hermes-agent/hermes_cli
ethernet c363db81e0
fix(desktop, ink): don't wipe messages before final message (#65919)
* fix(desktop): preserve interim assistant text wiped at message.complete

When the agent emits interim text (commentary alongside tool calls, or the
attempted final answer before a verify-on-stop nudge), all UI surfaces
streamed it live but then wiped it at message.complete — keeping only the
final response. The user saw text appear during inference, then disappear.

This is the complete fix across all three layers: agent core, gateway
transport, and all UI surfaces (desktop + Ink TUI).

The verify-on-stop and pre_verify paths flagged the assistant's attempted
final answer as _verification_stop_synthetic, suppressing it from both
state.db and the UI. The user only saw the terse post-verification reply.

Now the assistant response is real content: it's persisted to state.db and
emitted as an interim message via _emit_interim_assistant_message(force_display=True)
before the verification loop runs. Only the synthetic nudge messages keep
the synthetic flags. The turn finalizer drops nudges from live history and
compares content (not just role) to avoid duplicating a published candidate.
Message sequence repair collapses verification candidates in the
consecutive-assistant merge.

Wire agent.interim_assistant_callback both at construction (_agent_cbs())
and per-turn (defense-in-depth), emitting a new message.interim event with
{text, already_streamed}. Gated on display.interim_assistant_messages
(default true). Cleared in the finally block so a stale closure can't
fire on a later turn.

Add message.interim to the GatewayEventName union (apps/shared) and a
typed payload to the TUI's GatewayEvent discriminated union.

The TUI already had the segment-anchoring machinery (flushStreamingSegment +
finalTail) but had no handler for message.interim. Added recordInterimMessage
+ interimBoundaryIndex to seal segments mid-turn, and updated
recordMessageComplete to only dedupe segments after the interim boundary.

Replaced the fragile sealed-set approach with a proper interimBoundaryPending
state flag on ClientSessionState. finalizeInterimAssistantMessage finalizes
the streaming bubble in place (or creates a standalone one), rotates the
stream ID so next deltas create a new bubble, and sets the flag. When the
final text equals an already-sealed interim, they stay as distinct messages.

Extracted mergeFinalAssistantText() as a pure function in chat-messages.ts,
used by both completeAssistantMessage and finalizeInterimAssistantMessage.
Split the bidirectional dedup predicate: reasoning is a restatement only when
the final FULLY covers it. A short final ("Done.") no longer swallows a
longer reasoning block that merely starts with it.

Honor display.interim_assistant_messages (default true) across all layers:
the tui_gateway gates the callback, the desktop wires it to a nanostores
atom via use-hermes-config. Updated hermes_cli/config.py and
cli-config.yaml.example comments to document the Desktop behavior.

_split_segment_tokens now accepts posix=False and _find_ad_hoc_match tries
both posix modes so ad-hoc verification scripts with Windows backslash
paths are matched correctly. (response_previewed forwarding from #53553
is not included — our emit-interim + persist approach makes it unnecessary
since the attempted answer is now surfaced before the verification loop.)

- tsc: clean (desktop + TUI + shared)
- vitest desktop: 73/73 pass (7 interim-sealing + 5 mergeFinalAssistantText + 4 config atom)
- vitest TUI: 83/83 pass (4 new message.interim tests)
- python: 390 tests pass (340 tui_gateway + 33 verification/finalizer + 6 config gating + 3 evidence + 8 continuation budget)

Co-authored-by: Liam Zhang <yingliang-zhang@users.noreply.github.com>
Co-authored-by: Lucas D'Alessandro <lucasfdale@users.noreply.github.com>
Co-authored-by: Eric Manganaro <superposition@users.noreply.github.com>
Co-authored-by: sweetcornna <sweetcornna@users.noreply.github.com>
Co-authored-by: DECK6 <DECK6@users.noreply.github.com>
Co-authored-by: matantsevs <matantsevs@users.noreply.github.com>
Co-authored-by: gitcommit90 <gitcommit90@users.noreply.github.com>

* fix: prefix-match interim streamed content to avoid benign duplicate bubbles

_interim_content_was_streamed used exact equality (streamed == visible_content),
so a final response that was the streamed text plus a trailing delta — or a
partial stream before the verify nudge fired — failed the match and left
_response_was_previewed false. The turn then showed two bubbles (interim +
identical final) instead of settling the interim in place.

Relax to a prefix check (visible_content.startswith(streamed)) in both the
core match and the desktop's settle-in-place gate. The TUI already used
prefix matching via finalTail. The reverse direction (streamed longer than
final) is intentionally not matched — that could suppress a needed resend
in the gateway path where already_streamed=True calls on_segment_break().

