fix(timeline): persist typed display events - #69771
Merged
Merged
Conversation
Contributor
૮ >ﻌ< ა ci reviewran on a567ae7 all good! |
alt-glitch
added
type/bug
ethernet8023
changed the title
Filter verification-stop nudges and context-compaction handoffs at the stored-history mapper boundary. Preserve a real reply when a compaction handoff shares its stored message.
Drive tui_gateway.entry over its stdio JSON-RPC transport against the mock provider, wait for real completion events, and persist normal session history through AIAgent and SessionDB. Migrate resume and hidden-history coverage, including real compression and live verify-on-stop scaffolding, then remove the unused direct SessionDB import scripts.
Run the stdio gateway through uv's synced project environment outside the Nix dev shell, while retaining the fully provisioned Nix Python when the shell advertises HERMES_PYTHON_SRC_ROOT.
Mark the Nix-built Python environment active in the dev shell so the shared E2E session builder can always run through `uv run --active --no-sync`.
ethernet8023
force-pushed
the
ethie/fix-verifiction-unhide
branch
from
July 23, 2026 04:41
e66907c to
19f3b3d
Compare
ethernet8023
added a commit
that referenced
this pull request
Jul 23, 2026
…erve through rewrites, fix /resume display history Three review findings from PR #69771: 1. Provider payload leak: display_kind and display_metadata were forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends can reject the next request after a model switch or resumed typed event. Strip both from the per-request api_msg copy in conversation_loop alongside the existing api_content pop. 2. Rewrite/import data loss: _insert_message_rows preserved display_kind but silently dropped display_metadata. After replace_messages, archive_and_compact, or session import, async-delegation completion events lost their task counts and fell back to generic display text. Add display_metadata to the INSERT columns and bind tuple. 3. CLI /resume stale recap: startup --resume A set _resume_display_history from A's lineage. A subsequent in-session /resume B loaded B only into conversation_history via get_messages_as_conversation, leaving the stale A display projection. _display_resumed_history preferentially read the stale attribute, showing A's recap for B. Switch /resume to get_resume_conversations and update _resume_display_history alongside conversation_history. Tests: 890 Python (5 files), 35 desktop TS — all green.
…erve through rewrites, fix /resume display history Three review findings from PR #69771: 1. Provider payload leak: display_kind and display_metadata were forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends can reject the next request after a model switch or resumed typed event. Strip both from the per-request api_msg copy in conversation_loop alongside the existing api_content pop. 2. Rewrite/import data loss: _insert_message_rows preserved display_kind but silently dropped display_metadata. After replace_messages, archive_and_compact, or session import, async-delegation completion events lost their task counts and fell back to generic display text. Add display_metadata to the INSERT columns and bind tuple. 3. CLI /resume stale recap: startup --resume A set _resume_display_history from A's lineage. A subsequent in-session /resume B loaded B only into conversation_history via get_messages_as_conversation, leaving the stale A display projection. _display_resumed_history preferentially read the stale attribute, showing A's recap for B. Switch /resume to get_resume_conversations and update _resume_display_history alongside conversation_history. Tests: 890 Python (5 files), 35 desktop TS — all green.
ethernet8023
force-pushed
the
ethie/fix-verifiction-unhide
branch
from
July 23, 2026 06:21
d58369f to
4d5180a
Compare
The TUI was not handling display_kind at all — model switch markers and
async delegation completions rendered as opaque user messages with the
full [System: ...] text, and hidden compaction handoffs were visible.
Wire display_kind through the full TUI chain:
- _history_to_messages (tui_gateway/server.py) forwards display_kind
and display_metadata to the gateway transcript payload.
- GatewayTranscriptMessage (gatewayTypes.ts) gains both fields.
- Msg.kind (types.ts) gains 'event' value.
