fix: isolate OAuth cache by resource and headers (#25)

[clouatre]

Fixes #25

Problem

Multiple instances connecting to the same server URL with different --resource flags share OAuth sessions and connect to the wrong tenant.

Solution

Include authorizeResource and custom headers in the serverUrlHash calculation to isolate OAuth sessions per unique configuration.

Changes

• Modified getServerUrlHash() to include resource and headers in hash calculation
• Calculate hash once in parseCommandLineArgs() with all parameters
• Pass hash through stack to NodeOAuthClientProvider
• Each unique config gets isolated lockfile and token storage

Testing

• ✅ Added 6 unit tests for hash generation (all passing, 50/50 total)
• ✅ Tested with real multi-tenant Atlassian setup
• ✅ Both instances connect to correct tenants
• ✅ Separate OAuth sessions maintained
• ✅ Backward compatible (one-time re-auth for existing users)

Example (multiple Atlassian tenants)

{
  "mcpServers": {
    "atlassian_tenant1": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.atlassian.com/v1/sse", "--resource", "https://tenant1.atlassian.net/"]
    },
    "atlassian_tenant2": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.atlassian.com/v1/sse", "--resource", "https://tenant2.atlassian.net/"]
    }
  }
}

[pkg-pr-new]

Open in StackBlitz

npx https://pkg.pr.new/mcp-remote@175

commit: 2e769c2