pi-cache-optimizer 2.8.1 → 2.8.3

package/README.md

@@ -55,7 +55,7 @@ Run `/reload` in Pi after install/update/remove so extension hooks refresh.
 
 On Pi 0.79.7 and newer, `pi update` updates Pi itself only. To update installed Pi packages such as this extension, run `pi update --extensions` (packages only) or `pi update --all` (Pi + packages).
 
-This extension is validated against Pi 0.83.0 and remains designed for Pi 0.82+. It uses the existing extension hooks, `getAgentDir()`, and prompt options shared by those versions; it does not depend on Pi 0.83-only APIs such as `ctx.scopedModels` or the bundled TypeBox 1.3 aliases.
+This extension requires Pi 0.82+ and is validated against Pi 0.84.2. It uses the official Pi package types directly for type-checking, along with extension hooks, `getAgentDir()`, and prompt options shared by those versions; it does not depend on Pi 0.83+ APIs such as `ctx.scopedModels` or the bundled TypeBox 1.3 aliases.
 
 ## Commands
 
@@ -109,7 +109,7 @@ The explicit setting is stored in `pi-cache-optimizer-config.json` under Pi's ag
 
 Third-party `openai-completions` proxies (LiteLLM / OneAPI / NewAPI / OpenRouter-like channels) often route one session across multiple upstream backends. That splits provider-side prompt caches.
 
-Pi 0.81+ also has a built-in `llama.cpp` provider using an OpenAI-shaped transport. Pi 0.82+ core generates a session `prompt_cache_key` for it when cache retention is enabled, so this extension preserves that key and may add the same conservative fallback when missing. The built-in provider's explicit compat fingerprint is excluded from generic proxy routing/session-affinity advice, but a custom or overridden provider that merely reuses the id `llama.cpp` is treated like any other OpenAI-compatible channel. `prompt_cache_retention` remains subject to the normal safety rule: keep it only for official OpenAI or an explicit effective `supportsLongCacheRetention: true` opt-in in `models.json`; otherwise strip it before sending. Effective values follow Pi's precedence: `modelOverrides[modelId].compat` first, then the matching `models[].compat`, then provider-level `compat`. An explicit `false` at a higher layer overrides `true` below it.
+Pi 0.84.1 also fixes built-in Fireworks compatibility for models that reject `prompt_cache_retention`; keep the extension's diagnostics focused on the effective model compat exposed by Pi rather than adding a provider-name special case. Pi 0.81+ also has a built-in `llama.cpp` provider using an OpenAI-shaped transport. Pi 0.82+ core generates a session `prompt_cache_key` for it when cache retention is enabled, so this extension preserves that key and may add the same conservative fallback when missing. The built-in provider's explicit compat fingerprint is excluded from generic proxy routing/session-affinity advice, but a custom or overridden provider that merely reuses the id `llama.cpp` is treated like any other OpenAI-compatible channel. `prompt_cache_retention` remains subject to the normal safety rule: keep it only for official OpenAI or an explicit effective `supportsLongCacheRetention: true` opt-in in `models.json`; otherwise strip it before sending. Effective values follow Pi's precedence: `modelOverrides[modelId].compat` first, then the matching `models[].compat`, then provider-level `compat`. An explicit `false` at a higher layer overrides `true` below it.
 
 For real proxies, start with session affinity:
 
@@ -135,7 +135,7 @@ Notes:
 
 - `sendSessionAffinityHeaders: true` is the safe default when your proxy supports sticky routing.
 - `supportsLongCacheRetention: true` is optional. Add it only when the endpoint explicitly supports OpenAI long prompt cache retention.
-- If you see `400 Unsupported parameter: prompt_cache_retention`, remove/avoid `supportsLongCacheRetention` for that channel. Keep `sendSessionAffinityHeaders` if supported.
+- If you see `400 Unsupported parameter: prompt_cache_retention`, remove/avoid `supportsLongCacheRetention` for that channel. Keep `sendSessionAffinityHeaders` if supported. The extension detects the explicit error from response headers or the finalized assistant error message and strips the parameter from subsequent requests in the current process.
 - Use `/cache-optimizer compat` or `/cache-optimizer doctor` to see model-specific advice.
 - For DeepSeek models, the Pi Mono guidance expects `compat.requiresReasoningContentOnAssistantMessages: true` and `compat.thinkingFormat: "deepseek"` alongside cache/session-affinity flags when the endpoint supports them.
 - This extension's `doctor` and `compat` commands only advise; they do not modify `models.json`.
@@ -150,7 +150,7 @@ Some proxies rewrite or insert hidden 5-minute breakpoints after Pi's request ho
 
