Problem

openclaw doctor warns about duplicate plugin id feishu even though only the built-in extension exists, printing the same path twice.

Fixes #16208

Root Cause

In manifest-registry.ts, the registry tracked seen plugin IDs with a Set<string>. When two discovery candidates had different source path strings but pointed to the same physical directory (via symlinks or differing path representations), the registry falsely warned about a duplicate.

Solution

• Changed seenIds from Set<string> to Map<string, PluginCandidate> to track the first-seen candidate per manifest ID
• Before emitting a duplicate warning, compare fs.realpathSync() of both candidates' rootDir
• If they resolve to the same physical directory, silently skip (false positive)
• If they differ, emit the duplicate warning as before

Tests

Added manifest-registry.test.ts with 3 test cases:

1. Genuine duplicates — distinct directories with same manifest id → warning emitted ✅
2. Symlink duplicates — same directory via symlink → warning suppressed ✅
3. Identical path duplicates — same rootDir string → warning suppressed ✅

All 14 plugin test files pass (83 tests total).

Greptile Overview

Greptile Summary

This PR fixes a false-positive "duplicate plugin id" warning in openclaw doctor when the same plugin directory is discovered via both its real path and a symlink (or differing path representation). The fix changes seenIds from a Set<string> to a Map<string, PluginCandidate>, then uses fs.realpathSync() to compare physical directories before emitting a duplicate warning.

• The core logic in manifest-registry.ts is correct: symlink/same-path duplicates are silently skipped, genuine duplicates still emit warnings, and realpathSync failures safely fall through to the warning path.
• Three new test cases cover genuine duplicates, symlink duplicates, and identical-path duplicates — all using real filesystem operations with proper temp directory cleanup.
• Minor style note: the symlink test silently passes (via bare return) when symlinks aren't available, rather than logging a skip reason.

Confidence Score: 5/5

• This PR is safe to merge — it's a targeted fix with correct logic, proper error handling, and good test coverage.
• The change is minimal and well-scoped: Set → Map plus a realpathSync comparison guard. The error handling is sound (catch falls through to warning). The fix preserves the original behavior for genuine duplicates while correctly suppressing false positives. Three new tests cover the key scenarios.
• No files require special attention.

_{Last reviewed commit: 63f1d5e}