* test(desktop): add partial-stream-then-nudge dedup edge case

Third edge case for the interim-sealing dedup: model streams part of its
answer via message.delta, verify nudge fires, interim seals the streamed
prefix, then the final response is the same text plus a trailing delta.
Asserts one bubble (not two) containing the full final text.

Acceptance protocol #2 — covers all three dedup edges:
  1. interim == final (existing)
  2. interim = strict prefix of final (existing)
  3. partial-stream-then-nudge (this commit)

---------

Co-authored-by: Liam Zhang <yingliang-zhang@users.noreply.github.com>
Co-authored-by: Lucas D'Alessandro <lucasfdale@users.noreply.github.com>
Co-authored-by: Eric Manganaro <superposition@users.noreply.github.com>
Co-authored-by: sweetcornna <sweetcornna@users.noreply.github.com>
Co-authored-by: DECK6 <DECK6@users.noreply.github.com>
Co-authored-by: matantsevs <matantsevs@users.noreply.github.com>
Co-authored-by: gitcommit90 <gitcommit90@users.noreply.github.com>
2026-07-20 11:42:29 -04:00
..
dashboard_auth fix(mcp): complete OAuth through hosted dashboards 2026-07-17 04:50:47 -07:00
proxy security(gateway): set explicit client_max_size on 3 uncapped aiohttp servers (#59180) 2026-07-05 14:48:28 -07:00
subcommands fix(config): warn-after-write for unrecognized keys instead of refusing 2026-07-20 05:38:59 -07:00
__init__.py chore: release v0.18.2 (2026.7.7.2) (#60651) 2026-07-07 20:11:08 -07:00
_parser.py feat(cli): restore cwd on resume (--no-restore-cwd) 2026-07-12 05:08:13 -04:00
_subprocess_compat.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
active_sessions.py fix(tui): preserve live session identity across compression (#49041) 2026-06-24 00:54:18 +05:30
auth.py fix(auth): detect configured providers absent from registry (#66017) 2026-07-16 22:56:45 -04:00
auth_commands.py fix(auth): make xAI OAuth pools multi-account resilient 2026-07-17 11:37:25 +05:30
azure_detect.py fix(security): secure Azure catalog probes 2026-07-11 12:28:55 +05:30
backup.py fix(discord): harden reconnect message recovery 2026-07-18 14:01:33 -07:00
banner.py feat(install): warn pip/Homebrew installs are unsupported (CLI, TUI, desktop) (#57225) 2026-07-07 21:13:19 -07:00
blueprint_cmd.py refactor(cron): rebrand Cron Recipes -> Automation Blueprints 2026-06-11 10:49:47 -07:00
browser_connect.py fix: /browser connect times out when another app squats the CDP port 2026-07-18 02:49:28 -07:00
build_info.py
bundles.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
callbacks.py
checkpoints.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
claw.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
cli_agent_setup_mixin.py fix(cli): sanitize terminal escapes when replaying stored history (/resume recap, /status recap) 2026-07-17 04:53:38 -07:00
cli_billing_mixin.py feat(tui+cli): change your Nous plan from the terminal (/subscription, /topup, terminal-billing UX) (#51639) 2026-07-18 14:30:24 +05:30
cli_commands_mixin.py fix: session-scoped /fast + full /new reset to config defaults (#67979) 2026-07-20 03:27:22 -07:00
cli_output.py
clipboard.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
codex_models.py feat(openai): complete gpt-5.6 E2E — codex catalog + 272K compaction auto-raise 2026-07-10 00:47:51 +05:30
codex_runtime_plugin_migration.py
codex_runtime_switch.py refactor(codex-runtime): tidy reapply-migration control flow 2026-07-01 23:51:54 +05:30
colors.py
commands.py fix: session-scoped /fast + full /new reset to config defaults (#67979) 2026-07-20 03:27:22 -07:00
completion.py fix: batch of small robustness/correctness fixes from @kyssta-exe 2026-06-01 19:51:03 -07:00
config.py fix(desktop, ink): don't wipe messages before final message (#65919) 2026-07-20 11:42:29 -04:00
console_engine.py refactor(console): remove hosted-context command blocking from Hermes Console (#66144) 2026-07-17 04:33:34 -07:00
container_boot.py fix(container): keep named multiplex gateway slots down (#65368) 2026-07-16 14:30:05 +10:00
