[dhairya (dhairyajangir)]

Fixes #339843

Non-ASCII MCP resource-template values currently produce percent-encoded UTF-16 code units: café becomes caf%E9, and an emoji becomes two surrogate escapes. Encode valid Unicode strings as UTF-8 while retaining the existing ASCII punctuation escaping. Count prefix lengths by Unicode code point so a supplementary character remains intact. This follows RFC 6570 §3.2.1.

The two added tests cover accented/CJK/supplementary characters, punctuation, and supplementary-character prefixes; all 26 existing tests, including control-character zero-padding, remain unchanged.

Validation:

• Actual-source targeted transpilation followed by node test/unit/node/index.js --run src/vs/base/test/common/uriTemplate.test.ts: 29 passing (28 URI-template tests plus the runner's error check); the two additions fail before the fix. Node 24.19.0 on Linux.
• Repository ESLint on both changed files and git diff --check: pass.
• Independent actual-source checks: 192 Unicode cases pass; 3,120 ASCII scalar and 32 string-composite controls remain unchanged.

No Electron/UI or live MCP-server test was run. Malformed lone-surrogate inputs now throw URIError, as reserved expansion already does; the encoder also string-coerces non-string array elements. Existing reserved-percent and composite-operator behavior is otherwise outside this fix.

AI assistance was used for implementation, testing, and an independent review.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[dhairya (dhairyajangir)]

Could a maintainer authorize the fork workflows so the repository checks can run? The 29 targeted tests and changed-file ESLint pass locally; the CLA and dependency checks have passed. The reproduction and validation limits are in the PR description.

[Connor Peet (connor4312)]

thanks!