Agent Recipes

These recipes show common multi-tool workflows. Each includes the exact tool sequence, expected responses, and decision points.

Questions This Page Answers

What exact tool sequence should an operator run for common workflows?
Which calls prove system health before dispatching high-value work?
How do trust, autonomous sessions, and ROI queries connect in practice?

Recipe 1: First Connection Handshake

Verify auth, discover workspace, and get org context.

Verify Connection

{ "tool": "workspace", "args": { "action": "get" } }

Expected: { "workspace_id": "ws_...", "name": "My Workspace" }If you get workspace_not_set, call workspace with action=list first, then workspace with action=set.

Get Org Overview

{ "tool": "orgx_recommend", "args": { "view": "summary" } }

Expected: Summary with active_initiatives, total_tasks, health indicators.

Check Pending Work

{
  "tool": "orgx_search",
  "args": { "type": "decision", "status": "pending", "limit": 5 }
}

Expected: entities array of pending decisions with id, title, status, and urgency.

Recipe 2: Create and Launch an Initiative

Build a full initiative hierarchy and start execution.

Scaffold the Initiative

{
  "tool": "scaffold_initiative",
  "args": {
    "title": "Q1 Product Launch",
    "summary": "Ship new dashboard features by end of quarter",
    "workstreams": [
      {
        "title": "Frontend Development",
        "tasks": [
          { "title": "Build hero section" },
          { "title": "Implement responsive layouts" },
          { "title": "A/B test CTAs" }
        ]
      },
      {
        "title": "Documentation",
        "tasks": [
          { "title": "Update API docs" },
          { "title": "Create migration guide" }
        ]
      }
    ],
    "milestones": [
      { "title": "Beta launch", "due_date": "2026-04-01" },
      { "title": "GA release", "due_date": "2026-06-01" }
    ]
  }
}

Expected: Returns initiative ID, workstream IDs, task IDs. Auto-launches by default.

Check Health

{
  "tool": "orgx_recommend",
  "args": { "initiative_id": "<returned_id>" }
}

Expected: health_score, milestone progress, blockers.

Browse Created Entities

{
  "tool": "orgx_search",
  "args": { "type": "task", "parent_id": "<workstream_id>", "limit": 20 }
}

Expected: List of tasks with status, titles, and IDs.

Recipe 3: Decision Triage

Review and process pending decisions efficiently.

Fetch Pending Decisions

{
  "tool": "orgx_search",
  "args": { "type": "decision", "status": "pending", "limit": 10 }
}

Expected: Pending decisions. Filter the results client-side for critical items if you want the strictest triage queue.

Get Context for a Decision

Use orgx_search to understand the background:

{
  "tool": "orgx_search",
  "args": {
    "query": "<decision title or topic>",
    "scope": "decisions",
    "limit": 5
  }
}

Approve or Reject

{
  "tool": "orgx_decide",
  "args": {
    "decision_id": "dec_...",
    "note": "Approved — budget within limits"
  }
}

Or reject with feedback:

{
  "tool": "orgx_decide",
  "args": {
    "decision_id": "dec_...",
    "reason": "Budget too high — reduce by 20%"
  }
}

Recipe 4: Overnight Autonomous Session

Set up a bounded autonomous session and review results the next morning.

Check Trust Levels

{
  "tool": "get_my_trust_context",
  "args": { "workspace_id": "ws_...", "agent_type": "engineering-agent" }
}

Expected: Trust levels per capability. Only autonomous capabilities will execute without approval.

Start Session

{
  "tool": "start_autonomous_session",
  "args": {
    "workspace_id": "ws_...",
    "session_type": "overnight",
    "max_cost_usd": 5.00,
    "max_receipts": 50
  }
}

Expected: Session ID, status active.

Morning Review

{
  "tool": "orgx_recommend",
  "args": { "workspace_id": "ws_..." }
}

Expected: Curated receipts, exceptions that need attention, ROI delta.

Check ROI

{
  "tool": "orgx_recommend",
  "args": {}
}

Expected: Session summary, cost, completed work, pending decisions, and artifacts from the latest autonomous run.

Recipe 5: Planning Session

Use collaborative planning tools to improve an initiative plan.

Start Planning

{
  "tool": "orgx_plan",
  "args": {
    "initiative_id": "init_...",
    "focus": "Refine Q1 launch workstreams"
  }
}

Get AI Suggestions

{
  "tool": "orgx_plan",
  "args": { "session_id": "<session_id>" }
}

Expected: AI-generated suggestions for gaps, risks, and improvements.

Record Edits

{
  "tool": "orgx_plan",
  "args": {
    "session_id": "<session_id>",
    "edit_type": "add_workstream",
    "description": "Added QA workstream for regression testing"
  }
}

Complete Planning

{
  "tool": "orgx_plan",
  "args": { "session_id": "<session_id>" }
}

Recipe 6: Entity CRUD Lifecycle

Manage the complete lifecycle of an entity — create, update, comment, complete, and verify.

