Content Tools

archival.dev

An MCP server for Archival that lets AI write website source files, start preview sessions, await approval, check status, and reference documentation.

1 endpoint9 known toolsFirst detected September 14, 2026Last detected September 14, 2026

ENDPOINT 1

https://api.archival.dev/mcp

No auth detected

MCP server metadata

Name: archival
Version: 1

Capabilities

tools.listChangedprompts.listChanged

Server instructions

These tools build and publish an Archival website - a static site built from TOML content files and Liquid templates - to a shareable preview URL. Everything is scoped to a session. Call archival_start_session first, show the person the link and code it returns, then poll archival_await_approval until it reports ready. Nothing else works until then. You do not need a local working directory, a checkout, or the archival CLI. Write the site's source with archival_write_files and call archival_publish; the build happens on our side with a pinned archival binary, and build errors come back as tool output for you to fix. archival_publish is also the only way to find out whether what you wrote is valid, so publish as soon as one page renders rather than at the end. A session gets 20 publishes: spend them on a preview the person can look at early and react to. Give the person the "url" these tools return, and only that. The "siteUrl" beside it is the raw origin the site is served from - it has no way back to keeping the site, so it is not a link to hand anyone. Read archival_reference with topic "building" for the whole flow, and topic "authoring" before writing any template. Do not write Archival syntax from memory - the object/manifest format is specific and it has changed. Previews are public, expire, and are reviewed. Never invent facts about a real person or business: ask, or leave it out.

Known tools 9

archival_start_session

Begins a preview session and returns a link plus a code.

Inferred read-only

archival_await_approval

Returns {"status":"waiting"} until the person approves, then the preview's name and URL.

Inferred read-only

archival_reference

Archival's own documentation.

Inferred read-only

archival_status

The preview's name, URL, expiry, how many publishes are left, and every source file it currently holds.

Inferred read-only

archival_write_files

Writes the site's source - archival.toml, archival_objects.toml, objects/, pages/, layout/, public/.

Inferred read-only

archival_read_file

The current contents of one source file, for when you need to see what is actually there rather than what you last sent.

Inferred read-only

archival_delete_files

Deletes source files.

Inferred read-only

archival_upload_media

Uploads media to the CDN.

Inferred read-only

archival_publish

Builds the source you have written with Archival and publishes the result.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.archival]
url = "https://api.archival.dev/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "archival": {
      "type": "http",
      "url": "https://api.archival.dev/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: archival
Remote MCP URL: https://api.archival.dev/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "archival": {
      "url": "https://api.archival.dev/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "archival": {
      "type": "http",
      "url": "https://api.archival.dev/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "archival",
  "transport": "streamable-http",
  "url": "https://api.archival.dev/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

TRUST AND VERIFICATION EVIDENCE

Loading Trust v2 evidence…

Checking the associated registrable domain. The BuiltWith key remains server-side.

Indexed

Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.