# Flower Company Information and Published Content MCP

Use this service when you need Flower Computer Company's current public company information or published writing through MCP.

Company profile last reviewed: 2026-08-24.

Endpoint: https://mcp.flowercomputer.com/

No authentication is required. All tools are read-only and do not change Flower data.

## Interfaces and limits

Browser WebMCP exposes four camelCase tools (`getFlowerCompanyOverview`, `listFlowerPosts`, `getFlowerPost`, `getFlowerApiStatus`) on pages that register it. The remote MCP service is the canonical cross-client surface and exposes six snake_case tools at https://mcp.flowercomputer.com/; its tools are `get_company_overview`, `list_posts`, `get_post`, `search_posts`, `get_contact_information`, `get_api_status`. Both interfaces are unauthenticated and read-only.

No enforced quota is currently advertised. Clients should avoid rapid polling and must honor any future `429 Too Many Requests` response, `Retry-After` header, and `RateLimit` headers.

## Connect

Send Streamable HTTP requests to the endpoint. Include `Accept: application/json, text/event-stream` and `Content-Type: application/json` on JSON-RPC POST requests.

1. Send an `initialize` request with a supported MCP protocol version.
2. Send `notifications/initialized` after the server replies.
3. After initialization, include `MCP-Protocol-Version: 2025-11-25` on every subsequent HTTP request.
4. Call `tools/list` to discover the executable tools before calling `tools/call`.

## Read-only tools

| Tool | Purpose | Required input |
| --- | --- | --- |
| `get_company_overview` | Use when an agent needs Flower's canonical description, mission, current or prior work, public links, or high-level contact data. | None |
| `list_posts` | Use to browse Flower's published posts or discover valid slugs before calling get_post. | None |
| `get_post` | Use when an agent already has a published Flower slug and needs the full Markdown article. | `slug` — Published post slug returned by list_posts or search_posts. |
| `search_posts` | Use to search Flower post metadata by title, excerpt, category, author, tag, or slug when a full article body is not needed. | `query` — Words or phrase to match against published post metadata. |
| `get_contact_information` | Use when an agent needs Flower's canonical public contact details and public links without the full company overview. | None |
| `get_api_status` | Use when an agent needs to check availability and version of Flower's public content API. | None |

## Representative requests

Initialize with a JSON-RPC request such as `{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"flower-quickstart","version":"1.0.0"}}}`. Then use `tools/list` to inspect the current schemas.

Call the company overview with `{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"get_company_overview","arguments":{}}}`.

Call `list_posts` before `get_post`; pass the returned `slug` to `get_post` rather than deriving a slug from a title. Use `search_posts` when article metadata is sufficient, and `get_contact_information` or `get_api_status` for their narrower read-only queries.

## Reliability and citations

- Retry transient network failures and errors marked `retryable: true` with bounded exponential backoff. Honor a `retryAfter` value when present.
- Do not retry an unchanged terminal error. Read its resolution, refresh discovery with `tools/list`, and use a valid tool name and input schema.
- Cite a post with its returned `htmlUrl`. Keep `markdownUrl` for readable-source provenance and `apiUrl` for machine-retrieval provenance. Use `htmlUrl` when rendered media matters.

## Discovery

- [MCP Server Card](https://www.flowercomputer.com/.well-known/mcp/server-card.json)
- [Public content API](https://www.flowercomputer.com/api/)
- [Agent Skill](https://www.flowercomputer.com/.well-known/agent-skills/flower-public-content/SKILL.md)