# How to Build a Company Brain Workflow for Employee Onboarding

An employee onboarding company brain workflow centralizes handbooks, standard operating procedures, and team playbooks into an intelligent workspace accessible to new hires and AI agents. By organizing documents into designated workspaces and indexing content on arrival, organizations eliminate setup bottlenecks and support self-serve inquiries. With citation-backed search and Model Context Protocol access, new employees get checkable answers from authoritative company files.

Source: https://fast.io/resources/company-brain-workflow-employee-onboarding/
Author: [Tom Langridge](https://fast.io/authors/tom-langridge/)
Last reviewed: 2026-11-13

## Why Employee Onboarding Stalls Across Fragmented Documentation

A survey by Gartner found that 47% of digital workers struggle to find information or data needed to effectively perform their jobs. For new employees entering an organization, this information friction is sharp during their initial ninety days. New hires must absorb operational procedures, understand cross-functional dependencies, and configure software environments. Yet the critical knowledge required to get up to speed remains scattered across disconnected silos: HR policies in Google Drive, IT setup runbooks in personal folders, departmental playbooks in wikis, and administrative answers buried in chat channels.

A 2018 BambooHR onboarding study found that 31% of employees have quit a job within the first six months, so early orientation and enablement deserve attention. When new hires cannot find authoritative answers independently, they spend hours waiting for colleagues to reply or navigating contradictory documentation left behind by former team members.

An employee onboarding company brain is a centralized, AI-indexed workspace containing organizational policies, setup guides, and team playbooks that new hires and onboarding agents query for source-attributed guidance.

Without a centralized, intelligent workspace, employee onboarding encounters three distinct operational breakdowns:

1. **Documentation Drift:** Without centralized version tracking, onboarding documentation quickly becomes outdated as company policies and engineering toolchains change. New hires follow stale setup guides, encounter broken installation links, and reach out to team leads with problems that were resolved months earlier.
2. **The Asynchronous Support Bottleneck:** In distributed and hybrid teams, mentors cannot sit beside new hires to answer every operational question. When new employees must ask basic questions about hardware procurement or staging credentials in public chat rooms, response delays slow their ramp-up and create friction.
3. **Unverified Knowledge Retrieval:** General chat assistants without direct access to company records hallucinate policies or guess internal procedures. When new hires ask an ungrounded bot how to request software licenses or submit expense reports, the assistant invents plausible-sounding steps that contradict actual company rules.

Establishing an intelligent [company brain](/product/company-brain/) solves this fragmentation by treating onboarding assets as an active knowledge layer. Rather than leaving files stranded across individual desktop drives and isolated cloud silos, teams centralize their onboarding records in a dedicated workspace. With Intelligence Mode enabled, files are indexed on arrival for full-text and semantic search. Human new hires and automated AI agents query the workspace using natural language and receive checkable answers that cite the source document, the page where one applies, and the supporting passage.

### Comparing Employee Onboarding Knowledge Architectures

Human resources and engineering leaders evaluating onboarding strategies often compare traditional wikis, static cloud drives, enterprise search overlays, and intelligent workspaces. The comparison table below outlines how these architectures perform across core onboarding requirements:

| Knowledge Architecture | File Ingestion & Indexing | Search Grounding & Accuracy | AI Agent Integration | Access Governance |
| :--- | :--- | :--- | :--- | :--- |
| **Static Cloud Drives** (Google Drive, Dropbox, Box, OneDrive) | Stores static files; indexing varies by provider and configuration | Cloud search and assistant answers over stored files | MCP servers and connectors available; working files live across external folders | Drive-level or folder-level permissions; sharing set per file or folder in each drive |
| **Traditional Wikis** (Notion, Confluence) | Requires copying policies into wiki pages; attachments often managed separately | Search across wiki pages and supported attachments | MCP servers connect to wiki pages, while external working files remain separate | Page-level permissions maintained per space and page |
| **Enterprise Search Overlays** (Glean, Guru) | Connects search crawlers over external silos; files remain stored elsewhere | Semantic search across connected tools; search quality depends on external connector health | MCP servers query across silos, without direct workspace file coordination | Inherits permissions from external sources; sync lags can cause permission drift |
| **Intelligent Workspaces** (Fast.io) | Indexes PDFs, docs, spreadsheets, and notes on arrival via Intelligence Mode | Hybrid search combining semantic meaning, keyword matching, and typed metadata with cited sources | Remote MCP server with working files in the workspace to search, lock, and co-edit Notes | Role-based isolation per organization, workspace, and share; no folder-level complexity |

Intelligent workspaces bridge the gap between static storage and active enablement. Instead of expecting team leads to maintain duplicate documentation in an internal wiki, the workspace indexes the actual policy documents, software runbooks, and orientation decks the business already maintains.

