mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-06-11 08:42:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 13
hermes-agent/tests
History
0xyg3n 9f95f72b98 fix(agent): strip api_messages in thinking-signature recovery so the retry actually omits thinking blocks 
The thinking-signature recovery in agent/conversation_loop.py popped
reasoning_details from messages, then continued to retry. That had two
defects.

First, the strip never reached the wire payload. api_messages is built
once at the start of the turn by shallow-copying every entry in messages
(line 919 area). Each api_messages entry has its own reference to the
same reasoning_details list. When build_api_kwargs runs on every retry
iteration of the inner while-loop, it consumes api_messages, not
messages. Popping reasoning_details from messages left api_messages
untouched, so the retry's request still carried the same thinking
blocks Anthropic had just rejected. The classifier latched
thinking_sig_retry_attempted = True after the first attempt, and the
loop terminated with max_retries_exhausted on the same 400.

Second, the pop mutated the canonical message list. messages is the
same list _persist_session writes to state.db and the session
transcript, so a single recovery permanently wiped every signed
thinking block from the stored conversation. Subsequent turns reloaded
the stripped state, hit the same 400 ('invalid signature' or 'cannot
be modified', see #24107), and the agent stopped responding entirely.
Cascading compaction-ended sessions then chained off the corrupted
parent and the affected chat could not produce a response on any
future turn.

Move the strip onto api_messages, which is the API-call-time list
rebuilt into kwargs on every retry. messages is no longer touched, so
disk I/O stays clean and the recovery actually reaches the wire.

Observed against the native Anthropic Messages API on claude-opus-4-7
and claude-opus-4-8 with the interleaved-thinking-2025-05-14 beta on
hermes-agent 0.12.0 and 0.14.0. PR #24107 narrows the trigger; this
change makes the recovery do what it always claimed to do, and
prevents the destructive aftermath.

Tests cover the api_messages strip in isolation: pop on a shallow copy
does not affect the source, the canonical messages list survives the
strip, idempotency on a duplicate firing path, and a no-op when no
reasoning_details exist on the messages.

