Incorrect styling for customGPT name #29
Comments
|
Can you provide a few examples with your characters so I can test what the expected result should be? In English, the custom name looks fine with |
danorlando
added a commit
that referenced
this issue
May 7, 2023
* server-side JWT auth implementation * move oauth routes and strategies, fix bugs * backend modifications for wiring up the frontend login and reg forms * Add frontend data services for login and registration * Add login and registration forms * Implment auth context, functional client side auth * protect routes with jwt auth * finish local strategy (using local storage) * Start setting up google auth * disable token refresh, remove old auth middleware * refactor client, add ApiErrorBoundary context * disable google and facebook strategies * fix: fix presets not displaying specific to user * fix: fix issue with browser refresh * fix: casing issue with User.js (#11) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * fix: issue with api error watcher (#12) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * feat: add google auth social login * fix: make google login url dynamic based on dev/prod * fix: bug where UI is briefly displayed before redirecting to login * fix: fix cookie expires value for local auth * Update README.md * Update LOCAL_INSTALL structure * Add local testing instructions * Only load google strategy if client id and secret are provided * Update .env.example files with new params * fix issue with not redirecting to register form * only show google login button if value is set in .env * cleanup log messages * Add label to button for google login on login form * doc: fix client/server url values in .env.example * feat: add error message details to registration failure * Restore preventing paste on confirm password * auto-login user after registering * feat: forgot password (#24) * make login/reg pages look like openai's * add password reset data services * new form designs similar to openai, add password reset pages * add api's for password reset * email utils for password reset * remove bcrypt salt rounds from process.env * refactor: restructure api auth code, consolidate routes (#25) * add api's for password reset * remove bcrypt salt rounds from process.env * refactor: consolidate auth routes, use controller pattern * refactor: code cleanup * feat: migrate data to first user (#26) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes after refactor (#27) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: issue with auto-login when logging out then logging in with new browser window (#28) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: Update README and .env.example files with user system information (#29) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: update README and .env.example files * Fixup: LOCAL_INSTALL.md PS instructions (#200) (#30) Co-authored-by: alfredo-f <[email protected]> * feat: send user with completion to protect against abuse (#31) * Fixup: LOCAL_INSTALL.md PS instructions (#200) * server-side JWT auth implementation * move oauth routes and strategies, fix bugs * backend modifications for wiring up the frontend login and reg forms * Add frontend data services for login and registration * Add login and registration forms * Implment auth context, functional client side auth * protect routes with jwt auth * finish local strategy (using local storage) * Start setting up google auth * disable token refresh, remove old auth middleware * refactor client, add ApiErrorBoundary context * disable google and facebook strategies * fix: fix presets not displaying specific to user * fix: fix issue with browser refresh * fix: casing issue with User.js (#11) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * feat: add google auth social login * fix: make google login url dynamic based on dev/prod * fix: bug where UI is briefly displayed before redirecting to login * fix: fix cookie expires value for local auth * Only load google strategy if client id and secret are provided * Update .env.example files with new params * fix issue with not redirecting to register form * only show google login button if value is set in .env * cleanup log messages * Add label to button for google login on login form * doc: fix client/server url values in .env.example * feat: add error message details to registration failure * Restore preventing paste on confirm password * auto-login user after registering * feat: forgot password (#24) * make login/reg pages look like openai's * add password reset data services * new form designs similar to openai, add password reset pages * add api's for password reset * email utils for password reset * remove bcrypt salt rounds from process.env * refactor: restructure api