context_switch_guard.py fix(cli): log instead of swallow preflight-warning errors; consistent TUI warning field 2026-06-21 16:31:56 +05:30
copilot_auth.py fix(copilot): prefer endpoints.api for base URL, guard empty chat base URL 2026-06-30 03:27:41 -07:00
credential_lifecycle.py fix(credentials): unified provider key delete/update across .env, auth.json, config.yaml (#67213) 2026-07-19 03:02:21 -07:00
cron.py fix(cron): harden execution attempt ledger 2026-07-17 04:58:35 -07:00
curator.py feat(curator): add hermes curator usage — all-skills usage view 2026-07-06 12:26:21 -07:00
curses_ui.py feat(cli): ranked fuzzy search in the curses model picker 2026-06-01 16:58:58 -07:00
dashboard_register.py fix(cli): correct stale hermes auth login nous hints to hermes auth add nous (#53929) 2026-06-27 21:30:37 -07:00
debug.py fix: remove dead f-string prefixes via ruff F541 (216 sites) (#52336) 2026-07-05 13:42:46 -07:00
default_soul.py fix(soul): installers seed the real default persona, upgrade legacy empty templates (#52246) 2026-06-24 18:56:26 -07:00
dep_ensure.py fix(security): centralize credential-safe subprocess env (#29157) 2026-06-27 20:45:31 -07:00
diagnostics_upload.py feat(debug): drop dead confirm step from --nous upload (stateless NAS) 2026-06-30 17:29:23 -07:00
dingtalk_auth.py refactor(setup): route dependency installs through the canonical uv→pip→ensurepip ladder 2026-07-07 04:09:35 -07:00
doctor.py feat(providers): add `enabled: false` flag to hide a provider 2026-07-20 03:06:02 -07:00
dump.py fix(cron): accept UTF-8 BOM when reading jobs.json 2026-07-18 02:31:20 -07:00
env_loader.py fix(cli): warn once per path for UTF-32 .env refuse-to-mangle 2026-07-18 19:01:10 -04:00
fallback_cmd.py
fallback_config.py fix(auth): honor per-entry key_env when resolving fallback providers 2026-07-16 07:19:36 -07:00
gateway.py fix(gateway): recover Discord websocket and event-loop stalls 2026-07-18 20:01:55 +05:30
gateway_enroll.py feat(gateway): generic OIDC client-credentials relay provisioning (NAS-free) (#60730) 2026-07-08 16:55:32 +10:00
gateway_windows.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
goals.py fix(goals): auto-pause goal loop on consecutive transport failures 2026-07-20 05:38:25 -07:00
gui_uninstall.py feat: uninstall the Chat GUI without removing the agent (CLI + desktop UI) (#40355) 2026-06-06 18:22:38 -07:00
hooks.py feat(agent): add pre_verify hook and verify-on-stop coding guidance 2026-06-30 00:59:29 -05:00
input_sanitize.py fix(input): strip bracketed-paste leaks before prompt persistence (#62557) 2026-07-15 07:39:42 -07:00
inventory.py feat(model-switch): excluded_providers config to hide providers from /model picker 2026-07-20 03:06:02 -07:00
journey.py fix(cli): render /journey color instead of leaking raw ANSI 2026-07-01 16:25:48 -05:00
kanban.py fix: migrate CLI kanban gate + remaining mocks to 5-value judge contract 2026-07-20 05:38:25 -07:00
kanban_db.py refactor(kanban): unify attachment size cap on KANBAN_ATTACHMENT_MAX_BYTES 2026-07-16 07:33:14 -07:00
kanban_decompose.py fix(auxiliary): route direct-create aux callers through call_llm (#65029) 2026-07-15 07:39:17 -07:00
kanban_diagnostics.py fix(kanban): clear failure/crash diagnostics while a retry is in flight 2026-07-09 16:13:59 -05:00
kanban_specify.py fix(auxiliary): route direct-create aux callers through call_llm (#65029) 2026-07-15 07:39:17 -07:00
kanban_swarm.py refactor(kanban): fold worker/orchestrator skills into injected guidance (#50473) 2026-06-21 17:06:48 -07:00
logs.py fix: remove dead f-string prefixes via ruff F541 (216 sites) (#52336) 2026-07-05 13:42:46 -07:00
main.py feat(model-switch): excluded_providers config to hide providers from /model picker 2026-07-20 03:06:02 -07:00
managed_scope.py fix(managed-scope): honor managed scope in all standalone config loaders 2026-06-19 07:46:33 -07:00
managed_uv.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
mcp_catalog.py feat(mcp): enforce exact version pins across the whole MCP catalog 2026-07-15 04:56:25 -07:00
