Cover byte identity without neighbors, ECC's decision unchanged in every mode, the anchored and clear-leader listing rule, Claude user, plugin and project discovery, the CLI snapshot round trip, the launch and hook surfaces, and a fixture on the shipped registry. Require the new module in the npm pack surface.
The pattern only accepted settings.json and settings.<suffix>.json, so a plain settings-dev.json stayed editable on the subagent paths that only gate sensitive or hard-linked targets. Use settings[^/]*.json and cover the dash-suffixed variants plus a non-settings control.
GitHub Windows runners use an 8.3 short path for os.tmpdir()
(C:\Users\RUNNER~1\...). git rev-parse --show-toplevel returns the long
path, but JS fs.realpathSync does not expand 8.3 short names, so
isGitWorkTreeRoot compared unequal for a valid worktree and source
preparation fell through to the registry branch and threw.
Prefer fs.realpathSync.native (falling back to realpathSync for injected
fakes) in isGitWorkTreeRoot and findRegistrySource. Fail-closed behavior
is unchanged. Adds cross-platform and Windows-only regression tests.
Fixes the windows npm/full lane on #3882.
CLASSIFIER_NOTICE matched by prefix, so a pasted notice followed by a
user question was dropped from both summary paths. Anchor it to the
whole message so trailing user text is kept, and add a regression test.
Also pass ECC_SKIP_LLM_SUMMARY=1 in the withTranscript fixture helper so
an inherited ECC_LLM_SUMMARY_INTERVAL=1 cannot trigger claude -p.
Refs #3867
Assisted-by: Claude <noreply@anthropic.com>
Bash ends a decoded $'...' string at its first NUL, so `cat <<$'EOF\0X'`
ends at the line EOF, but the splitter looked for EOF\0X and swallowed
the commands after the body. A carriage return or a no-break space is
not a blank to Bash, so `echo foo\r# && npm run dev` runs the dev
server, but the splitter read the # as a comment. And spreading a long
substitution's segments into push() threw RangeError, which the hook's
catch turned into an allow. Truncate the decoded part at the NUL, use
Bash's blanks for word starts, and append the segments one at a time.
The safety-classifier notice filter only covered the mechanical summary.
When generateSessionSummary runs (low context or message interval),
extractConversationText read the raw transcript and passed the notice to
the LLM as a user turn, so it could still reach saved context.
Move the check into one shared isClassifierNotice() in
scripts/lib/llm-summary.js and use it in both userMessageText() and
extractConversationText(). Add LLM-path regression tests, and split the
session-end classifier-notice tests into small functions (<50 lines).
Refs #3867
Assisted-by: Claude <noreply@anthropic.com>