{
  "title": "AI project management agent benchmark 2026",
  "published": "2026-09-13",
  "lastReviewed": "2026-09-13",
  "publisher": "Stellary",
  "method": "Documentation-backed capability benchmark using one fixed sprint packet. No hands-on account trial is claimed.",
  "disclosure": "Stellary publishes this benchmark and is one of the evaluated products. No total score is calculated.",
  "states": {
    "confirmed": "The complete gate is explicit in official documentation or a verified Stellary product contract.",
    "partial": "Only part of the gate is explicit, or it depends on a particular configuration, integration, or workflow.",
    "not_confirmed": "No reliable current public evidence was found for the complete gate; this is not proof of absence."
  },
  "criteria": [
    { "id": "structure", "label": "Structure the sprint" },
    { "id": "context", "label": "Use source context" },
    { "id": "actions", "label": "Act on live work" },
    { "id": "continuation", "label": "Continue without prompting" },
    { "id": "scope", "label": "Limit access and actions" },
    { "id": "checkpoint", "label": "Stop for a human checkpoint" },
    { "id": "trace", "label": "Trace and recover" }
  ],
  "products": [
    {
      "name": "Linear",
      "results": {
        "structure": { "status": "confirmed", "note": "Linear Agent and MCP document project, issue, milestone, and relationship creation or updates.", "evidence": "https://linear.app/docs/mcp" },
        "context": { "status": "confirmed", "note": "Agent context includes issues, projects, teams, and workspace history.", "evidence": "https://linear.app/docs/linear-agent" },
        "actions": { "status": "confirmed", "note": "The agent can create and update core Linear work objects.", "evidence": "https://linear.app/docs/linear-agent" },
        "continuation": { "status": "partial", "note": "Issues can be delegated to agent app users; a general scheduled agent runner is not established in the reviewed docs.", "evidence": "https://linear.app/docs/agents-in-linear" },
        "scope": { "status": "confirmed", "note": "Agents operate through workspace, team, and app-user access controls.", "evidence": "https://linear.app/docs/agents-in-linear" },
        "checkpoint": { "status": "not_confirmed", "note": "No general native pre-write approval gate was found in the reviewed agent documentation.", "evidence": "https://linear.app/docs/linear-agent" },
        "trace": { "status": "partial", "note": "Work and progress are visible, but a dedicated recoverable agent-run ledger is not explicit.", "evidence": "https://linear.app/docs/agents-in-linear" }
      }
    },
    {
      "name": "ClickUp",
      "results": {
        "structure": { "status": "partial", "note": "Agents create tasks, subtasks, documents, and field updates; full sprint-model creation from the packet is not explicit.", "evidence": "https://help.clickup.com/hc/en-us/articles/37045015737111-What-are-Autopilot-Agents" },
        "context": { "status": "confirmed", "note": "Agents use configured knowledge, workspace context, and connected information.", "evidence": "https://help.clickup.com/hc/en-us/articles/31012020810775-Create-and-configure-Autopilot-Agents" },
        "actions": { "status": "confirmed", "note": "Tools cover task, document, subtask, and field actions.", "evidence": "https://help.clickup.com/hc/en-us/articles/37045015737111-What-are-Autopilot-Agents" },
        "continuation": { "status": "confirmed", "note": "Autopilot Agents run from configured triggers and conditions.", "evidence": "https://help.clickup.com/hc/en-us/articles/31012020810775-Create-and-configure-Autopilot-Agents" },
        "scope": { "status": "confirmed", "note": "Super Agents are users with configurable locations and permissions.", "evidence": "https://help.clickup.com/hc/en-us/articles/36926065055127-Super-Agent-privacy-security-and-permissions" },
        "checkpoint": { "status": "partial", "note": "Instructions can involve a person as a checkpoint, but a universal enforced gate is not documented.", "evidence": "https://help.clickup.com/hc/en-us/articles/36926065055127-Super-Agent-privacy-security-and-permissions" },
        "trace": { "status": "confirmed", "note": "Activity and audit views expose automation and agent runs.", "evidence": "https://help.clickup.com/hc/en-us/articles/30953763592087-View-your-Automations-and-Autopilot-Agents-activity" }
      }
    },
    {
      "name": "Notion",
      "results": {
        "structure": { "status": "partial", "note": "Pages and databases can model the packet, but Notion does not impose a first-class sprint structure.", "evidence": "https://www.notion.com/help/custom-agents" },
        "context": { "status": "confirmed", "note": "Custom Agents read permitted pages, databases, and connected applications.", "evidence": "https://www.notion.com/help/custom-agents" },
        "actions": { "status": "confirmed", "note": "Agents can update records, publish reports, create files, and send messages.", "evidence": "https://www.notion.com/help/custom-agents" },
        "continuation": { "status": "confirmed", "note": "Custom Agents run on schedules or events and can hand work to other agents.", "evidence": "https://www.notion.com/help/custom-agents" },