auth code, consolidate routes (#25) * add api's for password reset * remove bcrypt salt rounds from process.env * refactor: consolidate auth routes, use controller pattern * refactor: code cleanup * feat: migrate data to first user (#26) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes after refactor (#27) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: issue with auto-login when logging out then logging in with new browser window (#28) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: Update README and .env.example files with user system information (#29) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: update README and .env.example files * Send user id to openai to protect against abuse * add meilisearch to gitignore * Remove webpack --------- Co-authored-by: alfredo-f <[email protected]> --------- Co-authored-by: Danny Avila <[email protected]> Co-authored-by: Alfredo Fomitchenko <[email protected]>
BlitzWrecker
pushed a commit
to BlitzWrecker/AITok
that referenced
this issue
Jul 2, 2023
…vila#29) * Delete container.yml * Add release.yml to run upon production release * only trigger on released, not on preleases * query the right task-definition family * add client/version.txt for verification * release can check /assets/version.txt
cnkang
referenced
this issue
in cnkang/LibreChat
Feb 6, 2024
* server-side JWT auth implementation * move oauth routes and strategies, fix bugs * backend modifications for wiring up the frontend login and reg forms * Add frontend data services for login and registration * Add login and registration forms * Implment auth context, functional client side auth * protect routes with jwt auth * finish local strategy (using local storage) * Start setting up google auth * disable token refresh, remove old auth middleware * refactor client, add ApiErrorBoundary context * disable google and facebook strategies * fix: fix presets not displaying specific to user * fix: fix issue with browser refresh * fix: casing issue with User.js (#11) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * fix: issue with api error watcher (#12) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * feat: add google auth social login * fix: make google login url dynamic based on dev/prod * fix: bug where UI is briefly displayed before redirecting to login * fix: fix cookie expires value for local auth * Update README.md * Update LOCAL_INSTALL structure * Add local testing instructions * Only load google strategy if client id and secret are provided * Update .env.example files with new params * fix issue with not redirecting to register form * only show google login button if value is set in .env * cleanup log messages * Add label to button for google login on login form * doc: fix client/server url values in .env.example * feat: add error message details to registration failure * Restore preventing paste on confirm password * auto-login user after registering * feat: forgot password (#24) * make login/reg pages look like openai's * add password reset data services * new form designs similar to openai, add password reset pages * add api's for password reset * email utils for password reset * remove bcrypt salt rounds from process.env * refactor: restructure api auth code, consolidate routes (#25) * add api's for password reset * remove bcrypt salt rounds from process.env * refactor: consolidate auth routes, use controller pattern * refactor: code cleanup * feat: migrate data to first user (#26) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes after refactor (#27) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: issue with auto-login when logging out then logging in with new browser window (#28) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: Update README and .env.example files with user system information (#29) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: update README and .env.example files * Fixup: LOCAL_INSTALL.md PS instructions (danny-avila#200) (#30) Co-authored-by: alfredo-f <[email protected]> * feat: send user with completion to protect against abuse (#31) * Fixup: LOCAL_INSTALL.md PS instructions (danny-avila#200) * server-side JWT auth implementation * move oauth routes and strategies, fix bugs * backend modifications for wiring up the frontend login and reg forms * Add frontend data services for login and registration * Add login and registration forms * Implment auth context, functional client side auth * protect routes with jwt auth * finish local strategy (using local storage) * Start setting up google auth * disable token refresh, remove old auth middleware * refactor client, add ApiErrorBoundary context * disable google and facebook strategies * fix: fix presets not displaying specific to user * fix: fix issue with browser