fix(agents): any spelling of a claude uuid is checked as claude's lowercase id (Issue #31, review)

_is_uuid took an uppercase or mixed-case uuid, and the paths built from it
are the live lowercase session's own files on macOS's default
case-insensitive APFS. The in-use checks compare exactly, though: session
files and command lines carry the lowercase id, so `purge claude <UPPER>`
found no holder and could remove a live session's files. _canonical now
lowercases a canonical uuid, and the paths, the session-file and job
matches, the in-use check and the residual all use it; a command line is
searched for both spellings.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

.agents/skills/lib_py/agents/adapters/claude.py

@@ -7,11 +7,15 @@ from lib_py.verify_session import workspace_key
 _UUID = re.compile(r"[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}", re.IGNORECASE)
 
 
-def _is_uuid(uuid) -> bool:
-    """claude names every session by a canonical uuid. Any other name in a
-    session's place is not one: projects/<key>/memory/ (auto memory) sits
-    beside the session dirs."""
-    return isinstance(uuid, str) and bool(_UUID.fullmatch(uuid))
+def _canonical(uuid) -> Optional[str]:
+    """The id claude itself writes, in lowercase, for any spelling of a
+    canonical uuid; None for anything else. Any other name in a session's
+    place is not one: projects/<key>/memory/ (auto memory) sits beside the
+    session dirs. Another spelling must go through the same checks as
+    claude's own: on macOS's default case-insensitive APFS it names the
+    same files, while the ids in session files and on command lines are
+    compared exactly."""
+    return uuid.lower() if isinstance(uuid, str) and _UUID.fullmatch(uuid) else None
 
 
 def _config_dir(ctx: DiscoveryContext) -> str:
@@ -306,14 +310,16 @@ class ClaudeAgentAdapter(BaseAgentAdapter):
         # pre-#31 reach: the `<id>.jsonl` file alone.
         if not _plausible_id(uuid):
             return []
+        canon = _canonical(uuid)
         cfg = _config_dir(ctx)
         found = []
         for pdir in _projects_dirs(ctx):
-            base = os.path.join(pdir, ctx.ws_key, uuid)
-            found += [(pdir, p) for p in ((base + ".jsonl", base) if _is_uuid(uuid) else (base + ".jsonl",))
+            base = os.path.join(pdir, ctx.ws_key, canon or uuid)
+            found += [(pdir, p) for p in ((base + ".jsonl", base) if canon else (base + ".jsonl",))
                       if os.path.lexists(p)]
-        if not _is_uuid(uuid):
+        if not canon:
             return found
+        uuid = canon
         found += [(cfg, p) for p in (os.path.join(cfg, "file-history", uuid),
                                      os.path.join(cfg, "session-env", uuid)) if os.path.lexists(p)]
         job = self._owned_job(uuid, ctx)
@@ -344,15 +350,17 @@ class ClaudeAgentAdapter(BaseAgentAdapter):
     def conversation_in_use(self, uuid: str, ctx: DiscoveryContext) -> bool:
         # No claude process holds a name that is not a uuid; searching
         # command lines for one ("a/b", "memory") only finds strangers.
-        if not _is_uuid(uuid):
+        canon = _canonical(uuid)
+        if not canon:
             return False
-        if conversation_in_use(uuid, [self.artifact_path(uuid, ctx)]):
+        # claude's own spelling and the one asked about, on a command line.
+        if any(conversation_in_use(u, [self.artifact_path(canon, ctx)]) for u in {canon, uuid}):
             return True
-        return any(_session_file_live(data, pid) for _, pid, data in self._session_files(uuid, ctx))
+        return any(_session_file_live(data, pid) for _, pid, data in self._session_files(canon, ctx))
 
     def in_use_hints(self, uuid: str, ctx: DiscoveryContext) -> list:
         return [f"a claude process whose session file under {os.path.join(_config_dir(ctx), 'sessions')}/ names it"] \
-            + super().in_use_hints(uuid, ctx)
+            + super().in_use_hints(_canonical(uuid) or uuid, ctx)
 
     def purge_artifacts(self, uuid: str, ctx: DiscoveryContext) -> list:
         # Best effort: never raises (stop runs this inside its registry
@@ -360,6 +368,7 @@ class ClaudeAgentAdapter(BaseAgentAdapter):
         # shows up in residual_artifacts instead.
         # Ownership evidence (a job's state.json, a session file) is removed
         # only after what it vouches for, so a retry can still find the rest.
+        uuid = _canonical(uuid) or uuid
         purged = []
         job = self._owned_job(uuid, ctx)
         for path in self._owned_paths(uuid, ctx):
@@ -379,8 +388,9 @@ class ClaudeAgentAdapter(BaseAgentAdapter):
         roots (two levels), plus the content-matched jobs dir and session
         files with their (masked) key files. projects is not walked deeper;
         it can be very large."""
-        if not _is_uuid(uuid):
+        if not _canonical(uuid):
             return sorted(p for _, p in self._candidates(uuid, ctx))
+        uuid = _canonical(uuid)
         cfg = _config_dir(ctx)
         pdirs = _projects_dirs(ctx)
         found = set(scan_names(cfg, uuid, skip=pdirs))