For Agents & Builders

Typed read/write access
to a person's full context.
MCP + CLI.

Calendar, tasks, people, projects, knowledge, code, and memory — exposed as 60+ typed MCP tools with idempotent writes, scoped permissions, and full audit logging. Connect via MCP server or CLI. Every response includes timestamps and pagination.

State Primitives

Your agent gets typed access to:

Primitive	Read	Write	Examples
tasks	✓	✓	List by status, project, priority. Create, update, complete.
events	✓	✓	Calendar queries by date range. Create, update events.
people	✓	—	Search contacts, get org relationships, interaction history.
projects	✓	—	Full project context: tasks, signals, notes, members.
knowledge	✓	✓	Search notes, get documents. Create notes, upload docs.
code	✓	—	Semantic code search, git history, blame, expertise, conventions.
memory	✓	—	User preferences, learned attributes, behavioral patterns.

Trust Infrastructure

Every write is idempotent, scoped, and logged.

Idempotent writes

Pass idempotency_key on any create or update. Network timeouts, retries, race conditions — handled. Same key returns the existing resource, not a duplicate.

Permission scoping

Personal access tokens scope per domain: calendar, tasks, knowledge, code, people, projects. A calendar-only token returns 403 on task_create. Least privilege by default.

Approval gates

Destructive mutations (delete, archive) can require human confirmation. The agent pauses, the human approves in-app or via push notification, the agent resumes.

Audit logs

Every tool call logged: timestamp, agent identity, PAT ID, request parameters, response summary, latency. Query the audit trail via API. 90-day retention.

Budget ceilings

Set credit limits per agent, per run, per day. When a ceiling is hit, the agent stops and surfaces a notification — no silent cost overruns.

Freshness metadata

Every response includes last_updated and created_at timestamps. Your agent knows exactly how stale its context is. Unchanged resources return fast.

Connect in 5 Minutes

Two interfaces. 60+ tools. Same data.

MCP Server
// Claude Code, Cursor, Windsurf, etc.
{
  "mcpServers": {
    "execufunction": {
      "url": "https://mcp.execufunction.com/sse",
      "headers": {
        "Authorization": "Bearer exf_pat_..."
      }
    }
  }
}

Connect any MCP-compatible agent. 60+ tools across every domain.

CLI Tool
# Install and authenticate
npm install -g @execufunction/cli
exf auth login

# Same tools, from your terminal
exf tasks list --status next_action
exf calendar today
exf notes search "architecture"
exf code search "auth middleware"

Terminal-native access to your full workspace. Pipe into scripts, automate workflows, or use interactively.

"What's on my human's plate this week?"

calendar_list_events(startDate: "2026-02-17", endDate: "2026-02-23")

task_list(status: "in_progress")

task_list(status: "next_action", limit: 5)

Agent now knows: 3 meetings, 4 tasks in progress, 5 queued. Real state, not pasted context.

Why agents choose ExecuFunction

Fewer LLM calls to reach a decision.

One project_get_context call returns tasks, notes, signals, and members. No chain-of-thought steps to re-derive state that already exists.

Decisions grounded in real state.

Agents with actual calendar constraints, actual task backlogs, and actual relationship data outperform agents operating on pasted context.

Context the human generates for free.

The human uses the workspace normally. The agent benefits from that accumulated context automatically. No custom memory infrastructure to build.

Built for `cron`, not demos.

Structured errors with HTTP status codes. Idempotent writes. Scoped tokens. Audit trails. Pagination. Rate limits. Infrastructure for agents that run unsupervised.

Give your agent a workspace.

Get started in 2 minutes. Install the CLI or connect via MCP — same tools, same data, your choice of interface.

MCP Quickstart CLI Installation → Full Docs →

Typed read/write access to a person's full context. MCP + CLI.