refresh * fix: casing issue with User.js (#11) * delete user.js to be renamed * fix: fix casing issue with User.js * comment out api error watcher temporarily * feat: add google auth social login * fix: make google login url dynamic based on dev/prod * fix: bug where UI is briefly displayed before redirecting to login * fix: fix cookie expires value for local auth * Only load google strategy if client id and secret are provided * Update .env.example files with new params * fix issue with not redirecting to register form * only show google login button if value is set in .env * cleanup log messages * Add label to button for google login on login form * doc: fix client/server url values in .env.example * feat: add error message details to registration failure * Restore preventing paste on confirm password * auto-login user after registering * feat: forgot password (#24) * make login/reg pages look like openai's * add password reset data services * new form designs similar to openai, add password reset pages * add api's for password reset * email utils for password reset * remove bcrypt salt rounds from process.env * refactor: restructure api auth code, consolidate routes (#25) * add api's for password reset * remove bcrypt salt rounds from process.env * refactor: consolidate auth routes, use controller pattern * refactor: code cleanup * feat: migrate data to first user (#26) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes after refactor (#27) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: issue with auto-login when logging out then logging in with new browser window (#28) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: Update README and .env.example files with user system information (#29) * refactor: use /api for auth routes * fix: use user id instead of username * feat: migrate data to first user on register * fix: fix social login routes * fix: fix issue with auto-login in new tab * doc: update README and .env.example files * Send user id to openai to protect against abuse * add meilisearch to gitignore * Remove webpack --------- Co-authored-by: alfredo-f <[email protected]> --------- Co-authored-by: Danny Avila <[email protected]> Co-authored-by: Alfredo Fomitchenko <[email protected]>
danny-avila
added a commit
that referenced
this issue
Aug 29, 2024
* chore: copy assistant files * feat: frontend and data-provider * feat: backend get endpoint test * fix(MessageEndpointIcon): switched to AgentName and AgentAvatar * fix: small fixes * fix: agent endpoint config * fix: show Agent Builder * chore: install agentus * chore: initial scaffolding for agents * fix: updated Assistant logic to Agent Logic for some Agent components * WIP first pass, demo of agent package * WIP: initial backend infra for agents * fix: agent list error * wip: agents routing * chore: Refactor useSSE hook to handle different data events * wip: correctly emit events * chore: Update @librechat/agentus npm dependency to version 1.0.9 * remove comment * first pass: streaming agent text * chore: Remove @librechat/agentus root-level workspace npm dependency * feat: Agent Schema and Model * fix: content handling fixes * fix: content message save * WIP: new content data * fix: run step issue with tool calls * chore: Update @librechat/agentus npm dependency to version 1.1.5 * feat: update controller and agent routes * wip: initial backend tool and tool error handling support * wip: tool chunks * chore: Update @librechat/agentus npm dependency to version 1.1.7 * chore: update tool_call typing, add test conditions and logs * fix: create agent * fix: create agent * first pass: render completed content parts * fix: remove logging, fix step handler typing * chore: Update @librechat/agentus npm dependency to version 1.1.9 * refactor: cleanup maps on unmount * chore: Update BaseClient.js to safely count tokens for string, number, and boolean values * fix: support subsequent messages with tool_calls * chore: export order * fix: select agent * fix: tool call types and handling * chore: switch to anthropic for testing * fix: AgentSelect * refactor: experimental: OpenAIClient to use array for intermediateReply * fix(useSSE): revert old condition for streaming legacy client tokens * fix: lint * revert `agent_id` to `id` * chore: update localization keys for agent-related components * feat: zod schema handling for actions * refactor(actions): if no params, no zodSchema * chore: Update @librechat/agentus npm dependency to version 1.2.1 * feat: first pass, actions * refactor: empty schema for actions without params * feat: Update createRun function to accept additional options * fix: message payload formatting; feat: add more client options * fix: ToolCall component rendering when action has no args but has output * refactor(ToolCall): allow non-stringy args * WIP: first pass, correctly formatted tool_calls between providers * refactor: Remove duplicate import of 'roles' module * refactor: Exclude 'vite.config.ts' from TypeScript compilation * refactor: fix agent related types > - no need to use endpoint/model fields for identifying agent metadata > - add `provider` distinction for agent-configured 'endpoint' - no need for agent-endpoint map - reduce complexity of tools as functions into tools as string[] - fix types related to above changes - reduce unnecessary variables for queries/mutations and corresponding react-query keys * refactor: Add tools and tool_kwargs fields to agent schema * refactor: Remove unused code and update dependencies * refactor: Update updateAgentHandler to use req.body directly * refactor: Update AgentSelect component to use localized hooks * refactor: Update agent schema to include tools and provider fields * refactor(AgentPanel): add scrollbar gutter, add provider field to form, fix agent schema required values * refactor: Update AgentSwitcher component to use selectedAgentId instead of selectedAgent * refactor: Update AgentPanel component to include alternateName import and defaultAgentFormValues * refactor(SelectDropDown): allow setting value as option while still supporting legacy usage (string values only) * refactor: SelectDropdown changes - Only necessary when the available values are objects with label/value fields and the selected value is expected to be a string. * refactor: TypeError issues and handle provider as option * feat: Add placeholder for provider selection in AgentPanel component * refactor: Update agent schema to include author and provider fields * fix: show expected 'create agent' placeholder when creating agent * chore: fix localization strings, hide capabilities form for now * chore: typing * refactor: import order and use compact agents schema for now * chore: typing * refactor: Update AgentForm type to use AgentCapabilities * fix agent form agent selection issues * feat: responsive agent selection * fix: Handle cancelled fetch in useSelectAgent hook * fix: reset agent form on accordion close/open * feat: Add agent_id to default conversation for agents endpoint * feat: agents endpoint request handling * refactor: reset conversation model on agent select * refactor: add `additional_instructions` to conversation schema, organize other fields * chore: casing * chore: types * refactor(loadAgentTools): explicitly pass agent_id, do not pass `model` to loadAgentTools for now, load action sets by agent_id * WIP: initial draft of real agent client initialization * WIP: first pass, anthropic agent requests * feat: remember last selected agent * feat: openai and azure connected * fix: prioritize agent model for runs unless an explicit override model is passed from client * feat: Agent Actions * fix: save agent id to convo * feat: model panel (#29) * feat: model panel * bring back comments * fix: method still null * fix: AgentPanel FormContext * feat: add more parameters * fix: style issues; refactor: Agent Controller * fix: cherry-pick * fix: Update AgentAvatar component to use AssistantIcon instead of BrainCircuit * feat: OGDialog for delete agent; feat(assistant): update Agent types, introduced `model_parameters` * feat: icon and general `model_parameters` update * feat: use react-hook-form better * fix: agent builder form reset issue when switching panels * refactor: modularize agent builder form --------- Co-authored-by: Danny Avila <[email protected]> * fix: AgentPanel and ModelPanel type issues and use `useFormContext` and `watch` instead of `methods` directly and `useWatch`. * fix: tool call issues due to invalid input (anthropic) of empty string * fix: handle empty text in Part component --------- Co-authored-by: Marco Beretta <[email protected]>
danny-avila
added a commit
that referenced
this issue
Aug 31, 2024
* agents - phase 1 (#30) * chore: copy assistant files * feat: frontend and data-provider * feat: backend get endpoint test * fix(MessageEndpointIcon): switched to AgentName and AgentAvatar * fix: small fixes * fix: agent endpoint config * fix: show Agent Builder * chore: install agentus * chore: initial scaffolding for agents * fix: updated Assistant logic to Agent Logic for some Agent components * WIP first pass, demo of agent package * WIP: initial backend infra for agents * fix: agent list error * wip: agents routing * chore: Refactor useSSE hook to handle different data events * wip: correctly emit events * chore: Update @librechat/agentus npm dependency to version 1.0.9 * remove comment * first pass: streaming agent text * chore: Remove @librechat/agentus root-level workspace npm dependency * feat: Agent Schema and Model * fix: content handling fixes * fix: content message save * WIP: new content data * fix: run step issue with tool calls * chore: Update @librechat/agentus npm dependency to version 1.1.5 * feat: update controller and agent routes * wip: initial backend tool and tool error handling support * wip: tool chunks * chore: Update @librechat/agentus npm dependency to version 1.1.7 * chore: update tool_call typing, add test conditions and logs * fix: create agent * fix: create agent * first pass: render completed content parts * fix: remove logging, fix step handler typing * chore: Update @librechat/agentus npm dependency to version 1.1.9 * refactor: cleanup maps on unmount * chore: Update BaseClient.js to safely count tokens for string, number, and boolean values * fix: support subsequent messages with tool_calls * chore: export order * fix: select agent * fix: tool call types and handling * chore: switch to anthropic for testing * fix: AgentSelect * refactor: experimental: OpenAIClient to use array for intermediateReply * fix(useSSE): revert old condition for streaming legacy client tokens * fix: lint * revert `agent_id` to `id` * chore: update localization keys for agent-related components * feat: zod schema handling for actions * refactor(actions): if no params, no zodSchema * chore: Update @librechat/agentus npm dependency to version 1.2.1 * feat: first pass, actions * refactor: empty schema for actions without params * feat: Update createRun function to accept additional options * fix: message payload formatting; feat: add more client options * fix: ToolCall component rendering when action has no args but has output * refactor(ToolCall): allow non-stringy args * WIP: first pass, correctly formatted tool_calls between providers * refactor: Remove duplicate import of 'roles' module * refactor: Exclude 'vite.config.ts' from TypeScript compilation * refactor: fix agent related types > - no need to use endpoint/model fields for identifying agent metadata > - add `provider` distinction for agent-configured 'endpoint' - no need for agent-endpoint map - reduce complexity of tools as functions into tools as string[] - fix types related to above changes - reduce unnecessary variables for queries/mutations and corresponding react-query keys * refactor: Add tools and tool_kwargs fields to agent schema * refactor: Remove unused code and update dependencies * refactor: Update updateAgentHandler to use req.body directly * refactor: Update AgentSelect component to use localized hooks * refactor: Update agent schema to include tools and provider fields * refactor(AgentPanel): add scrollbar gutter, add provider field to form, fix agent schema required values * refactor: Update AgentSwitcher component to use selectedAgentId instead of selectedAgent * refactor: Update AgentPanel component to include alternateName import and defaultAgentFormValues * refactor(SelectDropDown): allow setting value as option while still supporting legacy usage (string values only) * refactor: SelectDropdown changes - Only necessary when the available values are objects with label/value fields and the selected value is expected to be a string. * refactor: TypeError issues and handle provider as option * feat: Add placeholder for provider selection in AgentPanel component * refactor: Update agent schema to include author and provider fields * fix: show expected 'create agent' placeholder when creating agent * chore: fix localization strings, hide capabilities form for now * chore: typing * refactor: import order and use compact agents schema for now * chore: typing * refactor: Update AgentForm type to use AgentCapabilities * fix agent form agent selection issues * feat: responsive agent selection * fix: Handle cancelled fetch in useSelectAgent hook * fix: reset agent form on accordion close/open * feat: Add agent_id to default conversation for agents endpoint * feat: agents endpoint request handling * refactor: reset conversation model on agent select * refactor: add `additional_instructions` to conversation schema, organize other fields * chore: casing * chore: types * refactor(loadAgentTools): explicitly pass agent_id, do not pass `model` to loadAgentTools for now, load action sets by agent_id * WIP: initial draft of real agent client initialization * WIP: first pass, anthropic agent requests * feat: remember last selected agent * feat: openai and azure connected * fix: prioritize agent model for runs unless an explicit override model is passed from client * feat: Agent Actions * fix: save agent id to convo * feat: model panel (#29) * feat: model panel * bring back comments * fix: method still null * fix: AgentPanel FormContext * feat: add more parameters * fix: style issues; refactor: Agent Controller * fix: cherry-pick * fix: Update AgentAvatar component to use