discourse_read_post

Retrieve content from a specific post in Discourse forums by providing its post ID for reading and analysis.

Instructions

Read a specific post.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`post_id`	Yes

Implementation Reference

src/tools/builtin/read_post.ts:16-34 (handler)

The asynchronous handler function that executes the tool logic: fetches the post data from Discourse API using the selected site's client, extracts relevant information like username, content, and URL, truncates if necessary, formats into a text response, and handles errors.

async ({ post_id }, _extra: any) => {
  try {
    const { base, client } = ctx.siteState.ensureSelectedSite();
    // Prefer raw by asking API for include_raw
    const data = (await client.getCached(`/posts/${post_id}.json?include_raw=true`, 10000)) as any;
    const username = data?.username || data?.user_id || "user";
    const created = data?.created_at || "";
    const raw: string = data?.raw || data?.cooked || "";
    const limit = Number.isFinite(ctx.maxReadLength) ? ctx.maxReadLength : 50000;
    const content = raw.slice(0, limit);
    const url = data?.topic_slug && data?.topic_id
      ? `${base}/t/${data.topic_slug}/${data.topic_id}/${data.post_number}`
      : `${base}/posts/${post_id}`;
    const text = `Post by @${username} (${created})\n\n${content}${raw.length > content.length ? `\n… (+${raw.length - content.length} more)` : ""}\n\nLink: ${url}`;
    return { content: [{ type: "text", text }] };
  } catch (e: any) {
    return { content: [{ type: "text", text: `Failed to read post ${post_id}: ${e?.message || String(e)}` }], isError: true };
  }
}

src/tools/builtin/read_post.ts:5-7 (schema)
Zod schema defining the input parameters for the tool: a positive integer post_id.
```
const schema = z.object({
  post_id: z.number().int().positive(),
});
```

src/tools/builtin/read_post.ts:9-35 (registration)

The server.registerTool call that registers the 'discourse_read_post' tool, specifying its name, metadata (title, description), input schema, and handler function.

server.registerTool(
  "discourse_read_post",
  {
    title: "Read Post",
    description: "Read a specific post.",
    inputSchema: schema.shape,
  },
  async ({ post_id }, _extra: any) => {
    try {
      const { base, client } = ctx.siteState.ensureSelectedSite();
      // Prefer raw by asking API for include_raw
      const data = (await client.getCached(`/posts/${post_id}.json?include_raw=true`, 10000)) as any;
      const username = data?.username || data?.user_id || "user";
      const created = data?.created_at || "";
      const raw: string = data?.raw || data?.cooked || "";
      const limit = Number.isFinite(ctx.maxReadLength) ? ctx.maxReadLength : 50000;
      const content = raw.slice(0, limit);
      const url = data?.topic_slug && data?.topic_id
        ? `${base}/t/${data.topic_slug}/${data.topic_id}/${data.post_number}`
        : `${base}/posts/${post_id}`;
      const text = `Post by @${username} (${created})\n\n${content}${raw.length > content.length ? `\n… (+${raw.length - content.length} more)` : ""}\n\nLink: ${url}`;
      return { content: [{ type: "text", text }] };
    } catch (e: any) {
      return { content: [{ type: "text", text: `Failed to read post ${post_id}: ${e?.message || String(e)}` }], isError: true };
    }
  }
);

src/tools/registry.ts:52-52 (registration)
The call to registerReadPost during the initialization of all tools in the registry, passing server, context, and options.
```
registerReadPost(server, ctx, { allowWrites: false });
```

Discourse MCP