Three ways to use bitfence.
HTTP API, MCP, or native Rig tools — every path returns the same assessment and pays per query via x402. No accounts. No API keys. No subscriptions.
HTTP API
One GET for a full assessment, one POST for position-aware risk. x402 handles payment on the wire.
# Full assessment (Solana) curl https://api.bitfence.ai/v1/risk/solana/So1111…1112 # Full assessment (Base) curl https://api.bitfence.ai/v1/risk/base/0x8335…2913 # Position-aware risk POST https://api.bitfence.ai/v1/risk/contextual
MCP
Hosted endpoint for Claude Desktop, Claude Code, Cursor, and other MCP-compatible agents. Three tools: analyze_token, contextual_risk, quick_check.
{
"mcpServers": {
"bitfence": {
"type": "streamable-http",
"url": "https://api.bitfence.ai/mcp"
}
}
}
Rig
Native rig::Tool implementations for Rust agents — the same scoring engine, called in-process.
use bitfence::rig::{
AnalyzeTokenTool,
ContextualRiskTool,
QuickCheckTool,
};
let agent = client
.agent("claude-sonnet-4-5")
.tool(AnalyzeTokenTool { state })
.build();
Give your agent the rules.
bitfence publishes its mandatory risk-check instructions as a skill file agents can ingest directly. One command installs it for Claude Code, Copilot, Cline, and any agent that supports the open SKILL.md format:
# Install the bitfence skill
npx skills add bitfenceai/bitfence
Or point your agent at the served file:
api.bitfence.ai/skill.mdListed where agents look.
- x402 Bazaar (Coinbase CDP) — discoverable by x402 agents; three resources listed:
/v1/risk,/v1/risk/contextual,/mcp. - OKX Agent Tradekit — published skill, passed OKX security scan.
- ClawHub — OpenClaw skills hub listing.
- Smithery — MCP server listing.
- MCP Registry — published as
io.github.bitfenceai/bitfencein the community registry that PulseMCP, Glama, and other aggregators ingest. - skills.sh — open skills directory, ranked by installs.
Full API reference — endpoints, response schema, error codes, and the x402 payment flow.
api-reference.md on GitHub →