Setup & usage guide
Connect the litscan-rag knowledge base to your coding agent in a couple of minutes. Two ways: the one-file fast path, or manual setup.
Fast path — one file, zero typing
Sign in to your dashboard and click ↓ AI setup file. You download a doc with your API key already filled in. Open Claude Code (or Codex) in any folder, hand it the file, and say “set this up” — the agent adds the MCP server and installs the skills for you.
Manual setup
1 · Get your key
Register at the dashboard with an academic (.edu / .ac) or Gmail address. Your key looks like sk-litscan-… and goes live within ~3 minutes. 1000 calls/day, free for researchers.
2 · Connect Claude Code
Run this in your terminal (paste your key):
claude mcp add --transport http -s user litscan-rag \ https://mcp.acceptpaper.com/mcp \ -H "X-API-Key: YOUR_KEY"
3 · Connect Codex (alternative client)
Add to ~/.codex/config.toml. It is an HTTP streamable MCP — adapt the field names to your Codex version if they differ:
[mcp_servers.litscan-rag]
url = "https://mcp.acceptpaper.com/mcp"
http_headers = { "X-API-Key" = "YOUR_KEY" }4 · Install the two skills
One line drops research-mentor and research-explorer into ~/.claude/skills/ (Claude Code). For Codex, paste each SKILL.md into your project's AGENTS.md.
curl -fsSL https://acceptpaper.com/skills/install.sh | sh
5 · Use it
- research-mentor — describe an idea: “Is this novel? What baselines must I beat? What are the landmines? What's my next step?”
- research-explorer — name a field: “Map the frontier and rank the open problems.”
The skills automatically orchestrate search / deep / quote / find_baselines / survey / trends / … and ground every claim in a real source_id.
Auth & limits
Every request carries X-API-Key: YOUR_KEY (or Authorization: Bearer YOUR_KEY). 1000 calls/day. Each call is logged to your dashboard — tool, query, papers returned, latency. Rotate or revoke a key any time.
What's inside
59,631 CV/ML papers (99.6% from 2024–2026), 1.15M paragraph chunks, a citation graph, and 16,864 figures — distilled into a structured schema. 18 MCP tools: search, deep, quote, compare_methods, find_baselines, find_citing, find_lineage, survey, trends, narrative_threads, get_figure, find_experts, graph, analyze, compare, bibtex, status, discover_tools.
Troubleshooting
- 401 / unauthorized — key not active yet (wait ~3 min after signup) or a header typo.
- Tools don't appear — restart your agent after adding the server; run /mcp to confirm litscan-rag + 18 tools.
- Skill doesn't trigger — skills load when your prompt matches their description: ask to critique an idea (mentor) or map a field (explorer).