## Architecting Document Hierarchy and Access Boundaries for New Hires

Designing an effective onboarding company brain requires deliberate workspace architecture and structured file categories rather than an unmanaged folder dump. If administrators place sensitive personnel records in the same directory as engineering runbooks, search relevance degrades and private compensation data risks exposure.

In Fast.io, permissions are configured per organization, workspace, and share through member roles: owner, admin, member, guest, and view. There are no folder-level or file-level permissions within a workspace; every workspace member reaches all workspace contents according to their assigned role. 

Because permissions do not exist at the folder or file level, material that only certain employees or managers may see must live in its own workspace or share. For employee onboarding, organizations structure workspaces into clean operational containers based on audience and role requirements.

### Workspace Isolation and Role-Based Permissions

To balance open knowledge sharing with necessary confidentiality, organizations establish isolated workspaces across three functional tiers:

* **Company-Wide Onboarding Workspaces (`onboarding-general`):** Accessible to all incoming employees. Holds employee handbooks, healthcare coverage overviews, 401(k) enrollment policies, annual holiday calendars, organizational charts, and standard communication norms.
* **Departmental Onboarding Workspaces (`onboarding-engineering`, `onboarding-sales`, `onboarding-marketing`):** Dedicated workspaces tailored to specific job functions. For instance, the engineering onboarding workspace holds Git branch guidelines, staging environment architecture, API documentation, local Docker setup instructions, and testing frameworks.
* **Confidential HR Operations Workspaces:** Restricted to human resources staff and hiring managers. Contains compensation bands, offer letters, background check records, and signed legal agreements.

A workspace join setting defaults to letting any member of the organization join. To keep departmental or restricted onboarding workspaces confined to designated mentors and incoming hires, administrators should restrict this setting upon workspace creation.

By default, Fast.io hybrid search and Intelligence Mode answer questions only from the workspaces and shares the asking person or agent belongs to. An incoming sales representative querying the company brain cannot retrieve technical deployment runbooks from a private engineering workspace or salary data from the HR workspace, preserving clean information boundaries.

### Core Documentation Pillars for New Hire Enablement

A complete onboarding company brain organizes documents into four primary operational pillars:

1. **Operational Policies and Governance Handbooks:** Official company handbooks, code of conduct guides, travel and expense reimbursement rules, remote work hardware policies, and compliance disclosures.
2. **Technical Setup Runbooks and Environment Blueprints:** Step-by-step installation guides for development environments, VPN and network authentication steps, database seeding runbooks, and software license request procedures.
3. **Departmental SOPs and Functional Playbooks:** Product roadmaps, customer support escalation matrices, code review rubrics, outbound sales email sequences, and design system asset libraries.
4. **Collaborative Notes and 30-60-90 Day Milestones:** Collaborative Notes kept right beside onboarding files. Notes co-edit live with labeled cursors to maintain personal 30-60-90 day milestone roadmaps, mentor pairing details, and open questions recorded during orientation check-ins.

### Automating Document Ingestion with Scheduled Cloud Sync

Organizations rarely author all onboarding materials in a single tool. HR teams draft policies in Google Drive, IT teams store software license instructions in Dropbox, and sales operations maintain pitch decks in OneDrive or Box. 

To keep the onboarding company brain synchronized without manual copying, Fast.io provides Cloud Sync. Cloud Sync connects Google Drive, Dropbox, Box, and OneDrive (including SharePoint libraries through the OneDrive connector) directly to a workspace. Sync operates one-way or two-way, on a schedule or on demand. It is never continuous, live, or real-time, allowing administrators to establish predictable synchronization windows, such as a nightly sync that pulls updated policy PDFs from HR's shared drive.

