TeamDay Admin Tool

The TeamdayAdmin tool gives agents full control over platform resources. Available as a built-in MCP tool at mcp__teamday-admin__TeamdayAdmin.

Use it to configure agents, install MCPs, schedule missions, manage skills, and store secrets --- all from within a conversation.

Files and folders are not managed through this tool. Agents read and write files directly with their filesystem tools in the folder their chat or mission works from; see Folders and Files.

---

Resources Overview

---

Agents

Agents are AI team members with a name, role, system prompt, skills, and MCPs. They belong to the whole organization and carry the same equipment into every folder they work in; there are no per-folder copies to create.

Create an Agent

{
  "action": "createAgent",
  "data": {
    "name": "Sarah",
    "role": "SEO Analyst",
    "category": "marketing",
    "systemPrompt": "You are Sarah, an SEO specialist...",
    "skillIds": ["seo", "content-analysis"],
    "advanced_tools": ["ahrefs"],
    "image": "https://www.teamday.ai/agents/agent-robot-1.png",
    "visibility": "organization"
  }
}

Agent categories: marketing, finance, hr, engineering, operations, general, data

Once created, the agent can be given work in any folder by starting a chat or mission with that folder as its working directory.

---

MCPs (Tool Integrations)

Install MCP servers so agents can use external tools like Google Analytics, Ahrefs, or custom APIs. MCP connections are attached to agents, not to folders.

Create an MCP

{
  "action": "createMcp",
  "data": {
    "name": "google-analytics",
    "type": "stdio",
    "command": "npx",
    "args": ["-y", "@anthropic-ai/mcp-google-analytics"],
    "env": {
      "GA_PROPERTY_ID": "12345678"
    }
  }
}

Browse Available MCPs

{
  "action": "browseMcpRegistry"
}

---

Missions (Scheduled Tasks)

Missions are scheduled AI tasks that run automatically. Bind them to an agent to run with that agent's personality, skills, and MCPs, and give them a working folder (cwd) that every run reads from and writes to.

Create a Mission

{
  "action": "createMission",
  "data": {
    "title": "Weekly SEO Report",
    "goal": "Analyze this week's search performance, compare to last week, and email the summary.",
    "cwd": "/marketing/seo",
    "characterId": "char-sarah",
    "schedule": {
      "type": "cron",
      "value": "0 9 * * MON"
    }
  }
}

Schedule Types

Update a Mission

Pause, change schedule, or update the goal:

{
  "action": "updateMission",
  "resourceId": "mission-abc",
  "data": {
    "status": "paused",
    "schedule": { "type": "cron", "value": "0 9 * * 1-5" }
  }
}

---

Skills

Skills are reusable capabilities that teach agents how to perform specialized tasks.

Browse Available Skills

{
  "action": "browseSkillsRegistry"
}

Create a Skill via API

{
  "action": "createSkill",
  "data": {
    "name": "report-generator",
    "description": "Generate formatted PDF reports from data",
    "content": "---\nname: report-generator\ndescription: Generate reports\nallowed-tools: Bash, Read, Write\n---\n\n# Report Generator\n\n..."
  }
}

Create a Skill via Files

Write the skill into the .claude/skills/ directory of the folder the agent works in. Skills defined this way are discovered from the run's working directory; see Skill & Agent Discovery.

.claude/skills/report-generator/
  SKILL.md          # Skill definition
  scripts/
    generate.ts     # Executable script

---

Secrets

Store encrypted environment variables for your organization. They are injected into agent sessions at runtime and shared by every agent, whichever folder it works in.

Store Secrets

{
  "action": "storeSecrets",
  "data": {
    "secrets": {
      "OPENAI_API_KEY": "sk-...",
      "DATABASE_URL": "postgres://..."
    }
  }
}

List Secrets

Returns secret names (not values):

{
  "action": "listSecrets"
}

Remove Secrets

{
  "action": "removeSecrets",
  "data": {
    "keys": ["OPENAI_API_KEY"]
  }
}

---

Agent Images

Use built-in avatar images from TeamDay's collection:

https://www.teamday.ai/agents/{style}-{number}.png

Available styles (numbers 1-9 each):

• agent-new --- Modern, clean
• agent-sketch --- Hand-drawn style
• agent-geo --- Geometric
• agent-robot --- Robot/mechanical
• agent-animal --- Animal characters
• agent-fruit --- Playful fruit characters

Or use any public image URL.

---

File-Based Definitions

Agents and skills can also be defined as files inside the folder a run works from. They are discovered from that working directory's .claude/ directory.

1. Write agent file to .claude/agents/{name}.md:

---
name: Sarah
role: PPC Specialist
skills: [campaign-reporting]
tools: [WebSearch, Read, Write]
---
You are Sarah, a PPC specialist who helps manage advertising campaigns...

2. Write skill file to .claude/skills/{name}/SKILL.md:

---
name: campaign-reporting
description: Generate campaign performance reports
allowed-tools: Bash, Read, Write
env: GA_API_KEY, OPENAI_API_KEY?
---
# Campaign Reporting Skill
...

The env field in skills supports these patterns:

• API_KEY --- required
• A | B | C --- any one of (OR)
• A, B --- all required (AND)
• KEY? --- optional

---

Error Handling

All actions return consistent response formats:

Success:

✅ Success
agent: "Sarah" (agent-abc123)

Validation error:

❌ Validation failed for createAgent: name: Required

API error:

❌ Error: Insufficient permissions

The tool validates data client-side before making API calls, catching common mistakes like missing required fields.