- toTranscriptMessages (domain/messages.ts) maps:
- hidden → skip entirely
- model_switch → event "model changed"
- async_delegation_complete → event "N background agents finished"
(or "background agent work finished" without metadata)
- messageGroup (blockLayout.ts) routes event to its own group, with
SELF_SPACED + PAINTS_TRAILING_GAP so it owns its margins.
- messageLine.tsx renders event-kind as a dim ◈ marker with no gutter,
matching the CLI's ◈ event rendering.
- 4 new TUI tests for hidden/model_switch/async_delegation mapping.
TUI typecheck: clean. TUI lint: 0 errors (2 pre-existing warnings).
TUI tests: 9 passed (1 pre-existing failure on main, unrelated).
randlee
pushed a commit
to randlee/hermes-agent
that referenced
this pull request
Aug 11, 2026
* fix(desktop): hide persisted agent-only history scaffolding Filter verification-stop nudges and context-compaction handoffs at the stored-history mapper boundary. Preserve a real reply when a compaction handoff shares its stored message. * test(desktop): build persisted E2E sessions through the real agent Drive tui_gateway.entry over its stdio JSON-RPC transport against the mock provider, wait for real completion events, and persist normal session history through AIAgent and SessionDB. Migrate resume and hidden-history coverage, including real compression and live verify-on-stop scaffolding, then remove the unused direct SessionDB import scripts. * fix(desktop): use the provisioned Python for real-session E2Es Run the stdio gateway through uv's synced project environment outside the Nix dev shell, while retaining the fully provisioned Nix Python when the shell advertises HERMES_PYTHON_SRC_ROOT. * fix(nix): expose the provisioned Python environment to uv Mark the Nix-built Python environment active in the dev shell so the shared E2E session builder can always run through `uv run --active --no-sync`. * fix(timeline): persist typed display events * fix(timeline): strip display-only fields from provider payloads, preserve through rewrites, fix /resume display history Three review findings from PR NousResearch#69771: 1. Provider payload leak: display_kind and display_metadata were forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends can reject the next request after a model switch or resumed typed event. Strip both from the per-request api_msg copy in conversation_loop alongside the existing api_content pop. 2. Rewrite/import data loss: _insert_message_rows preserved display_kind but silently dropped display_metadata. After replace_messages, archive_and_compact, or session import, async-delegation completion events lost their task counts and fell back to generic display text. Add display_metadata to the INSERT columns and bind tuple. 3. CLI /resume stale recap: startup --resume A set _resume_display_history from A's lineage. A subsequent in-session /resume B loaded B only into conversation_history via get_messages_as_conversation, leaving the stale A display projection. _display_resumed_history preferentially read the stale attribute, showing A's recap for B. Switch /resume to get_resume_conversations and update _resume_display_history alongside conversation_history. Tests: 890 Python (5 files), 35 desktop TS — all green. * feat(tui): render typed display events as ◈ markers in the Ink TUI The TUI was not handling display_kind at all — model switch markers and async delegation completions rendered as opaque user messages with the full [System: ...] text, and hidden compaction handoffs were visible. Wire display_kind through the full TUI chain: - _history_to_messages (tui_gateway/server.py) forwards display_kind and display_metadata to the gateway transcript payload. - GatewayTranscriptMessage (gatewayTypes.ts) gains both fields. - Msg.kind (types.ts) gains 'event' value. - toTranscriptMessages (domain/messages.ts) maps: - hidden → skip entirely - model_switch → event "model changed" - async_delegation_complete → event "N background agents finished" (or "background agent work finished" without metadata) - messageGroup (blockLayout.ts) routes event to its own group, with SELF_SPACED + PAINTS_TRAILING_GAP so it owns its margins. - messageLine.tsx renders event-kind as a dim ◈ marker with no gutter, matching the CLI's ◈ event rendering. - 4 new TUI tests for hidden/model_switch/async_delegation mapping. TUI typecheck: clean. TUI lint: 0 errors (2 pre-existing warnings). TUI tests: 9 passed (1 pre-existing failure on main, unrelated).