Files synced from an external cloud drive adopt the Fast.io workspace permissions, not the sharing settings of the source drive. This design ensures that files imported from a broadly shared corporate drive automatically inherit the role-based access rules of the target onboarding workspace. Teams can also bring files in by direct upload, by Cloud Import, which brings files in once from Google Drive, Dropbox, Box, or OneDrive through OAuth, or by URL import from a public URL, so imported files do not pass through a local machine.

## Five Steps to Build an Employee Onboarding Company Brain Workflow

Building a structured employee onboarding company brain transforms static orientation manuals into a responsive self-service environment. Following a clear, sequential implementation plan ensures that documentation remains organized, permissions remain secure, and new hires can find checkable answers on their first day.

### Step 1: Curate Core Onboarding Docs and Structure Workspaces
Begin by establishing your organization's onboarding workspace layout and gathering authoritative documentation:

1. In your Fast.io organization, create a new workspace.
2. Name the workspace according to function, such as `onboarding-general` for company-wide orientation or `onboarding-engineering` for technical staff.
3. Adjust the workspace access policy: restrict the join setting so that organization members cannot join automatically without an explicit invitation.
4. Organize the workspace directory with numbered folders to maintain intuitive visual hierarchy: `01-handbooks-and-policies`, `02-it-and-security`, `03-departmental-playbooks`.
5. Upload core operational documents, including employee handbooks, benefits summaries, and IT setup checklists. Maximum single upload sizes are `25 GB` on Starter, `50 GB` on Business, and `100 GB` on Enterprise.

### Step 2: Enable Workspace Intelligence for Hybrid Semantic and Full-Text Search
Transform static files into an active, searchable knowledge base:

1. Navigate to workspace settings and toggle on **Intelligence Mode**.
2. When Intelligence is enabled, all existing and incoming files are automatically processed and indexed.
3. Fast.io builds a hybrid search index combining full-text keyword indexing with semantic vector retrieval. The workspace indexes narrative text across PDFs, Word documents, spreadsheets, presentations, and Markdown notes without requiring external vector databases or manual chunking scripts.

### Step 3: Connect Cloud Sync to Centralize Legacy Drive Folders
Reduce document duplication by synchronizing existing drive folders:

1. In workspace settings, configure **Cloud Sync** to connect external storage folders from Google Drive, Dropbox, Box, or OneDrive. If your organization uses SharePoint document libraries, connect them directly via the OneDrive connector.
2. Select one-way synchronization from your HR or IT master drive into the onboarding workspace. This allows HR administrators to continue updating documents in their familiar drive while automatically pushing current policies to the company brain.
3. Set a recurring synchronization schedule (such as every 24 hours) or trigger an immediate sync when new company policies are published.

### Step 4: Configure Member Access and Scoped Folder Shares
Provision access for incoming personnel and external contractors:

1. Invite new full-time hires to the workspace using the `member` or `guest` role. Members can read documents, participate in Collaborative Notes, and execute search queries across all workspace files.
2. When hiring temporary contractors, external recruiters, or seasonal interns who only require access to specific documents (such as an independent contractor compliance guide), create a scoped [file share](/product/sharing/).
3. Configure custom share controls: set optional expiration dates, limit access to named recipients, or disable file downloads entirely so that contractors review sensitive policies in a browser without saving local copies. External guests who view shares do not consume user seats and can open links without creating an account.

### Step 5: Deploy AI Onboarding Query Prompts and Check Source Citations
Equip new hires to query the company brain and check answers against source citations:

1. Share recommended query prompts during orientation check-ins to help new hires find answers independently:
   * *What is the standard reimbursement policy for home office internet expenses?*
   * *Where are the local Docker setup instructions for running the payment microservice?*
   * *What are the company holidays observed during the fourth quarter?*
2. Fast.io Intelligence returns concise, natural-language answers backed by source citations. Each citation shows the file name, the page where one applies, and the supporting snippet.
3. New hires can click the citation link to open the authoritative document directly, confirming the context and checking the underlying source text.

## Connecting AI Onboarding Assistants via Model Context Protocol

While human employees query the company brain through browser search bars, modern engineering and operations teams increasingly deploy autonomous AI agents to guide new hires. Developers use coding assistants in Cursor or Claude Code to inspect repository architecture, while HR teams use conversational assistants in Claude or ChatGPT to triage onboarding inquiries.

