stage-362: 8-PR follow-up batch — Ollama routing + legacy toolset + cancel copy + cleanup + custom provider mismatch + cron metadata + dead-code removal (with #2323 revert after Opus-caught silent regression) - #2335
Merged
Conversation
Replace the hardcoded Skyly cancellation wording with the configured bot_name from settings, falling back to Hermes when unset. Keep the client-side fallback in sync by using window._botName if the session refresh after cancellation fails. Co-authored-by: Obryn 🐉 <obryn-ai@dotbeeps.dev>
fix: expand legacy Hermes CLI toolset alias (Michaelyklam, closes #2232)
fix: route endpoint-discovered Ollama models correctly (Michaelyklam)
fix: skip provider mismatch for named custom providers (Michaelyklam)
fix: show readable live activity progress (Michaelyklam) # Conflicts: # CHANGELOG.md
fix: use assistant name in cancel copy (dotBeeps)
feat: show cron output usage metadata (Michaelyklam)
fix: clean session attachment and stream recovery leftovers (Michaelyklam, closes #2325)
fix: isolate background worker profile env (Michaelyklam, closes #2321) # Conflicts: # CHANGELOG.md
…ed for 9-PR stage-362 batch with proper attribution
SysAdminDoc
pushed a commit
to SysAdminDoc/hermes-webui
that referenced
this pull request
Jun 26, 2026
stage-362: 8-PR follow-up batch — Ollama routing + legacy toolset + cancel copy + cleanup + custom provider mismatch + cron metadata + dead-code removal (with nesquena#2323 revert after Opus-caught silent regression)
bernyforce
pushed a commit
to bernyforce/hermes-webui
that referenced
this pull request
Jul 29, 2026
stage-362: 8-PR follow-up batch — Ollama routing + legacy toolset + cancel copy + cleanup + custom provider mismatch + cron metadata + dead-code removal (with nesquena#2323 revert after Opus-caught silent regression)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
stage-362 — 8-PR follow-up batch (v0.51.69)
Tenth full PR sweep of the day. Eight PRs in stage-362 — #2323 reverted after Opus caught a silent regression (separately filed as #2334 for follow-up).
+432 / -78 across ~22 files (after #2323 revert).
Composition
Added (1 PR)
Fixed (7 PRs)
hermesCLI toolset alias expansion tohermes-cli+hermes-api-servercustom:zenmux, etc.)_save_pre_compression_snapshothelper; retargeted test to exercise production helperReverted from this stage (Opus-caught silent regression)
hermes_cli.config.load_config()readsHERMES_HOMEfromos.environdirectly with no thread-local consumer anywhere in hermes-agent. Tests passed only because they mocked the production reader. Without the revert, this would have re-introduced the exact bug Fix profile-scoped auxiliary routing for background workers #2299 fixed in v0.51.67 (background workers reading default profile config). Reverted; follow-up filed as fix: lock turn journal appends #2334.Deferred to future sweeps:
Stage-362 maintainer fixes (THE CRITICAL ONE)
#2323 silent profile-routing regression
This was the most important catch of the day. The PR appeared to implement my own #2321 follow-up correctly (thread-local instead of
os.environfor background workers — Opus's recommended long-term fix). All targeted tests passed. CI green. The architectural shape made sense.But Opus did empirical verification by actually running the code path:
Chain of evidence Opus traced:
agent/auxiliary_client.py:3879 _get_auxiliary_task_config()→hermes_cli/config.py:4151 load_config()→get_config_path()→get_hermes_home()(hermes_constants.py:13) → readsos.environ["HERMES_HOME"]only, no thread-local fallback.api/config.pythread-local helpers exist, but background-worker threads don't have_tls.profileset (the streaming thread is separate from worker threads).os.environ["HERMES_HOME"] = str(profile_home_path)was the only line that made Fix profile-scoped auxiliary routing for background workers #2299 (stage-360) work.The three "verifying" tests (
test_title_aux_routing.py,test_update_banner_fixes.py,test_sprint46.py) all mock the production reader and don't exercisehermes_cli.config.load_config(). The tests verify the thread-local WAS SET — they don't verify any production code reads it.Production impact without revert: every non-default-profile session triggering background title generation, manual compression, or update-summary would silently use the default profile's auxiliary config — exact bug #2299 fixed.
Reverted with commit
2161b81d. Filed as #2334 with three options for re-attempt:os.environmutation under lock (compromise)hermes_cli.configto consult thread-local first (upstream change)Specific review (other 8 PRs)
#2322 — Endpoint-discovered Ollama models
Recognizes
custom:<host>-<port>:<model>picker values as UI routing hints when the configured local-server base URL matches; resolves via actualollamaprovider instead of looking forCUSTOM_*_API_KEY.#2330 —
startsWith('custom:')edgesOpus verified:
'custom:'(trailing colon, no name) → matches → skip check. Fails soft on malformed input, matches intent.'custom:openrouter'(theoretical collision) → matches → skip check. Correct, since named custom is always aggregator-shaped.#2327 — Bot name fallback
Opus verified
(window._botName||'Hermes').trim()||'Hermes'double-fallback handles empty/whitespacebot_name. Bothstatic/messages.jsandapi/streaming.pymirror the same logic. No regression vs. prior hardcoded "Skyly".#2328 — Cleanup hardening
shutil.rmtree(_session_attachment_dir(sid), ignore_errors=True)handles non-existent dirs cleanly(S.session&&S.session.session_id)!==activeSid—activeSidcaptured in closure,S.sessionread at firing time — correct scope#2333 — Dead helper removal
_save_pre_compression_snapshot()reachable only from a test (now retargeted to the production_preserve_pre_compression_snapshot()helper).Opus surfaced one SHOULD-FIX (non-blocking): branch 2 of
_preserve_pre_compression_snapshot()(the load-and-mark-only path, lines 1785-1804 — added in stage-359 as my SHOULD-FIX on Opus's previous review) has runtime-clear semantics that aren't covered by the retargeted test. Both pre-existing tests exercise branch 1 only. Worth adding branch 2 coverage in a follow-up.Verification
run-browser-tests.sh: 20/20 QA + 11/11 API checks PASSED in 103sOpus Advisor Review
After the revert applied, all 8 remaining PRs are SHIP-cleared.
_preserve_pre_compression_snapshot)Closes / refs
Closes #2232(via fix: expand legacy Hermes CLI toolset alias #2326 — legacy CLI toolset alias)Closes #2312 item #1(via test: retarget compression snapshot runtime regression #2333 — dead helper removed)Closes #2325(via fix: clean session attachment and stream recovery leftovers #2328 — both follow-ups landed)Closes #2329(via fix: skip provider mismatch for named custom providers #2330 — custom provider mismatch)Stats