AssistantIcon instead of BrainCircuit * feat: OGDialog for delete agent; feat(assistant): update Agent types, introduced `model_parameters` * feat: icon and general `model_parameters` update * feat: use react-hook-form better * fix: agent builder form reset issue when switching panels * refactor: modularize agent builder form --------- Co-authored-by: Danny Avila <[email protected]> * fix: AgentPanel and ModelPanel type issues and use `useFormContext` and `watch` instead of `methods` directly and `useWatch`. * fix: tool call issues due to invalid input (anthropic) of empty string * fix: handle empty text in Part component --------- Co-authored-by: Marco Beretta <[email protected]> * refactor: remove form ModelPanel and fixed nested ternary expressions in AgentConfig * fix: Model Parameters not saved correctly * refactor: remove console log * feat: avatar upload and get for Agents (#36) Co-authored-by: Marco Beretta <[email protected]> * chore: update to public package * fix: typing, optional chaining * fix: cursor not showing for content parts * chore: conditionally enable agents * ci: fix azure test * ci: fix frontend tests, fix eslint api * refactor: Remove unused errorContentPart variable * continue of the agent message PR (#40) * last fixes * fix: agentMap * pr merge test (#41) * fix: model icon not fetching correctly * remove console logs * feat: agent name * refactor: pass documentsMap as a prop to allow re-render of assistant form * refactor: pass documentsMap as a prop to allow re-render of assistant form * chore: Bump version to 0.7.419 * fix: TypeError: Cannot read properties of undefined (reading 'id') * refactor: update AgentSwitcher component to use ControlCombobox instead of Combobox --------- Co-authored-by: Marco Beretta <[email protected]>
BertKiv
pushed a commit
to BertKiv/LibreChat
that referenced
this issue
Dec 10, 2024
* agents - phase 1 (danny-avila#30) * chore: copy assistant files * feat: frontend and data-provider * feat: backend get endpoint test * fix(MessageEndpointIcon): switched to AgentName and AgentAvatar * fix: small fixes * fix: agent endpoint config * fix: show Agent Builder * chore: install agentus * chore: initial scaffolding for agents * fix: updated Assistant logic to Agent Logic for some Agent components * WIP first pass, demo of agent package * WIP: initial backend infra for agents * fix: agent list error * wip: agents routing * chore: Refactor useSSE hook to handle different data events * wip: correctly emit events * chore: Update @librechat/agentus npm dependency to version 1.0.9 * remove comment * first pass: streaming agent text * chore: Remove @librechat/agentus root-level workspace npm dependency * feat: Agent Schema and Model * fix: content handling fixes * fix: content message save * WIP: new content data * fix: run step issue with tool calls * chore: Update @librechat/agentus npm dependency to version 1.1.5 * feat: update controller and agent routes * wip: initial backend tool and tool error handling support * wip: tool chunks * chore: Update @librechat/agentus npm dependency to version 1.1.7 * chore: update tool_call typing, add test conditions and logs * fix: create agent * fix: create agent * first pass: render completed content parts * fix: remove logging, fix step handler typing * chore: Update @librechat/agentus npm dependency to version 1.1.9 * refactor: cleanup maps on unmount * chore: Update BaseClient.js to safely count tokens for string, number, and boolean values * fix: support subsequent messages with tool_calls * chore: export order * fix: select agent * fix: tool call types and handling * chore: switch to anthropic for testing * fix: AgentSelect * refactor: experimental: OpenAIClient to use array for intermediateReply * fix(useSSE): revert old condition for streaming legacy client tokens * fix: lint * revert `agent_id` to `id` * chore: update localization keys for agent-related components * feat: zod schema handling for actions * refactor(actions): if no params, no zodSchema * chore: Update @librechat/agentus npm dependency to version 1.2.1 * feat: first pass, actions * refactor: empty schema for actions without params * feat: Update createRun function to accept additional options * fix: message payload formatting; feat: add more client options * fix: ToolCall component rendering when action has no args but has output * refactor(ToolCall): allow non-stringy args * WIP: first pass, correctly formatted tool_calls between providers * refactor: Remove duplicate import of 'roles' module * refactor: Exclude 'vite.config.ts' from TypeScript compilation * refactor: fix agent related types > - no need to use endpoint/model fields for identifying agent metadata > - add `provider` distinction for agent-configured 'endpoint' - no need for