        "scope": { "status": "confirmed", "note": "Agents receive explicit page, database, and application access.", "evidence": "https://www.notion.com/help/custom-agents-sharing-and-permissions" },
        "checkpoint": { "status": "partial", "note": "MCP tools can always ask before running; the same universal gate is not established for internal actions.", "evidence": "https://www.notion.com/en-gb/help/mcp-connections-for-custom-agents" },
        "trace": { "status": "confirmed", "note": "Activity logs, version history, and configuration audit information are documented.", "evidence": "https://www.notion.com/help/custom-agents" }
      }
    },
    {
      "name": "Asana",
      "results": {
        "structure": { "status": "confirmed", "note": "Smart Import converts source documents into projects, tasks, sections, and fields with a preview.", "evidence": "https://help.asana.com/s/article/smart-import" },
        "context": { "status": "confirmed", "note": "AI Teammates operate from project and Work Graph context available to them.", "evidence": "https://help.asana.com/s/article/how-to-create-or-discover-an-ai-teammate" },
        "actions": { "status": "confirmed", "note": "AI can create and update tasks, comments, fields, and sections.", "evidence": "https://help.asana.com/s/article/understanding-asana-ai-requests-and-usage-allocation" },
        "continuation": { "status": "confirmed", "note": "AI Teammates can be assigned recurring or triggered project responsibilities.", "evidence": "https://help.asana.com/s/article/how-to-create-or-discover-an-ai-teammate" },
        "scope": { "status": "confirmed", "note": "Viewer, commenter, and editor access is explicit and agents cannot elevate themselves.", "evidence": "https://help.asana.com/s/article/understanding-access-control-for-ai-teammates?language=en_US" },
        "checkpoint": { "status": "partial", "note": "A notice-recommend-approve pattern is documented for client management, not as a universal gate.", "evidence": "https://help.asana.com/s/article/getting-started-with-ai-teammates-in-client-management" },
        "trace": { "status": "partial", "note": "Usage administration and work history provide visibility; a unified recoverable run ledger is not explicit.", "evidence": "https://help.asana.com/s/article/understanding-asana-ai-requests-and-usage-allocation" }
      }
    },
    {
      "name": "monday",
      "results": {
        "structure": { "status": "partial", "note": "Agents act in boards and workflows, but complete sprint creation from the fixed packet is not explicit.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "context": { "status": "confirmed", "note": "Agents use boards, documents, workflows, integrations, and workspace context.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "actions": { "status": "confirmed", "note": "Agents update items, owners, statuses, messages, outcomes, and follow-ups.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "continuation": { "status": "confirmed", "note": "Configured triggers and multi-agent flows support continued execution.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "scope": { "status": "confirmed", "note": "Agents use configured scope and workspace permissions.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "checkpoint": { "status": "not_confirmed", "note": "No general mandatory checkpoint before a protected write was found in the reviewed docs.", "evidence": "https://support.monday.com/hc/en-us/articles/33347027353746-AI-Agents-on-monday-com" },
        "trace": { "status": "confirmed", "note": "Run activity includes actions, reasoning, status, credit use, failures, partial results, and supported undo.", "evidence": "https://support.monday.com/hc/en-us/articles/38411316441874-Run-and-manage-your-AI-agents" }
      }
    },
    {
      "name": "Stellary",
      "results": {
        "structure": { "status": "confirmed", "note": "Agents and MCP clients operate on projects, cards, documents, and related project objects.", "evidence": "/docs/mcp/" },
        "context": { "status": "confirmed", "note": "Missions combine project context, history, memory, documents, and current work.", "evidence": "/docs/guide/agents/" },
        "actions": { "status": "confirmed", "note": "Agents can update cards, comments, statuses, events, and project work through governed tools.", "evidence": "/docs/guide/agents/" },
        "continuation": { "status": "confirmed", "note": "Mission queues and runs let assigned work continue asynchronously.", "evidence": "/docs/guide/agents/" },
        "scope": { "status": "confirmed", "note": "Agent roles, autonomy modes, workspace scope, and allow-ask-deny tool profiles are explicit.", "evidence": "/docs/guide/agents/" },
        "checkpoint": { "status": "confirmed", "note": "Approval mode, tool policy, proposals, and reviews provide enforced human checkpoints.", "evidence": "/docs/guide/agents/" },
        "trace": { "status": "confirmed", "note": "Mission history exposes timelines, tool executions, status, progress, token use, and cost.", "evidence": "/docs/guide/agents/" }
      }
    }
  ]
}