🤖 refactor: auto-cleanup#3695
Conversation
|
@codex review |
|
To use Codex here, create a Codex account and connect to github. |
|
Root cause: The Verification: Ran Recommendation: Re-run the failed CI jobs. No code change needed. |
|
@codex review Latest push rebases onto |
|
To use Codex here, create a Codex account and connect to github. |
|
@codex review New in this run: cleanup #3 — deduped the identical blockquote line-prefixing ( |
|
To use Codex here, create a Codex account and connect to github. |
|
@codex review |
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
78cd7b2 to
a676f79
Compare
|
@codex review Added auto-cleanup #5: |
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
a676f79 to
32928e3
Compare
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
32928e3 to
8afce4c
Compare
|
@codex review |
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
8afce4c to
990576b
Compare
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
2 times, most recently
from
cd778d7 to
08734b9
Compare
|
@codex review This run adds one behavior-preserving cleanup (#8 in the branch list): dedupes the two identical |
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
08734b9 to
b42facd
Compare
|
@codex review |
|
To use Codex here, create a Codex account and connect to github. |
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
b42facd to
750c86e
Compare
|
@codex review |
|
To use Codex here, create a Codex account and connect to github. |
Extract the byte-identical consolidation/harvest sweep `recordUsage` callbacks in MemoryConsolidationService into a shared `makeSweepUsageRecorder` helper. Behavior-preserving: same sidecar recording and analyticsIngest emit; only the workspaceId source and analyticsSource literal differ per call site. Auto-cleanup checkpoint: f7f0f02
Both fallback branches in prepareToolSearch inlined the identical
{ [TOOL_SEARCH_TOOL_NAME]: _removed, ...rest } destructure to drop the
built-in tool_search entry from the record. Extract a module-level
withoutToolSearch(tools) helper; output is byte-identical.
accumulateProviderMetadata inlined the same verbose (metadata.anthropic as { cacheCreationInputTokens?: number }).cacheCreationInputTokens ?? 0 cast twice. Extracted a module-private getAnthropicCacheCreateTokens(metadata) helper. Behavior-preserving; the displayUsage.ts site is intentionally left alone because its chain has an extra usage.inputTokenDetails.cacheWriteTokens fallback.
Both getThinkingPolicyForModel and hasExplicitThinkingPolicy inlined the identical getExplicitThinkingPolicy(resolveModelForMetadata(model, providersConfig ?? null)) call after #3708 added alias resolution to each. Extract a private getExplicitThinkingPolicyForModel helper; behavior-preserving.
Both eligibility gates in openaiExplicitPromptCachingAvailable inlined the
identical split(":", 2) + `origin !== "openai" || !modelName` check (once for
the request model, once for the resolved capability target). The destructured
origin/name locals were unused past their guard. Extracted into a module-private
isOpenAIOriginModel(canonical) helper. Behavior-preserving.
renameLegacyToolSearchCallPart and renameLegacyToolSearchResultPart were byte-identical except for their part type. Collapse them into a single generic renameLegacyToolSearchPart<T extends { toolName: string }> and drop the now-unused ToolCallPart/ToolResultPart imports. Behavior-preserving.
mux-bot
Bot
force-pushed
the
auto-cleanup
branch
from
750c86e to
69a01a6
Compare
|
@codex review Added one behavior-preserving cleanup (#11): unified the byte-identical |
Summary
This is the long-lived auto-cleanup PR. Each run, the auto-cleanup agent reviews new commits merged to
main, rebases onto the latestmain, and applies at most one extremely low-risk, behavior-preserving cleanup. The branch accumulates a small stack of independent cleanups until it is merged.Cleanups in this branch
recordUsagecallbacks (MemoryConsolidationService). The consolidation sweep and the harvest sweep each inlined the same 15-line callback that routes billed usage to the headless-usage sidecar and emitsanalyticsIngest. Extracted into a privatemakeSweepUsageRecorder(...)helper.MemoryService). ThecreateandsaveFile(new-file) paths each inlined a byte-identical block that calledstore.listFiles(), compared the count againstMEMORY_MAX_FILES_PER_SCOPE, and threw aMemoryCommandErrorwith the same message. Extracted into a privateassertScopeHasRoom(store, scope)helper.bashMonitorWakeStore.ts).buildBashMonitorWakePromptrendered both the matched-output lines and the lost-monitor script with the identical.map((line) => \> ${line}`).join("\n")blockquote pattern in two places. Extracted into a module-levelblockquoteLines(lines)` helper.tool_searchremoval inprepareToolSearch(toolCatalog.ts). Both fallback branches (PTC enabled, and empty deferred catalog) inlined the identical{ [TOOL_SEARCH_TOOL_NAME]: _removed, ...rest }destructure to drop the built-intool_searchentry from the tool record. Extracted into a module-levelwithoutToolSearch(tools)helper.accumulateProviderMetadata(usageHelpers.ts). The function inlined the same verbose(metadata.anthropic as { cacheCreationInputTokens?: number } | undefined)?.cacheCreationInputTokens ?? 