Fast.io provides native Model Context Protocol (MCP) connectivity, allowing AI agents to query the onboarding workspace directly over Streamable HTTP without local package installations or bridge daemons. Setup instructions are detailed at https://mcp.fast.io/docs, with full tool specifications at https://mcp.fast.io/skill.md and agent onboarding guidance at https://fast.io/llms.txt. Learn more on our [storage for agents page](/storage-for-agents/).

### Connecting MCP Clients over Streamable HTTP

Fast.io hosts a remote MCP server over Streamable HTTP with dedicated endpoints tailored to specific client environments:

* **Interactive Personal Assistants:** Claude (web, desktop, and mobile), Claude Cowork, Manus, and general personal-agent runtimes connect to `https://mcp.fast.io/mcp/tools`. Sign-in uses browser OAuth, where the person connecting reviews permissions, chooses Read Only or Read & Write, and selects allowed workspaces.
* **Coding Agents in IDEs and Terminals:** Claude Code, Cursor, Cline, Devin Desktop, VS Code Copilot agent mode, and Gemini CLI connect to `https://mcp.fast.io/mcp/code`.
* **ChatGPT and Codex:** Teams install the Fastio plugin from the ChatGPT plugin directory, or configure a custom MCP connector pointing to `https://mcp.fast.io/mcp/operations`.
* **Framework Agents and Headless Scripts:** Background automation running LangGraph, CrewAI, AutoGen, or custom Python scripts connects to `https://mcp.fast.io/mcp/tools` or `/mcp/code` and passes a scoped API key as an HTTP header: `Authorization: Bearer <api key>`.

To configure Cursor for an incoming developer, add the server configuration to `~/.cursor/mcp.json` or `.cursor/mcp.json` in the project root:

```json
{
  "mcpServers": {
    "fastio-onboarding": {
      "url": "https://mcp.fast.io/mcp/code"
    }
  }
}
```

In Claude Desktop or Claude Web, open **Customize**, select **Connectors**, add a custom connector with `https://mcp.fast.io/mcp/tools`, complete OAuth sign-in, and enable the connector in chat sessions.

### Consolidated MCP Toolset and Advisory File Locks

When interacting with the onboarding company brain, AI agents use Fast.io's consolidated MCP toolset. Read actions and write actions are cleanly partitioned across dedicated tools:

* On `https://mcp.fast.io/mcp/tools`, read operations use the `storage` tool, while modifications (moves, deletions, lock management) use `storage_manage`, and uploads use `upload_manage`.
* On `https://mcp.fast.io/mcp/code`, read requests use `search` and `execute`, while write operations use `execute_manage` and `upload_manage`.

When multiple autonomous agents or developer scripts generate onboarding documentation simultaneously, they coordinate modifications using advisory file locks:

1. An agent requests an advisory lease using `lock-acquire` on `storage_manage`.
2. Other connected agents check `lock-status` on `storage` to determine who holds the active lease.
3. The lease expires automatically unless heartbeated by the holding agent, and releases via `lock-release` upon task completion.

Because file locks are strictly advisory, concurrent writes land safely without throwing fatal lock errors, and per-file version history preserves every revision for complete auditability.

### Extracting Onboarding Records with Metadata Views

Employee onboarding generates dozens of structured records: signed policy acknowledgments, hardware receipt forms, benefits election sheets, and emergency contact forms. Manually entering this data into HR spreadsheets is time-consuming.

To automate this data extraction, teams use [Metadata Views](/product/document-data-extraction/). Metadata Views turn unstructured onboarding documents into a live, queryable database. Users describe the fields they want extracted in natural language, and AI designs a typed schema that extracts data across PDFs, scanned sheets, Word files, and spreadsheets without manual OCR rules.

Metadata Views support multiple typed column formats:

* **Text:** Employee Name, Department, Hardware Serial Number, Manager Name
* **Date & Time:** Employment Start Date, Policy Acknowledgment Date, Equipment Receipt Date
* **Integer & Decimal:** Office Stipend Amount, Hardware Value, Allocated Leave Days
* **Boolean:** NDA Signed, IT Security Training Completed, Direct Deposit Verified
* **URL & JSON:** Signed Document Link, Hardware Configuration Payload

When a new hire uploads a signed acknowledgment or an equipment receipt to the workspace, Metadata Views extract the required values automatically into a filterable spreadsheet view. HR managers can add new columns at any time without reprocessing original files.