Related: #24107, #26959, #17861.
2026-06-10 12:39:44 -07:00
..
acp feat(acp): emit session provenance metadata for compression rotation (#41724) 2026-06-07 22:22:21 -07:00
acp_adapter
agent fix(agent): route 'thinking blocks cannot be modified' 400 to recovery 2026-06-10 12:39:44 -07:00
cli fix(cli): prevent duplicate one-shot finalize on interrupted cleanup (#43320) 2026-06-09 22:41:04 -07:00
cron fix(cron): don't strict-scan script-injected output in no-skills jobs (#43223) 2026-06-10 08:27:24 +05:30
docker fix(gateway): auto-start after container restart via planned-stop marker (#42675) (#43236) 2026-06-10 14:01:34 +10:00
e2e chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
fakes
fixtures/plugins/example-dashboard/dashboard feat(dashboard): nous-blue theme, bulk sessions, schedule picker (#37383) 2026-06-02 12:37:40 -04:00
gateway test: port voice-reply suffix assertions, fix change-detector cap test, add AUTHOR_MAP entry 2026-06-10 02:57:39 -07:00
hermes_cli Add dashboard file browser paths 2026-06-10 09:53:12 -07:00
hermes_state
honcho_plugin fix(plugins): thread-safe lazy-singleton helpers; fix honcho TOCTOU (#24759) (#42150) 2026-06-08 09:35:22 -07:00
integration refactor(gateway): migrate Home Assistant adapter to bundled plugin 2026-06-06 11:46:24 -07:00
openviking_plugin fix(openviking): add missing /agent/{agent}/ segment to memory URI — fixes #36969 2026-06-04 17:40:33 -07:00
plugins fix(langfuse): redact base64 data URIs instead of truncating into invalid base64 2026-06-10 10:49:36 +05:30
providers fix(openrouter): route reasoning_effort to verbosity for adaptive Anthropic models (#43436) 2026-06-10 15:03:01 +05:30
run_agent fix(agent): strip api_messages in thinking-signature recovery so the retry actually omits thinking blocks 2026-06-10 12:39:44 -07:00
scripts fix(skills-hub): stop shipping a degenerate index when GitHub taps collapse (#42347) 2026-06-08 15:21:28 -07:00
skills fix(google-workspace): fall back to uv when venv has no pip (#39516) 2026-06-05 13:30:02 +10:00
stress chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
tools test: port voice-reply suffix assertions, fix change-detector cap test, add AUTHOR_MAP entry 2026-06-10 02:57:39 -07:00
tui_gateway fix(desktop): rebind sessions after websocket reconnect (salvage of #41740) (#43004) 2026-06-09 19:01:00 +00:00
website feat(skills): fix browse cap, add source links + copy buttons + category cleanup (#37143) 2026-06-01 19:52:28 -07:00
__init__.py
conftest.py fix: batch of small robustness/correctness fixes from @kyssta-exe 2026-06-01 19:51:03 -07:00
run_interrupt_test.py
test_account_usage.py
test_atomic_replace_symlinks.py
test_base_url_hostname.py
test_batch_runner_checkpoint.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_bitwarden_secrets.py fix(bitwarden): prevent zip-slip path traversal when extracting bws binary (#40569) 2026-06-06 18:33:44 -07:00
test_cli_file_drop.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_cli_manual_compress.py
test_cli_skin_integration.py
test_ctx_halving_fix.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_dashboard_sidecar_close_on_disconnect.py fix(tui-gateway): reap leaked slash_worker sessions on disconnect + active_list liveness (re-scoped onto current main) 2026-06-08 10:02:05 -07:00
test_desktop_mac_entitlements.py test(desktop): assert macOS device entitlements are inherited 2026-06-03 07:32:00 +07:00
test_docker_home_override_scripts.py Repair cron ownership on container restart (#41976) 2026-06-10 15:32:34 +10:00
test_docker_stage2_browser_discovery.py fix(docker): discover Playwright headless_shell browser (#35717) 2026-06-01 16:06:44 +10:00
test_dockerfile_tini_compat_shim.py fix(docker): add /usr/bin/tini compatibility shim for legacy wrappers (#34192) (#34382) 2026-06-01 13:32:55 +10:00
test_empty_model_fallback.py test(models): guard Nous silent default against expensive-flagship escalation 2026-06-05 02:54:34 -07:00
test_env_loader_secret_sources.py fix(secrets): only apply external secrets once per HERMES_HOME per process (#32271) 2026-05-25 15:18:55 -07:00
test_evidence_store.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_gateway_streaming_nested_config.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_get_tool_definitions_cache_isolation.py fix(gateway): close residual memory-leak sites under heavy scheduled workload 2026-06-08 06:32:42 -07:00
test_hermes_bootstrap.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_hermes_constants.py fix(constants): use windows native default hermes home 2026-06-03 19:37:29 -07:00
test_hermes_home_profile_warning.py
test_hermes_logging.py fix(gateway): tolerate Unicode in stderr log handlers on Windows 2026-06-06 19:57:44 -07:00
test_hermes_state.py fix(cron): bound the desktop run-history query to one job (#41088) 2026-06-07 02:41:01 -07:00
test_hermes_state_compression_locks.py fix(compression): prevent session-id fork from concurrent compressions (#34351) 2026-05-28 21:40:39 -07:00