0cast twice (once for the accumulated metadata, once for the current step). Extracted into a module-privategetAnthropicCacheCreateTokens(metadata)helper.thinking/policy.ts). After 🤖 feat: integrate GPT-5.6 Sol/Terra/Luna with native max effort and pro-mode toggle #3708 taught the thinking policy to resolvemappedToModelaliases, bothgetThinkingPolicyForModelandhasExplicitThinkingPolicyinlined the identicalgetExplicitThinkingPolicy(resolveModelForMetadata(modelString, providersConfig ?? null))call. Extracted into a privategetExplicitThinkingPolicyForModel(modelString, providersConfig)helper.messageQueue.ts). After 🤖 feat: queue messages behind special sends instead of erroring (FIFO message queue) #3696 rewroteMessageQueueinto FIFOQueueEntryitems, bothgetClearCallbacksandremoveWorkspaceTurninlined the identical spread that builds aQueueClearCallbacksobject from an entry's optionalonCanceled/onAcceptedPreStreamFailurefields. Extracted into a privateentryClearCallbacks(entry)helper.openaiExplicitPromptCachingAvailable(cacheStrategy.ts). After 🤖 feat: GPT-5.6 explicit prompt cache breakpoints for direct OpenAI #3712 added the GPT-5.6 explicit-prompt-caching eligibility gate, the function inlined the identicalsplit(":", 2)+origin !== "openai" || !modelNamecheck twice — once for the request model and once for the resolved capability target — and the destructuredorigin/modelNamelocals were unused past their guard in both places. Extracted into a module-privateisOpenAIOriginModel(canonical)helper.tool-call-execution-startemit inStreamManager(streamManager.ts). 🤖 fix: start tool elapsed timers when execute() actually runs #3716 introduced theToolCallExecutionStartEvent, emitted from two places:applyToolExecutionStart(part already stored) and the"tool-call"case that consumes apendingExecutionStartrecorded before the part landed. Both inlined the byte-identicalthis.emit("tool-call-execution-start", { type, workspaceId, messageId, toolCallId, timestamp } satisfies ToolCallExecutionStartEvent)block, differing only in thetoolCallId/timestampsource. Extracted into a privateemitToolCallExecutionStart(workspaceId, streamInfo, toolCallId, timestamp)helper.aiService.ts). After 🤖 feat: apply mid-turn thinking-level changes at the next model step #3718 added mid-turn thinking-level rebuilds, the initial-model path and the fallback-model path each inlined a byte-identical closure (mergeModelParameterExtras/mergeNextModelParameterExtras) that folds providers.jsoncproviderExtrasUNDER the Mux-built provider-options namespace (short-circuiting when there are no extras, deep-merging viamergeProviderExtrasUnderMuxwhen the namespace is a plain object). The two differed only in the namespace key and the overrides source. Extracted into a module-levelmakeModelParameterExtrasMerger(namespaceKey, providerExtras)factory that returns the merger closure.tool_searchpart-rename helper (toolCatalog.ts). 🤖 fix: avoid OpenAI tool search name collision #3719 renamed the built-in tool-search tool totool_catalog_searchand added request-time rewriting of historicaltool_searchcall/result parts. It introduced two byte-identical helpers —renameLegacyToolSearchCallPart(part: ToolCallPart)andrenameLegacyToolSearchResultPart(part: ToolResultPart)— that differ only in the part type; the rename body is identical. Collapsed both into a single genericrenameLegacyToolSearchPart<T extends { toolName: string }>(part: T)and dropped the now-unusedToolCallPart/ToolResultPartimports.This run
Considered
origin/mainthroughc18a8150e, which added two commits since the last checkpoint: avoid OpenAI tool search name collision (#3719) and boundtask_terminatewith abort support, timeouts, and per-task errors (#3720).normalizeLegacyToolSearchMessagespath fans out to two rename helpers that are identical except for theirToolCallPart/ToolResultPartargument type. Both call sites pass a part that carriestoolName, so a single genericrenameLegacyToolSearchPart<T extends { toolName: string }>preserves each call site's exact return type while removing the duplication (and the two type imports it required). Verified type-safe under bothtsconfigconfigs.raceWithAbortAndTimeouthelper, the bounded-stopStreamblocks intaskService/workspaceService, and the git-timeout plumbing are each single-purpose and already well-factored; the two bounded-stopStream blocks look similar but live in different classes with divergent downstream handling, so deduping them would be a cross-file change beyond the "extremely low-risk, local" bar for this agent.Validation
origin/main(c18a8150e); no conflicts.make static-checkpasses in full (ESLint, both TypeScript configs, Prettier, shell/Python formatting, shellcheck, hadolint, and code-to-docs link checks).bun test src/common/utils/tools/toolCatalog.test.ts— 32 pass / 0 fail, includingnormalizeLegacyToolSearchMessages > rewrites completed Mux search history without relabeling unrelated tools, which exercises the unified rename helper.Risks
Very low. All cleanups are pure same-file simplifications that preserve semantics. Every cleanup extracts byte-identical inlined logic into a helper — no control flow, arguments, thresholds, error types, emitted events, or rendered text change.
Auto-cleanup checkpoint: c18a815
Generated with
mux• Model:anthropic:claude-opus-4-8• Thinking:xhigh• Cost:$0.00