 ## Adaptive thinking models
 
-Claude models from opus-4.6 / sonnet-4.6 (including Opus 5 and Sonnet 5) / fable-5 onwards require `forceAdaptiveThinking: true` in compat. Kimi Coding K3 (`k3`) and `kimi-for-coding` also use adaptive thinking and need `allowEmptySignature: true` so replayed empty-signature thinking blocks remain valid. Without the required compat, Pi may send a legacy thinking payload or replay thinking incorrectly. Pi 0.83.0's native Opus 5 catalog is covered by the same adaptive-thinking detection; custom `anthropic-messages` channels still need the compat flag when Pi does not provide it.
+Claude models from opus-4.6 / sonnet-4.6 (including Opus 5 and Sonnet 5) / fable-5 onwards require `forceAdaptiveThinking: true` in compat. Kimi Coding K3 (`k3`) and `kimi-for-coding` also use adaptive thinking and need `allowEmptySignature: true` so replayed empty-signature thinking blocks remain valid. Without the required compat, Pi may send a legacy thinking payload or replay thinking incorrectly. Pi 0.83+ native Opus 5 catalogs are covered by the same adaptive-thinking detection; custom `anthropic-messages` channels still need the compat flag when Pi does not provide it.
 
 Pi's built-in catalog already sets this flag for official models. Custom channels in `models.json` that override these models must include the flag:
 
@@ -229,8 +229,9 @@ Pi 0.80.9+ already includes Kimi K3 in built-in Kimi Coding, Moonshot AI / China
 2. Warns: ① changes affect all sessions using that channel, ② automatic backup created at `models.json.backup-cache-optimizer-<timestamp>`, ③ Pi reload required
 3. Uses comment-preserving surgical editor — existing comments, indentation, and existing key order are preserved
 4. Requires explicit user confirmation (interactive prompt or `ui.select`)
-5. Writes atomically (temp + rename); self-validates after write
-6. Falls back to manual guidance if JSONC scanner cannot confidently locate the target
+5. Writes and restores atomically (temp + rename); self-validates after write
+6. Preserves the existing `models.json` access mode exactly — it does not tighten or loosen permissions (for example, `0600` stays `0600`, `0644` stays `0644`)
+7. Uses unique, non-overwriting backup names and falls back to manual guidance if the JSONC scanner cannot confidently locate the target
 
 Existing `modelOverrides[modelId]` entries have Pi's highest precedence, so `fix` repairs them directly. For built-in or API-login models without a custom `models[]` entry, `fix` creates a compat-only `modelOverrides` entry instead of inventing a custom model definition. Self-validation checks the effective three-layer compat result, so a lower-level edit shadowed by an override is rejected.
 
@@ -283,10 +284,10 @@ Pi 0.79+ also includes a built-in footer `CH` marker for the latest prompt cache
 Example footer:
 
 ```text
-· OpenAI cache 3/10 · 0.002M/0.005M tok (40%) ⚠️ compat
+· OpenAI cache 3/10·0.002M/0.005M 40.0% ⚠️ compat
 ```
 
-The leading `· ` is owned by this extension and separates its status from statuses published by other extensions in the same footer. It is present for normal, disabled, router-restored, and warning-suffixed statuses. The stats format remains `<label> <hit requests>/<total requests> · <cached input tokens>/<total input tokens> tok (<token hit rate>)`; `/cache-optimizer stats` output is unchanged. Some adapters may also append `· write <tokens> tok`, and runtime diagnostics may append `⚠️ compat` or `⚠️ integrity`.
+The leading `· ` is owned by this extension and separates its status from statuses published by other extensions in the same footer. It is present for normal, disabled, router-restored, and warning-suffixed statuses. The compact footer format is `<label> <hit requests>/<total requests>·<cached input tokens>/<total input tokens> <token hit rate>`; token hit rate keeps one decimal place and the footer omits the redundant `tok` suffix. `/cache-optimizer stats` output is unchanged. Some adapters may also append `·write <tokens>`, and runtime diagnostics may append `⚠️ compat` or `⚠️ integrity`.
 
