Skip to content

TOOLS.md

Latest commit

 

History

History
272 lines (204 loc) · 4.43 KB

TOOLS.md

File metadata and controls

272 lines (204 loc) · 4.43 KB
Raw

Tools

All tools return structured MCP content with:

  • data
  • instruction
  • optional safety

owlyn_start

Purpose: Start a work session with a goal and deadline.

Inputs:

  • goal: string
  • deadline: ISO datetime or HH:mm
  • timezone: optional, defaults to Europe/Warsaw
  • mode: optional, focused, night_shift, or marathon
  • max_work_minutes: optional
  • notes: optional
  • force_new: optional

Outputs:

  • session_id
  • goal
  • mode
  • timezone
  • started_at
  • deadline_at
  • elapsed_minutes
  • remaining_minutes
  • status
  • initial_instruction

Example:

{
  "goal": "Improve the repo safely until morning.",
  "deadline": "06:00",
  "mode": "night_shift"
}

owlyn_status

Purpose: Return current session status.

Inputs:

  • session_id: optional

Outputs:

  • session metadata
  • current time
  • elapsed and remaining minutes
  • deadline status
  • checkpoint count
  • latest checkpoint summary, next tasks, and blockers

Example:

{
  "session_id": "owl_..."
}

owlyn_checkpoint

Purpose: Save meaningful progress.

Inputs:

  • session_id: optional
  • summary: string
  • completed_tasks: string[]
  • next_tasks: string[]
  • blockers: optional string[]
  • files_changed: optional string[]
  • validation_results: optional string[]
  • confidence: low, medium, or high
  • risk_level: optional low, medium, or high

Outputs:

  • checkpoint metadata
  • completed tasks
  • next tasks
  • blockers
  • files changed
  • validation results
  • confidence and risk
  • status after checkpoint

Example:

{
  "summary": "Added policy tests.",
  "completed_tasks": ["Added night_shift continuation test"],
  "next_tasks": ["Document host setup"],
  "validation_results": ["npm test passed"],
  "confidence": "high",
  "risk_level": "low"
}

owlyn_should_continue

Purpose: Decide whether the agent should continue working.

Inputs:

  • session_id: optional
  • current_task_done: boolean
  • has_next_tasks: boolean
  • requires_user_approval: optional boolean
  • risk_level: low, medium, or high
  • destructive_action_pending: optional boolean
  • validation_passed: optional boolean
  • notes: optional

Outputs:

  • should_continue
  • reason
  • recommended action
  • next policy
  • safety notes
  • elapsed and remaining minutes
  • deadline status

Example:

{
  "current_task_done": true,
  "has_next_tasks": true,
  "requires_user_approval": false,
  "risk_level": "low",
  "destructive_action_pending": false,
  "validation_passed": true
}

owlyn_plan_next

Purpose: Rank candidate next tasks by impact, risk, approval requirement, time fit, and category.

Inputs:

  • session_id: optional
  • candidate_tasks: array of tasks

Each candidate task:

  • title
  • description: optional
  • impact: low, medium, or high
  • risk: low, medium, or high
  • estimated_minutes: optional
  • requires_user_approval: optional
  • category: optional

Outputs:

  • recommended task
  • ranked tasks
  • rejected tasks with reasons
  • planning notes

Example:

{
  "candidate_tasks": [
    {
      "title": "Add integration tests",
      "impact": "high",
      "risk": "low",
      "estimated_minutes": 20,
      "category": "tests"
    }
  ]
}

owlyn_end

Purpose: End a session and return a final report.

Inputs:

  • session_id: optional
  • final_summary: string
  • remaining_tasks: optional string[]
  • validation_results: optional string[]
  • status: optional completed or abandoned

Outputs:

  • session metadata
  • total elapsed minutes
  • checkpoint count
  • final summary
  • aggregated completed tasks
  • remaining tasks
  • validation results
  • final recommendation

Example:

{
  "final_summary": "Completed release hardening.",
  "remaining_tasks": ["Validate in more hosts"],
  "validation_results": ["npm test passed"],
  "status": "completed"
}

owlyn_list_sessions

Purpose: List previous sessions.

Inputs:

  • status: optional active, completed, abandoned, or all
  • limit: optional number

Outputs:

  • sessions with goal, mode, status, timestamps, and checkpoint count

Example:

{
  "status": "all",
  "limit": 20
}

owlyn_report

Purpose: Return a detailed report without ending the session.

Inputs:

  • session_id: optional

Outputs:

  • session metadata
  • elapsed and remaining minutes
  • checkpoint count
  • timeline
  • aggregated completed tasks
  • latest next tasks
  • latest blockers
  • decision summary

Example:

{
  "session_id": "owl_..."
}