index.js runs upsert --run-id R-T005 --task-id T-005 --kind implement --status running --phase decomposing --activity "正在拆分 Adapter" node .agent/skills/management-api/scripts/index.js runs event --run-id R-T005 --type agent_invoked --phase invoking_agent --message "已调用 editor-adapter-agent" node .agent/skills/management-api/scripts/index.js runs checkpoint --run-id R-T005 --task-id T-005 --type validation_started --phase validating --activity "Running focused tests" node .agent/skills/management-api/scripts/index.js queues upsert --queue-id Q-batc