Query the Indian Liberals archive (Centre for Civil Society) from any AI client or agent framework. Read-only, no authentication, free.
1531 works · 695 thinkers · 195 musings · 59 opinions · 52 organisations · 56 ThePrint columns
MCP (Streamable HTTP): https://mcp.indianliberals.in/mcp
REST: https://mcp.indianliberals.in/api/<tool>
OpenAPI 3.1 spec: https://mcp.indianliberals.in/openapi.jsonSettings → Connectors → Add custom connector → URL:
https://mcp.indianliberals.in/mcpclaude mcp add --transport http indian-liberals https://mcp.indianliberals.in/mcpConnector (Settings → Connectors → Advanced → Developer mode → Add): use the MCP URL above — the required search and fetch tools are provided.
Custom GPT Action: in the GPT editor choose "Create new action" → "Import from URL" → https://mcp.indianliberals.in/openapi.json (auth: none).
{
"mcpServers": {
"indian-liberals": { "url": "https://mcp.indianliberals.in/mcp" }
}
}Clients that only speak stdio: npx mcp-remote https://mcp.indianliberals.in/mcp
"mcp_servers": [{ "type": "url", "url": "https://mcp.indianliberals.in/mcp", "name": "indian-liberals" }]"tools": [{ "type": "mcp", "server_label": "indian-liberals", "server_url": "https://mcp.indianliberals.in/mcp", "require_approval": "never" }]Use the MCP URL with any MCP-capable SDK, or call the REST endpoints directly as function tools — try /api/search_corpus?query=swatantra party.
read_index | The curated index of the archive (llms.txt): what the collection holds, how it is organised, and the canonical browse URLs. Start here for orientation. |
list_thinkers | List thinker profiles with bio snippets. Filterable by tradition, canon status, vocation, or free-text query. Includes referenced non-liberal figures (canon_status="referenced") — do not describe those as Indian liberals. |
list_works | The works catalogue: books, pamphlets, speeches, essays, periodical issues, interviews and more. Filterable by author, work_type, theme, language, series, year range, or free-text query. |
get_work_metadata | Full structured metadata for one work: authors, publication, AI summary and key points, table of contents, per-essay summaries, provenance, rights, pdf_url (Tier B) or transcript pointer (interviews, Tier A). |
read_clean_content | Read the full clean-markdown body of a Tier A document (thinker profile, organisation page, musing/excerpt, opinion, interview transcript, ThePrint column). Paragraphs carry <!-- #p-xxxxxx --> anchors — cite "<url>#p-xxxxxx". Tier B works are refused (only AI summaries exist for them — use get_work_metadata). |
get_passage | Fetch specific paragraphs of a Tier A document by their stable paragraph IDs (the citation primitive). IDs come from read_clean_content annotations (<!-- #p-xxxxxx -->). |
search_corpus | Full-text search across every content kind (thinkers, works, musings, opinions, organisations, interviews, ThePrint columns). Results are tier-flagged: Tier A may be quoted with paragraph citations; Tier B hits matched an AI summary, not body text. |
find_related | TF-IDF cross-links: the most textually related entries to a given document, across content kinds (same data as the on-page "Related" sections). |
search | Search the archive. Returns matching documents with ids usable by the fetch tool. (Alias of search_corpus in the shape ChatGPT connectors require.) |
fetch | Fetch the full content of a document found via search. Tier A returns the complete clean markdown; Tier B returns the structured metadata + AI summary + PDF link. (ChatGPT-connector counterpart of read_clean_content / get_work_metadata.) |
Tier A (thinker profiles, organisations, musings, opinions, interview transcripts, ThePrint mirror): quote freely, cite <url>#p-xxxxxx paragraph anchors.
Tier B (primary-work PDFs): tools return AI summaries — attribute claims as "Indian Liberals' summary of <work>" and link the PDF. Full policy: indianliberals.in/AGENTS.md.
The server is a stateless proxy over the archive's build artifacts — it updates automatically as the collection grows. Maintained by CCS; rebuilt by Thothica.