-
Notifications
You must be signed in to change notification settings - Fork 94
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
Pasting LLxprt's own terminal output into the prompt causes runaway memory allocation (Windows/PowerShell)
ToolingIssues related to tools provided to the model or tool execution (scheduler etc).Issues related to tools provided to the model or tool execution (scheduler etc).windowsWindows specific issuesWindows specific issuesStatus: Open.Bound ast_read_file native memory use and remove unused repository scans
ToolingIssues related to tools provided to the model or tool execution (scheduler etc).Issues related to tools provided to the model or tool execution (scheduler etc).Status: Open.Finish interface-neutral Agent control surface and remove CLI/A2A runtime reach-through
APIAPIs and things to plugin and extend LLxprt CodeAPIs and things to plugin and extend LLxprt CodearchitectureArchitectural refactor / clean designArchitectural refactor / clean designCode Quality / ModularizationIssues to do with the quality of llxprt code and its maintainability.Issues to do with the quality of llxprt code and its maintainability.configurationllxprt system configurationllxprt system configurationStatus: Open.Make Agent API runtime assembly self-contained and delete provider-held agent factory registration
APIAPIs and things to plugin and extend LLxprt CodeAPIs and things to plugin and extend LLxprt CodearchitectureArchitectural refactor / clean designArchitectural refactor / clean designCode Quality / ModularizationIssues to do with the quality of llxprt code and its maintainability.Issues to do with the quality of llxprt code and its maintainability.packages/coreIssues related to packages/core componentIssues related to packages/core componentProvider Supportthings that support an inference provider but not a particular modelthings that support an inference provider but not a particular modelsubagentsasync / synch / other types of subagents.async / synch / other types of subagents.Status: Open.Stateful OpenAI Responses enforcement counts only transport delta, allowing retained context overflow and futile retries
codex / GPTrelated to codex auth and GPT model support.related to codex auth and GPT model support.Context ManagementIssues related to managing the context sent to the model.Issues related to managing the context sent to the model.Provider Supportthings that support an inference provider but not a particular modelthings that support an inference provider but not a particular modelrecoverabilityIssues related to being able to continue an implementation or session if something goes wrong.Issues related to being able to continue an implementation or session if something goes wrong.subagentsasync / synch / other types of subagents.async / synch / other types of subagents.Status: Open.#3219 In vybestack/llxprt-code;Move image history out of process memory with cache-aware media lifecycle
architectureArchitectural refactor / clean designArchitectural refactor / clean designContext ManagementIssues related to managing the context sent to the model.Issues related to managing the context sent to the model.Status: Open.Provider history normalization wipes tool-turn metadata, silently defeating the #3070 cache anchor
Context ManagementIssues related to managing the context sent to the model.Issues related to managing the context sent to the model.packages/coreIssues related to packages/core componentIssues related to packages/core componentStatus: Open.Prime the Codex stateful chain on resume and after compression instead of spending a full-history request
codex / GPTrelated to codex auth and GPT model support.related to codex auth and GPT model support.Context ManagementIssues related to managing the context sent to the model.Issues related to managing the context sent to the model.Provider Supportthings that support an inference provider but not a particular modelthings that support an inference provider but not a particular modelStatus: Open.dumpcontext fabricates request headers, method and URL, hiding which transport was used
ObservabilityIssues related to quotas being able to observe model or provider responses or behavior.Issues related to quotas being able to observe model or provider responses or behavior.Status: Open.Nightly workflow failed
ci/cdIssues with github and the workflow scripts and CI CD environment.Issues with github and the workflow scripts and CI CD environment.Status: Open.#3153 In vybestack/llxprt-code;Remove the vestigial gemini-cli Declarative Agent Framework (AgentExecutor/SubagentInvocation)
Code Quality / ModularizationIssues to do with the quality of llxprt code and its maintainability.Issues to do with the quality of llxprt code and its maintainability.Status: Open.Nightly: shard the Windows and macOS test jobs to cut wall clock from ~53 min to under 30
ci/cdIssues with github and the workflow scripts and CI CD environment.Issues with github and the workflow scripts and CI CD environment.Status: Open.