Skip to main content
Glama

Kavach KYA (io.github.srigenmedia/kavach-kya)

mcptry

Sovereign Agentic Spend Firewall & Streamable HTTP MCP Guard (https://kavach.srigenmedia.com)

Kavach KYA is a zero-custody (₹0.00 customer funds held) horizontal B2B Know-Your-Agent (KYA) spend firewall for autonomous AI agents, 402-in-a-Box gateways, and Model Context Protocol (MCP) tool servers.


Core Capabilities

  • Streamable HTTP /mcp Endpoint (https://kavach.srigenmedia.com/mcp):

    • kavach_check_spend_policy — Evaluates canonical 5-tuples [asset_or_currency, kya_agent_id, merchant_vpa_or_recipient, amount_minor_or_units, eligibility_flag] against velocity limits, merchant allowlists, and prompt-injection guards.

    • kavach_get_agent_status — Inspects live eligibility_flag (0/1), daily velocity spend, and pending 2FA holds.

    • kavach_trigger_kill_switch — Atomically flips eligibility_flag = 0 in < 50ms via Cloudflare SQLite Durable Objects (KyaSpendCoordinatorDO) and propagates revocation to 402-in-a-Box (PUT /api/admin/kya/agents/:id/eligibility).

  • Hibernatable 15-Minute 2FA Holds & 1-Tap Webhooks: Velocity/amount breaches enter AWAITING_HUMAN_2FA with 15-minute Durable Object alarm() expiry and constant-time HMAC-SHA256 verified Slack (X-Slack-Signature) and WhatsApp Business (X-Hub-Signature-256) 1-tap callbacks.

  • Tamper-Evident Audit Hash Chain: Every decision appends a SHA-256 hash-chained receipt (prev_receipt_hash -> receipt_hash) with verified bank/gateway UTR/RRN proofs and zero raw PII.

  • Zero-Custody & Indian Regulatory Safe-Harbor: Holds ₹0.00 customer funds; B2B SaaS plans are billed under SAC 998315 (18% GST INR) or 0% Export LUT (USD).


Related MCP server: Safe4

Quick MCP Client Configuration

{
  "mcpServers": {
    "kavach-kya": {
      "type": "streamable-http",
      "url": "https://kavach.srigenmedia.com/mcp",
      "headers": {
        "Authorization": "Bearer ${KAVACH_OPERATOR_JWT}"
      }
    }
  }
}

See docs/402-in-a-box-connector.md for the complete 5-minute HTTP & webhook integration guide and SECURITY.md for our fail-closed security architecture.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      B
      maintenance
      A governance proxy for AI tools — every MCP/agent tool call is policy-gated, secret-redacted, and written to a hash-chained, offline-verifiable audit trail.
      13
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI agents to verify proposed payments against their assigned task, budgets, permitted categories, and counterparties, returning an ALLOW or DENY decision with a tamper-evident audit trail.
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to inspect, check, and audit signed offline spending authorizations, and to issue hash-chained receipts for each action taken against them. Enforces per-transaction, total, expiry, merchant, and human-approval constraints, returning allow, needs_human, or deny with the specific failing constraint.
      Apache 2.0
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables autonomous AI agents to invoke metered upstream tool APIs through a policy firewall that enforces per-key cent-level spending caps and strict tool allowlists, blocking unauthorized or over-budget calls pre-flight. Every transaction is settled atomically and returned with an HMAC-SHA256 signed audit receipt, with instant key rotation and spread accounting built in.
      MIT