MCP¶

webarchiver exposes the same capability three ways: as MCP tools an agent calls, as a Python API you import, and as a CLI.

As an MCP server¶

Once deployed, the server registers one typed, read-only tool modules. Each module is independently togglable with a *TOOL environment flag.

As a Python API¶

from webarchiver.auth import get_client

api = get_client()        # builds the fetch policy from SDK settings
result = await api.extract("https://example.org/article")
print(result.page.title, result.page.canonical_url)

As a CLI¶

Optionally tune the fetch policy (WEBARCHIVER_MIN_INTERVAL_S, WEBARCHIVER_MAX_BYTES, WEBARCHIVER_OBEY_ROBOTS), then run webarchiver-mcp --transport stdio.

Explicit HTTP endpoint¶

HTTPS remains the default. For an existing HTTP-only endpoint you are authorized to use, set WEBARCHIVER_BASE_URL=http://service.example.invalid and explicitly set WEBARCHIVER_ALLOW_PLAINTEXT=true. This forwards the SDK's HttpClientOptions.allow_plaintext choice only to that vendor client. Requests remain bound to its exact scheme, host and port; redirects stay disabled. HTTP carries requests without TLS encryption. The flag never retries HTTPS over HTTP, changes certificate verification, or authorizes another endpoint. OIDC token exchanges retain their separate TLS policy. Leave the flag false for HTTPS.