### Coordinating Agent-to-Human Ownership Transfer

Automated onboarding workflows often begin before a new human manager is actively managing the workspace. For example, an IT provisioning agent or an automated HR bot might provision the initial organization, establish workspaces, ingest documentation, and invite team members.

Fast.io supports clean ownership transfer from agents to humans:

1. The provisioning agent sets up the onboarding workspaces and ingests all core documentation.
2. The agent hands off the organization to the human HR director or department lead using a claim link.
3. Once claimed, the human lead takes over organizational ownership and governance. An owner can also transfer a workspace or organization directly to another member, in which case the previous owner becomes an admin.

## Maintaining Knowledge Freshness and Tracking Onboarding Progress

Internal onboarding documentation quickly becomes outdated without centralized version tracking and active review cycles. When engineering toolchains update and company policies evolve, unmanaged guides begin to contradict current operational standards. Without a unified operational repository, documentation maintenance is often treated as an afterthought: changes are announced in ephemeral chat channels, while orientation PDFs remain untouched for quarters. Months later, incoming employees follow stale setup guides, causing environment errors and avoidable management escalations.

Fast.io reduces this operational drift by centralizing documentation in an intelligent workspace backed by per-file version history, detailed change logging, and predictable subscription capacity. Maintaining document accuracy ensures that the company brain continues to serve both new hires and tenured employees effectively throughout the organization's growth.

### Preserving Operational Auditability with Version History and the Activity Log

Whenever an HR manager updates a remote work policy, an engineer revises an architecture diagram, or an IT administrator issues an updated credential runbook, uploading the replacement file creates a new version entry. Every file in Fast.io maintains complete version history, allowing team members to review previous versions and restore one if a newly introduced instruction proves incorrect.

Alongside version control, Fast.io maintains a **detailed activity log**. The activity log records changes made across the organization and workspaces, including file uploads, edits, moves, deletions, membership changes, and share creations. The activity log focuses strictly on operational modifications: it does not record read-only events such as file views or downloads, and does not serve as an access log. It provides administrative visibility into when policies were revised, which team member uploaded new guidelines, and when shared onboarding folders were created.

For technical teams operating custom HR notification bots, Fast.io provides an **events feed**. External applications subscribe to workspace activity over a WebSocket or long-poll connection (`GET /current/activity/poll/{entity_id}?wait=95&lastactivity={timestamp}`). When an HR lead uploads a revised employee handbook or an incoming developer submits a hardware acknowledgment, the events feed alerts connected systems immediately.

### Transitioning New Hires from Onboarding to Daily Operations

As employees complete their initial ninety days, their relationship with the company brain evolves. Rather than querying basic setup guides, tenured employees transition into productive contributors who author documentation and collaborate in departmental workspaces.

1. **Expanding Workspace Membership:** Once general orientation concludes, employees are added to active project workspaces (such as `eng-payments` or `marketing-campaigns`) where daily collaboration occurs.
2. **Maintaining Personal Orientation Logs:** Collaborative Notes created during onboarding remain saved beside project documentation, serving as historical records of mentor check-ins and architectural decisions.
3. **Continuous Knowledge Contributions:** When an engineer discovers an undocumented edge case or resolves a tricky staging environment issue, they update the workspace runbook directly. Version history records the update, and Intelligence Mode re-indexes the new version so the next new hire benefits from the resolution.

### Transparent Subscription Plans and Resource Allocation

Fast.io operates on a transparent subscription model based on workspaces, storage, seats, and usage credits. Every organization runs on a paid subscription.

Monthly plans start with a 14-day free trial that requires a credit card:

| Plan | Monthly Pricing | Annual Pricing | Included Seats | Included Storage | Workspaces | Included Credits |
| :--- | :--- | :--- | :--- | :--- | :--- | :--- |
| Starter | `$29/mo` | `$290/yr` | `5` seats | `1 TB` | `10` | `300,000` / mo |
| Business | `$99/mo` | `$990/yr` | `20` seats | `10 TB` | `100` | `1,200,000` / mo |
| Enterprise | `$299/mo` | `$2,990/yr` | `50` seats | `50 TB` | `300` | `4,500,000` / mo |

* **Starter (`$29/mo`)**: Includes `5` seats, `1 TB` storage, `10` workspaces, and `300,000` credits per month. Maximum upload size is `25 GB`.
* **Business (`$99/mo`)**: Includes `20` seats, `10 TB` storage, `100` workspaces, and `1,200,000` credits per month. Maximum upload size is `50 GB`.
* **Enterprise (`$299/mo`)**: Includes `50` seats (with additional seats at `$1` a month each, up to `200`), `50 TB` storage, `300` workspaces, and `4,500,000` credits per month. Maximum upload size is `100 GB`.