 Supported footer labels include: DS, Claude, OpenAI, Gemini, Kimi, Qwen, GLM, MiniMax, Mimo, Hunyuan, Mistral, Grok, Llama, Nemotron, Cohere, Yi, Doubao, ERNIE, Baichuan, StepFun, Spark, InternLM, Gemma, Phi, Jamba, Solar, Sonar, Nova, Reka, Falcon, DBRX, MPT, StableLM, Aquila, EXAONE, HyperCLOVA, Luminous, Hermes, Granite, Arctic, Pangu, SenseNova, Zhinao, MiniCPM, XVERSE, Orion, OpenChat, Vicuna, Wizard, Zephyr, Dolphin, OpenOrca, Starling, BLOOM, RWKV, and Aya.
 

package/README.zh-CN.md

@@ -55,7 +55,7 @@ pi remove npm:pi-deepseek-cache-optimizer && pi install npm:pi-cache-optimizer
 
 Pi 0.79.7 及之后，`pi update` 默认只更新 Pi 本体。若要更新已安装的 Pi package（包括本扩展），请运行 `pi update --extensions`（只更新 packages）或 `pi update --all`（Pi 与 packages 一起更新）。
 
-本扩展已使用 Pi 0.83.0 验证，并继续面向 Pi 0.82+ 设计。它使用这些版本共有的 extension hooks、`getAgentDir()` 和 prompt options，不依赖 Pi 0.83 专有 API（例如 `ctx.scopedModels` 或 bundled TypeBox 1.3 aliases）。
+本扩展要求 Pi 0.82+，并已使用 Pi 0.84.2 验证。TypeScript 校验直接使用官方 Pi package 类型，同时只使用这些版本共有的 extension hooks、`getAgentDir()` 和 prompt options；不依赖 Pi 0.83+ 专有 API（例如 `ctx.scopedModels` 或 bundled TypeBox 1.3 aliases）。
 
 ## 命令
 
@@ -108,7 +108,7 @@ Pi 0.79.7 及之后，`pi update` 默认只更新 Pi 本体。若要更新已安
 
 LiteLLM / OneAPI / NewAPI / 类 OpenRouter 渠道等第三方 `openai-completions` 代理，常会把同一个 session 分散到多个上游后端，导致 provider 侧 prompt cache 被拆散。
 
-Pi 0.81+ 也内置了使用 OpenAI-shaped transport 的 `llama.cpp` provider。Pi 0.82+ core 在启用 cache retention 时会为它生成 session `prompt_cache_key`，因此本扩展会保留该 key，并在缺失时使用同样的保守 fallback。只有符合 Pi 内置 provider 明确 compat 指纹的模型会跳过通用 proxy 路由 / session-affinity 建议；仅复用 `llama.cpp` id 的自定义或覆盖 provider 仍按普通 OpenAI-compatible 渠道处理。`prompt_cache_retention` 继续遵循统一安全规则：仅官方 OpenAI 或 `models.json` 中有效配置为 `supportsLongCacheRetention: true` 时保留，否则发送前移除。有效值遵循 Pi 的优先级：先看 `modelOverrides[modelId].compat`，再看匹配的 `models[].compat`，最后看 provider 级 `compat`；高层显式 `false` 会覆盖低层的 `true`。
+Pi 0.84.1 还修复了内置 Fireworks 渠道对拒绝 `prompt_cache_retention` 的模型兼容性；本扩展继续依据 Pi 暴露的有效 model compat 做诊断，不按 provider 名称增加特殊分支。Pi 0.81+ 也内置了使用 OpenAI-shaped transport 的 `llama.cpp` provider。Pi 0.82+ core 在启用 cache retention 时会为它生成 session `prompt_cache_key`，因此本扩展会保留该 key，并在缺失时使用同样的保守 fallback。只有符合 Pi 内置 provider 明确 compat 指纹的模型会跳过通用 proxy 路由 / session-affinity 建议；仅复用 `llama.cpp` id 的自定义或覆盖 provider 仍按普通 OpenAI-compatible 渠道处理。`prompt_cache_retention` 继续遵循统一安全规则：仅官方 OpenAI 或 `models.json` 中有效配置为 `supportsLongCacheRetention: true` 时保留，否则发送前移除。有效值遵循 Pi 的优先级：先看 `modelOverrides[modelId].compat`，再看匹配的 `models[].compat`，最后看 provider 级 `compat`；高层显式 `false` 会覆盖低层的 `true`。
 
