feat(dashboard): configure main + auxiliary models from Models page#17802
Merged
Conversation
Dashboard Models page was analytics-only — no way to pick a model as main
for new sessions or override an auxiliary task slot without hand-editing
config.yaml or running a /model slash command inside a chat.
Changes:
- hermes_cli/web_server.py: three REST endpoints (GET /api/model/options,
GET /api/model/auxiliary, POST /api/model/set). Reuses
list_authenticated_providers() from model_switch.py so the REST path
surfaces the same curated model lists as the TUI-gateway model.options
JSON-RPC. POST /api/model/set writes model.provider + model.default for
scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary
(with task="" meaning 'all 8 slots' and task="__reset__" resetting them
to auto).
- web/src/components/ModelPickerDialog.tsx: accepts an optional loader +
onApply pair so it works without an open chat PTY. ChatSidebar's
gw-WebSocket path still works unchanged (back-compat).
- web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing
main model + collapsible list of 8 auxiliary tasks with per-row Change
buttons and Reset all to auto. Every existing model card gets a
'Use as' dropdown for one-click assignment to main or any aux slot.
Cards badged 'main' or 'aux · <task>' when currently assigned.
- website/docs/user-guide/configuring-models.md: new docs page walking
through both UI paths, aux task override patterns, troubleshooting,
plus REST/CLI alternatives.
- Screenshots under website/static/img/docs/dashboard-models/.
Applies to new sessions only — running sessions keep their model (use
/model slash command to hot-swap a live session). No prompt-cache
invalidation on existing sessions.
nickdlkk
pushed a commit
to nickdlkk/hermes-agent
that referenced
this pull request
May 11, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
4 tasks
02356abc
pushed a commit
to 02356abc/hermes-agent
that referenced
this pull request
May 14, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
jsboige
pushed a commit
to jsboige/hermes-agent
that referenced
this pull request
May 14, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
dannyJ848
pushed a commit
to dannyJ848/hermes-agent
that referenced
this pull request
May 17, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
gweeteve
pushed a commit
to gweeteve/hermes-agent
that referenced
this pull request
Jun 2, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
Egavasyug
pushed a commit
to Egavasyug/hermes-agent
that referenced
this pull request
Jun 10, 2026
…ousResearch#17802) Dashboard Models page was analytics-only — no way to pick a model as main for new sessions or override an auxiliary task slot without hand-editing config.yaml or running a /model slash command inside a chat. Changes: - hermes_cli/web_server.py: three REST endpoints (GET /api/model/options, GET /api/model/auxiliary, POST /api/model/set). Reuses list_authenticated_providers() from model_switch.py so the REST path surfaces the same curated model lists as the TUI-gateway model.options JSON-RPC. POST /api/model/set writes model.provider + model.default for scope=main, and auxiliary.<task>.{provider,model} for scope=auxiliary (with task="" meaning 'all 8 slots' and task="__reset__" resetting them to auto). - web/src/components/ModelPickerDialog.tsx: accepts an optional loader + onApply pair so it works without an open chat PTY. ChatSidebar's gw-WebSocket path still works unchanged (back-compat). - web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing main model + collapsible list of 8 auxiliary tasks with per-row Change buttons and Reset all to auto. Every existing model card gets a 'Use as' dropdown for one-click assignment to main or any aux slot. Cards badged 'main' or 'aux · <task>' when currently assigned. - website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths, aux task override patterns, troubleshooting, plus REST/CLI alternatives. - Screenshots under website/static/img/docs/dashboard-models/. Applies to new sessions only — running sessions keep their model (use /model slash command to hot-swap a live session). No prompt-cache invalidation on existing sessions.
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.
Summary
Dashboard Models page can now pick models for the main slot and the 8 auxiliary task slots (vision, web_extract, compression, session_search, skills_hub, approval, mcp, title_generation). Previously it was analytics-only — you had to hand-edit
config.yamlor run/model --globalfrom inside a chat. Writes go straight to~/.hermes/config.yamland apply to new sessions only — no prompt-cache invalidation on anything currently running.Changes
hermes_cli/web_server.py: 3 REST endpoints —GET /api/model/options(reuseslist_authenticated_providersso the curated model lists match the TUI-gatewaymodel.optionsJSON-RPC 1:1),GET /api/model/auxiliary,POST /api/model/set. Supports scope=main, scope=auxiliary+task=X (single slot), scope=auxiliary+task="" (all 8 slots at once), scope=auxiliary+task="reset" (reset all to auto).web/src/components/ModelPickerDialog.tsx: accepts optionalloader+onApplyfor HTTP mode. ChatSidebar's existinggw-WebSocket path keeps working unchanged (back-compat).web/src/pages/ModelsPage.tsx: Model Settings panel at the top showing the main model + collapsible 8 auxiliary-task rows (per-row Change + bulk Reset all to auto). Every usage card gets a Use as dropdown — one click to assign to main or any specific aux slot. Cards badgedmainoraux · <task>when currently assigned so you can see wiring at a glance.website/docs/user-guide/configuring-models.md: new docs page walking through both UI paths (Model Settings panel + per-card Use as), aux task override patterns (when to override vs stay on auto), troubleshooting, plus CLI and REST alternatives.website/static/img/docs/dashboard-models/.Screenshots
Models page — new Model Settings panel at the top
Expanded auxiliary section — 8 task slots with per-row Change + Reset all to auto
Picker dialog — reused from chat mode but HTTP-driven here
Per-card Use as dropdown — one-click assign to main or any aux task
Validation
scripts/run_tests.sh tests/hermes_cli/ -k "web_server or model"→ 772 passed.web/:npx tsc -bclean,npm run buildclean.website/:npm run buildclean (pre-existing unrelated broken-anchor warning on adding-platform-adapters).HERMES_HOME: seeds config, exercises all 3 endpoints (main set, aux single, aux bulk, aux reset, 3 negative inputs, auth gating) — all pass. Then loaded the page in a real browser, verified main/aux badges render, clicking Use as → Main model updatesconfig.yamlon disk (confirmed via filesystem read), Reset all to auto flips all 8 slots back toprovider: auto, model: ''.Notes
/modelinside the chat to hot-swap the live session. This is deliberate: swapping the main model invalidates the prompt cache, and we reserve that for the explicit slash command.DEFAULT_CONFIG["auxiliary"]. Adding a new aux task in the future means updating_AUX_TASK_SLOTSinweb_server.py+AUX_TASKSinModelsPage.tsx(there's a cross-reference comment on both).