prmartinow
pushed a commit
to prmartinow/hermes-agent
that referenced
this pull request
Aug 26, 2026
* fix(desktop): hide persisted agent-only history scaffolding Filter verification-stop nudges and context-compaction handoffs at the stored-history mapper boundary. Preserve a real reply when a compaction handoff shares its stored message. * test(desktop): build persisted E2E sessions through the real agent Drive tui_gateway.entry over its stdio JSON-RPC transport against the mock provider, wait for real completion events, and persist normal session history through AIAgent and SessionDB. Migrate resume and hidden-history coverage, including real compression and live verify-on-stop scaffolding, then remove the unused direct SessionDB import scripts. * fix(desktop): use the provisioned Python for real-session E2Es Run the stdio gateway through uv's synced project environment outside the Nix dev shell, while retaining the fully provisioned Nix Python when the shell advertises HERMES_PYTHON_SRC_ROOT. * fix(nix): expose the provisioned Python environment to uv Mark the Nix-built Python environment active in the dev shell so the shared E2E session builder can always run through `uv run --active --no-sync`. * fix(timeline): persist typed display events * fix(timeline): strip display-only fields from provider payloads, preserve through rewrites, fix /resume display history Three review findings from PR NousResearch#69771: 1. Provider payload leak: display_kind and display_metadata were forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends can reject the next request after a model switch or resumed typed event. Strip both from the per-request api_msg copy in conversation_loop alongside the existing api_content pop. 2. Rewrite/import data loss: _insert_message_rows preserved display_kind but silently dropped display_metadata. After replace_messages, archive_and_compact, or session import, async-delegation completion events lost their task counts and fell back to generic display text. Add display_metadata to the INSERT columns and bind tuple. 3. CLI /resume stale recap: startup --resume A set _resume_display_history from A's lineage. A subsequent in-session /resume B loaded B only into conversation_history via get_messages_as_conversation, leaving the stale A display projection. _display_resumed_history preferentially read the stale attribute, showing A's recap for B. Switch /resume to get_resume_conversations and update _resume_display_history alongside conversation_history. Tests: 890 Python (5 files), 35 desktop TS — all green. * feat(tui): render typed display events as ◈ markers in the Ink TUI The TUI was not handling display_kind at all — model switch markers and async delegation completions rendered as opaque user messages with the full [System: ...] text, and hidden compaction handoffs were visible. Wire display_kind through the full TUI chain: - _history_to_messages (tui_gateway/server.py) forwards display_kind and display_metadata to the gateway transcript payload. - GatewayTranscriptMessage (gatewayTypes.ts) gains both fields. - Msg.kind (types.ts) gains 'event' value. - toTranscriptMessages (domain/messages.ts) maps: - hidden → skip entirely - model_switch → event "model changed" - async_delegation_complete → event "N background agents finished" (or "background agent work finished" without metadata) - messageGroup (blockLayout.ts) routes event to its own group, with SELF_SPACED + PAINTS_TRAILING_GAP so it owns its margins. - messageLine.tsx renders event-kind as a dim ◈ marker with no gutter, matching the CLI's ◈ event rendering. - 4 new TUI tests for hidden/model_switch/async_delegation mapping. TUI typecheck: clean. TUI lint: 0 errors (2 pre-existing warnings). TUI tests: 9 passed (1 pre-existing failure on main, unrelated).