Create an Entity

{
  "tool": "orgx_write",
  "args": {
    "type": "task",
    "title": "Implement OAuth2 token refresh",
    "parent_id": "ws_abc123"
  }
}

Expected:

{
  "id": "tsk_abc123",
  "type": "task",
  "title": "Implement OAuth2 token refresh",
  "status": "pending",
  "parent_id": "ws_abc123",
  "created_at": "2026-02-26T10:00:00Z"
}

Update the Entity

{
  "tool": "update_entity",
  "args": {
    "entity_id": "tsk_abc123",
    "description": "Handle silent token refresh using refresh_token grant. Must support token rotation and revocation."
  }
}

Expected:

{
  "id": "tsk_abc123",
  "title": "Implement OAuth2 token refresh",
  "description": "Handle silent token refresh using refresh_token grant. Must support token rotation and revocation.",
  "status": "pending",
  "updated_at": "2026-02-26T10:02:00Z"
}

Comment on It

{
  "tool": "comment_on_entity",
  "args": {
    "entity_id": "tsk_abc123",
    "body": "Spike complete — refresh window should be 5 minutes before expiry. See design doc for sequence diagram."
  }
}

Expected:

{
  "comment_id": "cmt_def456",
  "entity_id": "tsk_abc123",
  "body": "Spike complete — refresh window should be 5 minutes before expiry. See design doc for sequence diagram.",
  "created_at": "2026-02-26T10:15:00Z"
}

Complete the Entity

{
  "tool": "orgx_act",
  "args": {
    "entity_id": "tsk_abc123",
    "action": "complete"
  }
}

Expected:

{
  "id": "tsk_abc123",
  "status": "completed",
  "completed_at": "2026-02-26T14:30:00Z"
}

Verify Parent Completion

{
  "tool": "orgx_act",
  "args": {
    "dry_run": true,
    "entity_id": "ws_abc123"
  }
}

Expected:

{
  "entity_id": "ws_abc123",
  "type": "workstream",
  "total_children": 5,
  "completed_children": 3,
  "completion_pct": 60,
  "is_complete": false,
  "remaining": ["tsk_ghi789", "tsk_jkl012"]
}

If is_complete is true, the parent workstream can be marked complete as well.

Recipe 7: Scoring & Queue Management

Manage the Next Up Queue — score items, inspect signals, pin priorities, tune weights, and skip blockers.

Score the Queue

{
  "tool": "orgx_recommend",
  "args": {
    "entity_type": "workspace",
    "entity_id": "default"
  }
}

Expected:

{
  "recommendation": {
    "action": "orgx_decide",
    "target_id": "dec_abc123",
    "reasoning": "Campaign brief matches brand guidelines and has strong CTAs.",
    "confidence": 0.92
  }
}

Get Scoring Signals

{
  "tool": "get_scoring_signals",
  "args": {
    "entity_id": "tsk_q01aaa"
  }
}

Expected:

{
  "entity_id": "tsk_q01aaa",
  "signals": {
    "critical_bug": { "weight": 1.5, "value": true, "contribution": 30 },
    "blocking_release": { "weight": 1.3, "value": true, "contribution": 26 },
    "recency": { "weight": 1.0, "value": 0.9, "contribution": 18 },
    "effort_estimate": { "weight": 0.8, "value": "small", "contribution": 20 }
  },
  "total_score": 94
}

Pin a High-Priority Item

{
  "tool": "pin_queue_item",
  "args": {
    "entity_id": "tsk_q01aaa",
    "reason": "P0 auth regression — must ship before release cut"
  }
}

Expected:

{
  "entity_id": "tsk_q01aaa",
  "pinned": true,
  "reason": "P0 auth regression — must ship before release cut",
  "pinned_at": "2026-02-26T09:05:00Z"
}

Pinned items always appear at the top of the queue regardless of score.

Adjust Scoring Weights

{
  "tool": "set_scoring_weights",
  "args": {
    "weights": {
      "critical_bug": 2.0,
      "blocking_release": 1.5,
      "user_impact": 1.2,
      "tech_debt": 0.5,
      "recency": 1.0
    }
  }
}

Expected:

{
  "updated": true,
  "weights": {
    "critical_bug": 2.0,
    "blocking_release": 1.5,
    "user_impact": 1.2,
    "tech_debt": 0.5,
    "recency": 1.0
  },
  "note": "The next orgx_recommend call will use these updated scoring weights"
}

Skip a Blocked Workstream

{
  "tool": "skip_workstream",
  "args": {
    "workstream_id": "ws_blocked99",
    "reason": "Waiting on third-party API access — ETA next week"
  }
}

Expected:

{
  "workstream_id": "ws_blocked99",
  "status": "skipped",
  "reason": "Waiting on third-party API access — ETA next week",
  "skipped_at": "2026-02-26T09:10:00Z",
  "tasks_dequeued": 4
}

Skipped workstreams and their tasks are removed from the queue until the workstream is resumed.

Getting Started

Connect Your Tools

Agent Ops

Platform

API Reference

Resources

Questions This Page Answers

Recipe 1: First Connection Handshake

Recipe 2: Create and Launch an Initiative

Recipe 3: Decision Triage

Recipe 4: Overnight Autonomous Session

Recipe 5: Planning Session

Recipe 6: Entity CRUD Lifecycle

Recipe 7: Scoring & Queue Management

Next Steps

Failure Playbooks

Tool Profiles

​Questions This Page Answers

​Recipe 1: First Connection Handshake

​Recipe 2: Create and Launch an Initiative

​Recipe 3: Decision Triage

​Recipe 4: Overnight Autonomous Session

​Recipe 5: Planning Session

​Recipe 6: Entity CRUD Lifecycle

​Recipe 7: Scoring & Queue Management

​Next Steps

Failure Playbooks

Tool Profiles

Questions This Page Answers

Recipe 1: First Connection Handshake

Recipe 2: Create and Launch an Initiative

Recipe 3: Decision Triage

Recipe 4: Overnight Autonomous Session

Recipe 5: Planning Session

Recipe 6: Entity CRUD Lifecycle

Recipe 7: Scoring & Queue Management

Next Steps