agent-endpoint map - reduce complexity of tools as functions into tools as string[] - fix types related to above changes - reduce unnecessary variables for queries/mutations and corresponding react-query keys * refactor: Add tools and tool_kwargs fields to agent schema * refactor: Remove unused code and update dependencies * refactor: Update updateAgentHandler to use req.body directly * refactor: Update AgentSelect component to use localized hooks * refactor: Update agent schema to include tools and provider fields * refactor(AgentPanel): add scrollbar gutter, add provider field to form, fix agent schema required values * refactor: Update AgentSwitcher component to use selectedAgentId instead of selectedAgent * refactor: Update AgentPanel component to include alternateName import and defaultAgentFormValues * refactor(SelectDropDown): allow setting value as option while still supporting legacy usage (string values only) * refactor: SelectDropdown changes - Only necessary when the available values are objects with label/value fields and the selected value is expected to be a string. * refactor: TypeError issues and handle provider as option * feat: Add placeholder for provider selection in AgentPanel component * refactor: Update agent schema to include author and provider fields * fix: show expected 'create agent' placeholder when creating agent * chore: fix localization strings, hide capabilities form for now * chore: typing * refactor: import order and use compact agents schema for now * chore: typing * refactor: Update AgentForm type to use AgentCapabilities * fix agent form agent selection issues * feat: responsive agent selection * fix: Handle cancelled fetch in useSelectAgent hook * fix: reset agent form on accordion close/open * feat: Add agent_id to default conversation for agents endpoint * feat: agents endpoint request handling * refactor: reset conversation model on agent select * refactor: add `additional_instructions` to conversation schema, organize other fields * chore: casing * chore: types * refactor(loadAgentTools): explicitly pass agent_id, do not pass `model` to loadAgentTools for now, load action sets by agent_id * WIP: initial draft of real agent client initialization * WIP: first pass, anthropic agent requests * feat: remember last selected agent * feat: openai and azure connected * fix: prioritize agent model for runs unless an explicit override model is passed from client * feat: Agent Actions * fix: save agent id to convo * feat: model panel (danny-avila#29) * feat: model panel * bring back comments * fix: method still null * fix: AgentPanel FormContext * feat: add more parameters * fix: style issues; refactor: Agent Controller * fix: cherry-pick * fix: Update AgentAvatar component to use AssistantIcon instead of BrainCircuit * feat: OGDialog for delete agent; feat(assistant): update Agent types, introduced `model_parameters` * feat: icon and general `model_parameters` update * feat: use react-hook-form better * fix: agent builder form reset issue when switching panels * refactor: modularize agent builder form --------- Co-authored-by: Danny Avila <[email protected]> * fix: AgentPanel and ModelPanel type issues and use `useFormContext` and `watch` instead of `methods` directly and `useWatch`. * fix: tool call issues due to invalid input (anthropic) of empty string * fix: handle empty text in Part component --------- Co-authored-by: Marco Beretta <[email protected]> * refactor: remove form ModelPanel and fixed nested ternary expressions in AgentConfig * fix: Model Parameters not saved correctly * refactor: remove console log * feat: avatar upload and get for Agents (danny-avila#36) Co-authored-by: Marco Beretta <[email protected]> * chore: update to public package * fix: typing, optional chaining * fix: cursor not showing for content parts * chore: conditionally enable agents * ci: fix azure test * ci: fix frontend tests, fix eslint api * refactor: Remove unused errorContentPart variable * continue of the agent message PR (danny-avila#40) * last fixes * fix: agentMap * pr merge test (danny-avila#41) * fix: model icon not fetching correctly * remove console logs * feat: agent name * refactor: pass documentsMap as a prop to allow re-render of assistant form * refactor: pass documentsMap as a prop to allow re-render of assistant form * chore: Bump version to 0.7.419 * fix: TypeError: Cannot read properties of undefined (reading 'id') * refactor: update AgentSwitcher component to use ControlCombobox instead of Combobox --------- Co-authored-by: Marco Beretta <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
When using custom name for a customGPT, the CSS style displayed on the web interface is incorrect.
The text was updated successfully, but these errors were encountered: