C

Ghost Protocol

mcp agent Online Last scanned: Mar 16, 2026

# MCP Directory Submissions This page tracks Ghost Protocol MCP listing readiness for agent/tool directories. ## Submission packet (canonical) Use these URLs in every directory submission: - MCP endpoint: `https://ghostprotocol.cc/api/mcp/read-only` - MCP metadata: `https://ghostprotocol.cc/api/mcp/read-only` (`GET`) - MCP manifest: `https://ghostprotocol.cc/.well-known/mcp.json` - OpenAPI: `https://ghostprotocol.cc/openapi.json` - LLM index: `https://ghostprotocol.cc/llms.txt` - AI plugin manifest: `https://ghostprotocol.cc/.well-known/ai-plugin.json` - Pricing endpoint: `https://ghostprotocol.cc/api/pricing` - Terms: `https://ghostprotocol.cc/terms` - Privacy: `https://ghostprotocol.cc/privacy` Tool list: - `list_agents` - `get_agent_details` - `get_payment_requirements` Scope statement: - "Read-only MCP server for agent discovery and pricing requirements. No settlement, ticket issuance, wallet actions, or state mutation." ## Directory targets Status meanings: - `READY`: packet complete and endpoint live. - `SUBMIT`: requires manual external listing submission. - `LISTED`: public listing confirmed. | Directory | Status | Notes | |---|---|---| | Smithery | SUBMIT | Submit endpoint + manifest + tool scope. | | MCP.so | SUBMIT | Submit server profile and canonical endpoint. | | mcpservers.org | SUBMIT | Submit public metadata and capability summary. | | Cursor directory (if/when open) | SUBMIT | Use same packet above. | Copy-ready listing profile: - `docs/developer-portal/mcp-submission-profile.md` ## Verification checklist before each submission 1. `GET /api/mcp/read-only` returns `ok: true` and expected tool list. 2. `POST /api/mcp/read-only` with `tools/list` returns 200 and tool schema. 3. `/.well-known/mcp.json` resolves and matches live endpoint URL. 4. `openapi.json` and `llms.txt` are reachable from production. ## Automated monitoring Workflow: - `.github/workflows/mcp-readonly-monitor.yml` Schedule: - Twice per hour (`:17` and `:47`, UTC) It verifies: 1. `GET /api/mcp/read-only` 2. `GET /.well-known/mcp.json` 3. MCP JSON-RPC `initialize` 4. MCP JSON-RPC `tools/list` 5. MCP JSON-RPC `tools/call` for: - `list_agents` - `get_payment_requirements`

Security Score

65/100

Details

Security Checks (14 criteria)

HTTPS Enabled
Authentication Required
No Admin Exposure
CORS Configured
No Credential Leak
No Server Info Leak
No Open Redirect
No Directory Listing
No Error Info Leak
Endpoint Reachable
Content-Security-Policy
Strict-Transport-Security
X-Frame-Options
X-Content-Type-Options