Feedoracle Compliance

FeedOracle Compliance Intelligence

FeedOracle provides verifiable compliance evidence for AI agents operating in regulated tokenized markets. Every response is ES256K-signed (JWKS-verifiable) and anchored on public networks (Polygon + XRPL).

27 MCP tools across compliance, risk, evidence, KYA (Know Your Agent), and audit verification — monitoring 105+ stablecoins across 18 MiCA articles.

MCP Server URL: https://feedoracle.io/mcp/ Auth: Free tier requires no API key (300 units/day). Optional FEEDORACLE_API_KEY for higher limits. Privacy Policy: https://feedoracle.io/privacy Trust Policy: https://github.com/feedoracle/feedoracle-mcp/blob/main/docs/TRUST_POLICY.md

Authentication

No environment variables are required. The free tier works without any configuration.

When to use this skill

Use this skill only when the user explicitly requests one of the following:

• Stablecoin MiCA compliance status or issuer due diligence
• Verifiable evidence for compliance review or audit workflows
• Stablecoin risk scores, peg monitoring, or reserve backing data

This skill does NOT auto-invoke. It should only be called in response to a direct user request about compliance, MiCA, or stablecoin risk topics.

Data Handling & Privacy

What each tool sends to feedoracle.io

Read-only tools (24 of 27) — send only a token symbol: These tools send a single parameter (e.g. token_symbol: "USDC") and receive signed evidence back. No conversation content, no user data, no PII is transmitted.

Includes: compliance_preflight, mica_status, mica_full_pack, mica_market_overview, peg_deviation, peg_history, significant_issuer, interest_check, document_compliance, reserve_quality, rlusd_integrity, evidence_profile, evidence_leaderboard, evidence_bundle, custody_risk, market_liquidity, macro_risk, ai_explain, ai_provenance, kya_status, audit_query, audit_verify, ping, generate_report

ai_query — sends question text (user-initiated only): This tool sends the user's natural language question to feedoracle.io for routing to the correct evidence API. Only invoke when the user explicitly asks a compliance question. Do not send conversation history, PII, or unrelated text.

kya_register — sends agent metadata (user-initiated only): Registers an agent identity for trust scoring. Only invoke when the user explicitly requests agent registration.

audit_log — sends decision text (user-initiated only): Logs a compliance decision with evidence references. Only invoke when the user explicitly requests decision logging.

Data retention

• Read-only queries: Stateless — no data stored
• KYA profiles: Stored until deletion requested
• Audit trail: Append-only, retained for compliance verification
• Full policy: https://feedoracle.io/privacy
• GDPR: Operated from Germany, Art. 6(1)(b)

MCP Tools (27)

Compliance — 11 tools (read-only, sends token symbol only)

Risk & Evidence — 6 tools (read-only, sends token symbol or protocol name only)

AI Gateway — 3 tools

KYA (Know Your Agent) — 2 tools

Audit Trail — 3 tools

System — 2 tools

Behavior Instructions

1. User-initiated only: Only call FeedOracle tools when the user explicitly asks about compliance, MiCA, stablecoins, or evidence. Never auto-invoke.
2. Verify before claiming: Do not assert compliance status without calling mica_status first. Present data and let the user decide.
3. Write tools require explicit consent: Only call audit_log, kya_register, or generate_report when the user explicitly requests these. Never auto-invoke write tools.
4. Minimal data in ai_query: Send only the specific compliance question — never include conversation history, PII, or unrelated context.
5. Cite evidence: Reference the ES256K signature, pack_id, and JWKS URL from responses.
6. Be precise: Use "verifiable evidence" and "signed compliance data" — not absolute compliance claims.

Controlling Invocation Scope

This skill is designed to be invoked only on explicit user request. If your agent framework supports trigger configuration:

• Restrict triggers to explicit compliance/MiCA/stablecoin keywords only
• Disable auto-invocation if your use case does not require automatic compliance checks
• Sandbox first — test with non-sensitive queries before production use

The skill contains no code, no installation payload, and no persistent background processes. All external communication is to feedoracle.io over HTTPS only.

Connection

# Claude Code
claude mcp add --transport http feedoracle https://feedoracle.io/mcp/

# Claude Desktop (claude_desktop_config.json)
{
  "mcpServers": {
    "feedoracle": {
      "url": "https://feedoracle.io/mcp/"
    }
  }
}

Error Handling

• 401: Invalid API key — use free tier without auth, or verify your key
• 404: Symbol not tracked — check supported assets at feedoracle.io
• 429: Rate limit — wait 60 seconds, retry once
• Trust level insufficient — register via kya_register (only if user requests)

API keys: feedoracle.io/pricing | Docs: github.com/feedoracle/feedoracle-mcp