mcp_config.py fix(mcp): close hosted OAuth lifecycle gaps 2026-07-17 04:50:47 -07:00
mcp_picker.py
mcp_security.py fix(security): close hermes-0day MCP-persistence attack surface 2026-06-21 19:05:27 -07:00
mcp_startup.py fix(mcp): late-refresh must see desktop/dashboard discovery thread owner (#55514) 2026-06-30 02:08:37 -07:00
memory_oauth.py feat(memory): Honcho OAuth connect — desktop and CLI flows + token refresh (#44335) 2026-06-22 19:16:47 -05:00
memory_setup.py refactor(setup): route dependency installs through the canonical uv→pip→ensurepip ladder 2026-07-07 04:09:35 -07:00
middleware.py fix(middleware): single-use next_call guard + deepcopy-safe request copies 2026-06-06 23:07:25 +05:30
migrate.py
moa_cmd.py feat(moa): support per-slot reasoning effort 2026-07-14 21:08:22 -07:00
moa_config.py fix(moa): surface stale presets without retries 2026-07-17 13:49:12 -07:00
model_catalog.py feat(models): catalog-labeled silent default — GLM-5.2 marked "default": true in the model catalog 2026-07-15 00:10:31 -07:00
model_cost_guard.py fix(model): require confirmation for expensive model selections 2026-06-10 00:24:06 -07:00
model_normalize.py fix(gemini): strip native self prefixes before generateContent (#36141) 2026-06-13 13:47:08 -07:00
model_setup_flows.py feat(pricing): refresh Fireworks snapshot to 2026-07, cover full serverless catalog + cached picker pricing 2026-07-16 04:24:14 -07:00
model_switch.py feat(providers): post-filter picker by `enabled: false` for built-ins 2026-07-20 03:06:02 -07:00
models.py fix: dedupe claude-sonnet-5 entries after overlapping salvages 2026-07-20 02:25:44 -07:00
nous_account.py feat(billing): /credits command — balance + portal top-up handoff (#44776) 2026-06-12 08:51:10 +00:00
nous_auth_keepalive.py fix Nous auth refresh for idle agents 2026-06-21 22:43:48 -07:00
nous_billing.py Terminal-billing client hardening: shared wire types, wire-layer tests, dead RPC removal (#61067) 2026-07-18 07:05:58 -07:00
nous_subscription.py fix(browser): validate agent-browser is runnable, not just present (#51740) 2026-06-24 00:14:49 -07:00
onepassword_secrets_cli.py fix(secrets): remove unused masked_secret_prompt import from onepassword CLI 2026-07-06 04:58:07 -07:00
oneshot.py refactor(cli): apply /simplify-code findings to oneshot teardown 2026-07-20 12:59:28 +05:30
pairing.py
partial_compress.py feat(commands): /compact alias + --preview/--dry-run flags for /compress (#3243 salvage) 2026-07-02 05:10:31 -07:00
pets.py feat(pets): generation RPCs, non-blocking gallery + gateway plumbing 2026-06-24 13:48:38 -05:00
platforms.py
plugins.py fix(approval): emit observer hooks for smart verdicts 2026-07-13 02:00:09 -07:00
plugins_cmd.py fix(dashboard): unblock basic auth plugin when setting password interactively (#54489) (#63786) 2026-07-16 19:49:39 -04:00
portal_cli.py fix: remove dead f-string prefixes via ruff F541 (216 sites) (#52336) 2026-07-05 13:42:46 -07:00
profile_describer.py fix(auxiliary): route direct-create aux callers through call_llm (#65029) 2026-07-15 07:39:17 -07:00
profile_distribution.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
profiles.py fix: remove dead f-string prefixes via ruff F541 (216 sites) (#52336) 2026-07-05 13:42:46 -07:00
projects_cmd.py feat(projects): add per-profile project store 2026-06-25 16:40:26 -05:00
projects_db.py feat(projects): add per-profile project store 2026-06-25 16:40:26 -05:00
prompt_size.py fix(prompt-size): respect enabled/disabled toolsets per platform 2026-07-05 19:13:20 -07:00
provider_catalog.py feat(moa): expose MoA presets as selectable virtual models (#46081) 2026-06-25 13:52:06 -07:00
providers.py fix(model-switch): override stale api_mode with host-mandated mode on OpenAI-direct switch 2026-07-17 15:47:08 -07:00
psutil_android.py
pt_input_extras.py fix(cli): ignore terminal focus reports (salvage of #16780) 2026-05-29 00:31:44 -07:00
pty_bridge.py fix(pty-bridge): mark os.killpg/getpgid windows-footgun-ok (POSIX-only module) 2026-06-08 07:03:12 -07:00