melon-xf
added a commit
to melon-xf/hermes-agent
that referenced
this pull request
Sep 3, 2026
* fix(desktop): hide persisted agent-only history scaffolding Filter verification-stop nudges and context-compaction handoffs at the stored-history mapper boundary. Preserve a real reply when a compaction handoff shares its stored message. * test(desktop): build persisted E2E sessions through the real agent Drive tui_gateway.entry over its stdio JSON-RPC transport against the mock provider, wait for real completion events, and persist normal session history through AIAgent and SessionDB. Migrate resume and hidden-history coverage, including real compression and live verify-on-stop scaffolding, then remove the unused direct SessionDB import scripts. * fix(desktop): use the provisioned Python for real-session E2Es Run the stdio gateway through uv's synced project environment outside the Nix dev shell, while retaining the fully provisioned Nix Python when the shell advertises HERMES_PYTHON_SRC_ROOT. * fix(nix): expose the provisioned Python environment to uv Mark the Nix-built Python environment active in the dev shell so the shared E2E session builder can always run through `uv run --active --no-sync`. * fix(timeline): persist typed display events * fix(timeline): strip display-only fields from provider payloads, preserve through rewrites, fix /resume display history Three review findings from PR NousResearch#69771: 1. Provider payload leak: display_kind and display_metadata were forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends can reject the next request after a model switch or resumed typed event. Strip both from the per-request api_msg copy in conversation_loop alongside the existing api_content pop. 2. Rewrite/import data loss: _insert_message_rows preserved display_kind but silently dropped display_metadata. After replace_messages, archive_and_compact, or session import, async-delegation completion events lost their task counts and fell back to generic display text. Add display_metadata to the INSERT columns and bind tuple. 3. CLI /resume stale recap: startup --resume A set _resume_display_history from A's lineage. A subsequent in-session /resume B loaded B only into conversation_history via get_messages_as_conversation, leaving the stale A display projection. _display_resumed_history preferentially read the stale attribute, showing A's recap for B. Switch /resume to get_resume_conversations and update _resume_display_history alongside conversation_history. Tests: 890 Python (5 files), 35 desktop TS — all green. * feat(tui): render typed display events as ◈ markers in the Ink TUI The TUI was not handling display_kind at all — model switch markers and async delegation completions rendered as opaque user messages with the full [System: ...] text, and hidden compaction handoffs were visible. Wire display_kind through the full TUI chain: - _history_to_messages (tui_gateway/server.py) forwards display_kind and display_metadata to the gateway transcript payload. - GatewayTranscriptMessage (gatewayTypes.ts) gains both fields. - Msg.kind (types.ts) gains 'event' value. - toTranscriptMessages (domain/messages.ts) maps: - hidden → skip entirely - model_switch → event "model changed" - async_delegation_complete → event "N background agents finished" (or "background agent work finished" without metadata) - messageGroup (blockLayout.ts) routes event to its own group, with SELF_SPACED + PAINTS_TRAILING_GAP so it owns its margins. - messageLine.tsx renders event-kind as a dim ◈ marker with no gutter, matching the CLI's ◈ event rendering. - 4 new TUI tests for hidden/model_switch/async_delegation mapping. TUI typecheck: clean. TUI lint: 0 errors (2 pre-existing warnings). TUI tests: 9 passed (1 pre-existing failure on main, unrelated).
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
2 participants
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.
What does this PR do?
Persists durable
display_kindand JSONdisplay_metadataseparately from model-facing message roles/content, so synthetic transcript entries (model switches, async delegation completions, compaction handoffs) render as timeline events instead of opaque user messages across all three chat surfaces: CLI, TUI, and desktop.The core problem: model switch markers, async delegation completions, and compaction handoffs were all persisted as