Annual plan pricing is available at `$290`, `$990`, and `$2,990` per year.

Credits cover AI work: chat, file ingestion, agent runs, and document processing. Overage is $10 per 100,000 credits. Storage beyond the plan is $0.015 per GB a month and extra bandwidth $0.04 per GB, billed separately and not in credits. Review full tier details on the [Fast.io pricing page](/pricing/).

External contractors and temporary hires who access onboarding materials through scoped shares do not consume member seats and can view shared files directly in their browser without an account.

## Frequently asked questions

### How do you organize documents for an onboarding company brain?

Organize documents by structuring isolated workspaces based on audience and confidentiality: a general workspace for company-wide handbooks and benefits, department-specific workspaces for technical runbooks or sales playbooks, and restricted workspaces for sensitive HR records. Within each workspace, use numbered folders (such as 01-handbooks-and-policies, 02-it-and-security, 03-departmental-playbooks) to maintain clear hierarchy before enabling Intelligence Mode for hybrid semantic search.

### Can AI agents assist new employees during onboarding?

Yes. AI agents connected via the Model Context Protocol (MCP) can query the onboarding company brain directly. Coding assistants like Cursor or Claude Code connect over Streamable HTTP to search architecture documentation, while conversational assistants like Claude or ChatGPT answer policy questions using cited excerpts from uploaded handbooks.

### How do permissions work when syncing employee handbooks from Google Drive?

Files synced from Google Drive, Dropbox, Box, or OneDrive adopt the role-based permissions of the destination Fast.io workspace, rather than the sharing permissions of the source drive. Access is managed by member role (owner, admin, member, guest, or view) at the organization, workspace, and share levels, ensuring that synchronized handbooks are governed by consistent internal access policies.

### How does Intelligence Mode provide citations for onboarding questions?

When Intelligence Mode is enabled on an onboarding workspace, files are indexed upon arrival using a hybrid engine combining full-text keyword matching with semantic vector search. When an employee or AI agent queries the workspace, the system retrieves relevant text segments and generates answers that cite the file name, the page where one applies, and the supporting passage, allowing new hires to check answers against official sources.

### How do Metadata Views help HR teams manage onboarding documents?

Metadata Views convert unstructured onboarding documents into a live, queryable database without manual OCR templates. HR teams define required fields in natural language, and the system extracts typed data (such as employee start dates, equipment serial numbers, signed policy acknowledgments, and emergency contact details) into a filterable spreadsheet view across PDFs, spreadsheets, and scanned forms.

### What happens when an employee handbook or setup guide is updated?

When an updated guide or handbook is uploaded, Fast.io retains full per-file version history, allowing teams to review previous revisions or restore earlier drafts. Intelligence Mode updates its search index automatically, and the detailed activity log records the file modification, timestamp, and author, so new hires query the current version while administrators keep a record of each change.

## Sources

- [EMSNow: Gartner Survey on Digital Workplace Information Friction](https://www.emsnow.com/gartner-survey-reveals-47-of-digital-workers-struggle-to-find-the-information-needed-to-effectively-perform-their-jobs/): A survey by Gartner found that 47% of digital workers struggle to find information or data needed to effectively perform their jobs.
- [BambooHR: HR and Employee Engagement Statistics (BambooHR 2018 Onboarding Study)](https://www.bamboohr.com/blog/58-hr-statistics-you-need-to-know/): A 2018 BambooHR onboarding study found that 31% of employees have quit a job within the first six months.

## About Fast.io

Fast.io provides shared workspaces where people and AI agents work on the same files, with built-in semantic search and citation-backed chat over what they hold. Agents reach it through a remote MCP server, a REST API at https://api.fast.io/current/, and a command line client published on npm as @vividengine/fastio-cli. MCP setup is at https://mcp.fast.io/docs: Claude and most MCP clients connect to https://mcp.fast.io/mcp/tools, ChatGPT to https://mcp.fast.io/mcp/operations, and coding agents to https://mcp.fast.io/mcp/code.