pty_session.py feat(pty): periodic reaper wired into dashboard lifespan 2026-07-07 15:15:37 -07:00
relaunch.py revert(windows): roll back terminal-popup PRs #53791 #53810 #53829 (#53853) 2026-06-27 15:59:00 -07:00
runtime_provider.py feat(providers): extend `enabled: false` gate to built-in resolution 2026-07-20 03:06:02 -07:00
secret_prompt.py feat(memory): improve OpenViking setup UX 2026-06-17 01:04:26 +08:00
secrets_cli.py fix(secrets): fail early with clear error when bitwarden setup runs without TTY (#40571) 2026-06-06 18:36:40 -07:00
security_advisories.py fix(stt,tts): restore mistralai — 2.4.8 is clean, ban lifted (#34841) 2026-05-29 13:24:12 -07:00
security_audit.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
security_audit_startup.py style(security-audit): add explicit encoding to read_text calls (ruff PLW1514) 2026-06-21 19:05:27 -07:00
send_cmd.py fix(managed-scope): honor managed scope in config→env bridges too 2026-06-19 07:46:33 -07:00
service_manager.py fix(s6): dot-prefix gateway staging dir so svscan ignores it mid-build (#54834) 2026-06-29 21:33:00 +10:00
session_export.py feat: add prompt-only session export 2026-07-07 13:29:58 -07:00
session_export_html.py fix(session-export): escape html tool call names 2026-07-09 19:54:02 -07:00
session_export_md.py feat(sessions): full prune-filter set + --redact on sessions export 2026-07-07 12:36:41 -07:00
session_filters.py feat(sessions): full filter surface for prune + bulk archive subcommand (#59327) 2026-07-05 22:04:52 -07:00
session_listing.py feat(gateway): add /sessions search <query> (#57685) 2026-07-03 13:44:00 -07:00
session_recap.py fix(cli): sanitize terminal escapes when replaying stored history (/resume recap, /status recap) 2026-07-17 04:53:38 -07:00
setup.py feat(models): add Claude Sonnet 5 support 2026-07-20 02:25:44 -07:00
setup_whatsapp_cloud.py fix(whatsapp-cloud): review follow-ups for #43921 2026-06-11 07:51:01 -07:00
skills_config.py fix(cli): normalize malformed skills config in get_disabled_skills (#61797) 2026-07-09 19:57:54 -07:00
skills_hub.py fix: widen UTF-8 BOM tolerance to all sibling frontmatter parsers 2026-07-17 04:52:02 -07:00
skin_engine.py
slack_cli.py fix(slack): complete agent view workspace routing 2026-07-14 13:58:36 -07:00
sqlite_util.py feat(projects): add per-profile project store 2026-06-25 16:40:26 -05:00
status.py fix(cron): accept UTF-8 BOM when reading jobs.json 2026-07-18 02:31:20 -07:00
stdio.py chore: prune unused imports and duplicate import redefinitions 2026-05-28 22:26:25 -07:00
suggestions_cmd.py fix: remove dead f-string prefixes via ruff F541 (216 sites) (#52336) 2026-07-05 13:42:46 -07:00
telegram_managed_bot.py Add CLI Telegram QR onboarding 2026-06-05 03:20:10 -07:00
timeouts.py
tips.py feat(browser): store full snapshots on truncation; make eval denylist opt-in (#65923) 2026-07-16 23:41:26 -07:00
tools_config.py feat(desktop): five Capabilities-tab UX fixes from live testing — hints, vision link, web split, key deep-links (#67482) 2026-07-19 05:45:01 -07:00
toolset_validation.py fix(config): surface invalid platform_toolsets instead of silently dropping tools (#38798) 2026-06-26 14:07:43 +05:30
uninstall.py feat: add uninstall dry-run mode 2026-07-07 05:12:24 -07:00
urllib_security.py fix(security): order sanitizer after installed hooks 2026-07-11 12:28:55 +05:30
voice.py
web_git.py feat(desktop): auto-fetch remote base branch before worktree add 2026-07-13 14:53:34 -04:00
web_server.py fix(dashboard): clear the model mirror when its custom endpoint is deleted 2026-07-20 08:37:14 -07:00
webhook.py fix(webhook): make dual-stack bind exclusive 2026-07-16 12:36:51 +05:30
win_pty_bridge.py feat(windows): enable dashboard /chat tab via ConPTY (win_pty_bridge) + tests (#42251) 2026-06-08 11:32:43 -07:00
write_approval_commands.py refactor(memory,skills): replace tri-state write_mode with boolean write_approval (default off) (#43354) 2026-06-09 23:21:14 -07:00
xai_retirement.py fix(config): guard xai migration writer + drop gratuitous annotation 2026-07-05 23:00:34 +05:30