role: usermessages with opaque content. On resume, every surface displayed them as if the user had typed that text. This PR introduces adisplay_kindcolumn on themessagestable and adisplay_metadataJSON sidecar, so the model still receives the originalrole/contentunchanged while each surface renders the row as a typed timeline event (or hides it entirely for compaction handoffs).Review fixes (commit
d58369f0c)Three findings from top-to-bottom code review, all fixed with regression tests:
Provider payload leak —
display_kind/display_metadatawere forwarded to the provider API as unknown message fields. Strict OpenAI-compatible backends could reject the next request after a model switch or resumed typed event. Fixed by stripping both from the per-requestapi_msgcopy inconversation_loop.pyalongside the existingapi_contentpop.Rewrite/import data loss —
_insert_message_rowspreserveddisplay_kindbut silently droppeddisplay_metadata. Afterreplace_messages,archive_and_compact, or session import, async-delegation events lost their task counts. Fixed by addingdisplay_metadatato the INSERT columns and bind tuple.CLI
/resumestale recap — Startup--resume Aset_resume_display_historyfrom A's lineage. A subsequent in-session/resume Bloaded B only intoconversation_history, leaving the stale A display projection. Fixed by switching/resumetoget_resume_conversationsand updating_resume_display_historyalongsideconversation_history.TUI event rendering (commit
a567ae795)The TUI was not handling
display_kindat all — model switch markers and async delegation completions rendered as opaque user messages, and hidden compaction handoffs were visible. Wireddisplay_kindthrough the full TUI chain:_history_to_messagesforwards the fields,toTranscriptMessagesmaps them to a new'event'message kind, andmessageLine.tsxrenders events as dim◈markers.Related Issue
Fixes #69629 (timeline events leak as user messages on resume)
Type of Change
Changes Made
hermes_state.py— Adddisplay_kindanddisplay_metadatacolumns tomessagestable.append_message()accepts and persists both.set_latest_matching_message_display_kind()stamps presentation metadata._insert_message_rows()preservesdisplay_metadatathrough rewrites.get_resume_conversations()returns(model_history, display_history)from one lineage SELECT.tui_gateway/server.py— Model switch marker and async delegation completion stampdisplay_kind/display_metadata._history_to_messages()forwards display fields to the TUI transcript payload.agent/conversation_loop.py— Stripdisplay_kind/display_metadatafrom the per-requestapi_msgcopy before provider dispatch.run_agent.py— Stampsdisplay_kind="hidden"on compaction handoff rows.hermes_cli/cli_agent_setup_mixin.py+cli_commands_mixin.py— CLI resume and/resumeuseget_resume_conversations(), render timeline events as◈markers, skip hidden rows.ui-tui/src/—GatewayTranscriptMessage+Msg+TranscriptRowgain display fields.toTranscriptMessages()mapsdisplay_kindto'event'kind.messageLine.tsxrenders events as dim◈markers.blockLayout.tsadds'event'group.apps/desktop/src/lib/chat-messages.ts+types/hermes.ts—toChatMessages()projectsdisplay_kindinto system-role timeline bubbles.apps/desktop/e2e/—RealSessionBuildercreates real persisted sessions.hidden-history-messages.spec.tsE2E.nix/devShell.nix— ExportVIRTUAL_ENVfor uv.How to Test
nix develop -c bash -lc 'export HERMES_PYTHON="$(command -v python)"; scripts/run_tests.sh tests/test_hermes_state.py tests/test_tui_gateway_server.py tests/tui_gateway/test_model_switch_marker_role.py tests/cli/test_resume_display.py tests/cli/test_cli_resume_command.py tests/run_agent/test_session_meta_filtering.py'— 911 passedcd ui-tui && npm run typecheck && npm run lint && npx vitest run src/__tests__/messages.test.ts— 9 passed (1 pre-existing failure on main)cd apps/desktop && npm run typecheck && npm run lint && npm test -- --run src/lib/chat-messages.test.ts— 35 passednix build .#then resume a seeded session in CLI, TUI, and desktop underWLR_BACKENDS=headless cageChecklist
Screenshots / Logs
CLI resume —
◈event markers, hidden rows suppressedTUI resume — dim
◈markers in the Ink TUIDesktop — timeline events rendered via Playwright E2E
Captured with the desktop E2E Playwright fixtures (
launchDesktop+page.screenshot) underWLR_BACKENDS=headless cage, using a real persisted session seeded viaRealSessionBuilder.toChatMessages()projectsdisplay_kindinto system-role timeline bubbles.Verification output
Timeline display evidence
All captures use the same seeded display event history: a model-switch marker,
a two-task delegation completion, and a hidden compaction row. The hidden row
does not render.
CLI and TUI were captured in Ghostty under headless Cage. Desktop was captured
by the real Electron Playwright fixture under headless Cage.