test_hermes_state_wal_fallback.py fix(kanban): skip redundant WAL pragma on already-WAL connections 2026-05-27 14:31:55 -07:00
test_honcho_client_concurrency.py fix(plugins): thread-safe lazy-singleton helpers; fix honcho TOCTOU (#24759) (#42150) 2026-06-08 09:35:22 -07:00
test_honcho_client_config.py fix(honcho): harden self-hosted setup paths 2026-05-29 22:29:48 -07:00
test_honcho_session_context.py fix(honcho): align user context peer perspective 2026-05-27 10:49:33 -07:00
test_honcho_startup_fail_open.py fix: make Honcho startup fail open 2026-06-01 20:13:42 -07:00
test_install_no_initial_commit.py fix(install): move broken checkout aside instead of deleting it 2026-06-08 02:18:21 -07:00
test_install_sh_browser_install.py
test_install_sh_pythonpath_sanitization.py
test_install_sh_root_fhs_uv_python_path.py test(install): harden uv-python-path regression test against future drift 2026-05-27 13:55:51 -07:00
test_install_sh_setup_wizard_tty_probe.py
test_install_sh_symlink_stomp.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_install_sh_termux_network_prereqs.py
test_ipv4_preference.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_lazy_session_regressions.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_lint_config.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_live_system_guard_self_test.py
test_mcp_serve.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_mini_swe_runner.py
test_minimax_model_validation.py
test_minimax_oauth.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_minisweagent_path.py
test_model_forces_max_completion_tokens.py fix(params): send max_completion_tokens for newer OpenAI families on custom endpoints 2026-06-09 23:22:10 -07:00
test_model_picker_scroll.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_model_tools.py feat(middleware): add adaptive execution intercepts 2026-06-03 11:22:06 -07:00
test_model_tools_async_bridge.py fix(web): run URL SSRF checks off the event loop in async paths 2026-06-04 18:04:47 -07:00
test_ollama_num_ctx.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_output_cap_parsing.py test(agent): cover char-based output-cap overflow parsing (#42741) 2026-06-09 03:17:12 -07:00
test_package_json_lazy_deps.py
test_packaging_metadata.py fix(packaging): ship optional-mcps catalog in wheel and sdist (#39859) 2026-06-09 14:03:20 -04:00
test_plugin_skills.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_plugin_utils.py fix(plugins): thread-safe lazy-singleton helpers; fix honcho TOCTOU (#24759) (#42150) 2026-06-08 09:35:22 -07:00
test_process_loop_event_loop_warning.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_project_metadata.py fix(deps): align anthropic extra pin with lazy pin + guard whole pin surface (#42335) 2026-06-08 12:11:54 -07:00
test_retry_utils.py
test_run_tests_parallel.py fix(ci): make parallel runner's exit-4 retry robust for newly-added test files (#42994) 2026-06-09 21:39:09 -07:00
test_sanitize_tool_error.py
test_slash_worker_watchdog.py feat(slash-worker): self-terminate on parent death via create_time watchdog 2026-06-08 07:03:12 -07:00
test_sql_injection.py
test_state_db_malformed_repair.py fix(state.db): recover from malformed sqlite_master so hidden sessions reappear (#43149) 2026-06-09 18:49:08 -05:00
test_subprocess_home_isolation.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_termux_all_extra_compat.py
test_timezone.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_toolset_distributions.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_toolsets.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_trajectory_compressor.py fix(research): keep tool_call/tool_response pairs intact when compressing trajectories 2026-06-07 05:01:27 -07:00
test_trajectory_compressor_async.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_transform_llm_output_hook.py
test_transform_tool_result_hook.py test: stub has_hook in transform_tool_result hook tests 2026-06-03 06:36:46 -07:00
test_tui_gateway_server.py fix(desktop): keep model runtime state per session (#43702) 2026-06-10 18:16:50 +00:00
test_tui_gateway_ws.py fix: resolve rebase conflict in _teardown_session worker cleanup 2026-06-08 10:02:05 -07:00
test_utils_truthy_values.py
test_web_server.py fix(tui-gateway): reap leaked slash_worker sessions on disconnect + active_list liveness (re-scoped onto current main) 2026-06-08 10:02:05 -07:00
test_wheel_locales_e2e.py fix(packaging): ship locales/ i18n catalogs in wheel, sdist, and Nix (#38383) 2026-06-03 12:00:27 -07:00
test_yuanbao_integration.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_yuanbao_markdown.py
test_yuanbao_pipeline.py test(yuanbao): add missing patch import to pipeline tests 2026-06-09 03:17:00 -07:00
test_yuanbao_proto.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
test_yuanbao_shutdown.py fix(yuanbao): bound ws.close() so an idle server can't stall shutdown ~5s (#40607) 2026-06-07 17:49:38 -07:00