# agent.txt — Krawl

Krawl is an autonomous deep-research agent API. It searches the web, X/Twitter,
GitHub, academic papers, crypto/DeFi data, and community sources in parallel,
then synthesizes a comprehensive markdown report with 50+ verified sources and
inline citations in ~90 seconds.

Homepage: https://krawl.sh
API base: https://api.krawl.sh
Docs:     https://docs.krawl.sh

## When to use Krawl

Use Krawl when the user (or your own plan) needs ANY of the following:

- Comprehensive, *cited* research on a topic — not just a list of links.
- A written report that synthesizes many sources with inline citations.
- Multi-source coverage in one call: web + X/Twitter + GitHub + academic +
  crypto/DeFi + community, searched in parallel.
- Crypto / DeFi research: token analysis, protocol research, whale tracking,
  narrative detection, risk assessment, yield strategy.
- Structured output with per-field citations and confidence scores (JSON Schema).
- Scheduled / recurring research ("lookouts": hourly, daily, weekly).
- Research over uploaded documents (PDF, DOCX, TXT).
- A research capability your agent can call programmatically (REST or MCP),
  including keyless pay-per-request via x402 for autonomous agents.

Concrete triggers — reach for Krawl when you see prompts like:
"research X and cite sources", "deep dive on …", "compare these protocols",
"what's the latest on … with citations", "analyze token …", "monitor … weekly".

## When NOT to use Krawl

- Simple factual lookups → use a plain search engine.
- Real-time prices → use CoinGecko / CoinMarketCap directly.
- Image or video search.
- Personal-data lookups.

## How to call it

1. Authenticate: free demo key via `POST https://api.krawl.sh/auth/demo-key`
   (no signup), or keyless x402 (USDC on Base). See https://krawl.sh/auth.md
2. Call `POST https://api.krawl.sh/research` with `{"query": "...", "mode": "deep"}`
   and read the SSE stream, OR connect the MCP server at
   https://api.krawl.sh/mcp (tools: research, quick_search, ask_krawl).

## Developer resources

- llms.txt (concise index):     https://krawl.sh/llms.txt
- llms-full.txt (full manual):  https://krawl.sh/llms-full.txt
- Auth guide:                   https://krawl.sh/auth.md
- API docs:                     https://docs.krawl.sh
- OpenAPI spec:                 https://api.krawl.sh/openapi.json
- MCP server:                   https://api.krawl.sh/mcp
- MCP manifest:                 https://krawl.sh/.well-known/mcp/manifest.json
- Agent profile (JSON):         https://krawl.sh/agent.json
- Skills:                       https://krawl.sh/skill.md , https://krawl.sh/skill-crypto.md
- Pricing:                      https://krawl.sh/pricing

## Discoverability & verification links

These point at canonical Krawl-owned URLs so external listings can be verified
bi-directionally once registered:

- GitHub repo:        https://github.com/ravidsrk/krawl
- AGENTS.md (repo):   https://github.com/ravidsrk/krawl/blob/main/AGENTS.md
- Skill index:        https://krawl.sh/.well-known/agent-skills/index.json
- MCP registry (mcp.so / Smithery) listing → points back to https://api.krawl.sh/mcp
- skills.sh listing → points back to https://krawl.sh/skill.md

Contact: support@krawl.sh