 对真正的代理，建议先启用 session affinity：
 
@@ -134,7 +134,7 @@ Pi 0.81+ 也内置了使用 OpenAI-shaped transport 的 `llama.cpp` provider。P
 
 - `sendSessionAffinityHeaders: true` 是安全默认项，前提是你的代理支持 sticky routing。
 - `supportsLongCacheRetention: true` 是可选项。只有 endpoint 明确支持 OpenAI long prompt cache retention 时才添加。
-- 如果出现 `400 Unsupported parameter: prompt_cache_retention`，请为该渠道移除 / 避免 `supportsLongCacheRetention`；如支持，可保留 `sendSessionAffinityHeaders`。
+- 如果出现 `400 Unsupported parameter: prompt_cache_retention`，请为该渠道移除 / 避免 `supportsLongCacheRetention`；如支持，可保留 `sendSessionAffinityHeaders`。扩展会从响应头或最终 assistant error message 中识别这条明确错误，并在当前进程的后续请求中移除该参数。
 - 使用 `/cache-optimizer compat` 或 `/cache-optimizer doctor` 查看当前模型的具体建议。
 - 对 DeepSeek 模型，Pi Mono 指南期望在支持时同时设置 `compat.requiresReasoningContentOnAssistantMessages: true` 和 `compat.thinkingFormat: "deepseek"`，再配合缓存 / session-affinity 相关 compat。
 - 本扩展的 `doctor` 和 `compat` 命令只给建议，不会修改 `models.json`。
@@ -149,7 +149,7 @@ Anthropic 按 `tools → system → messages` 顺序处理 cache breakpoint，
 
 ## Adaptive thinking 模型
 
-Claude 从 opus-4.6 / sonnet-4.6（含 Opus 5、Sonnet 5）/ fable-5 开始需要在 compat 中设置 `forceAdaptiveThinking: true`。Kimi Coding K3（`k3`）和 `kimi-for-coding` 也使用 adaptive thinking，并需要 `allowEmptySignature: true`，以正确重放空 signature 的 thinking block。缺少这些 compat 时，Pi 可能发送旧版 thinking payload 或错误重放 thinking。Pi 0.83.0 的原生 Opus 5 catalog 已覆盖在同一 adaptive-thinking 检测中；如果自定义 `anthropic-messages` 渠道没有继承该 compat，仍需手动设置。
+Claude 从 opus-4.6 / sonnet-4.6（含 Opus 5、Sonnet 5）/ fable-5 开始需要在 compat 中设置 `forceAdaptiveThinking: true`。Kimi Coding K3（`k3`）和 `kimi-for-coding` 也使用 adaptive thinking，并需要 `allowEmptySignature: true`，以正确重放空 signature 的 thinking block。缺少这些 compat 时，Pi 可能发送旧版 thinking payload 或错误重放 thinking。Pi 0.83+ 的原生 Opus 5 catalog 已覆盖在同一 adaptive-thinking 检测中；如果自定义 `anthropic-messages` 渠道没有继承该 compat，仍需手动设置。
 
 Pi 内置 catalog 已为官方模型设置此 flag。`models.json` 中覆盖这些模型的自定义渠道必须包含该 flag：
 
@@ -228,8 +228,9 @@ Pi 0.80.9+ 已在内置 Kimi Coding、Moonshot AI / 中国区、OpenRouter 和 V
 2. 警告：① 修改影响使用该渠道的所有 session，② 自动备份到 `models.json.backup-cache-optimizer-<timestamp>`，③ 需重启 Pi 或 reload
 3. 使用保留注释的精确编辑器 —— 现有注释、缩进和已有 key 顺序都会保留
 4. 需要用户明确确认（交互式提示或 `ui.select`）
-5. 原子写入（temp + rename）；写入后自我验证
-6. 如果 JSONC 扫描器无法置信定位目标，回退到手动修改指引
+5. 写入与失败恢复都使用原子替换（temp + rename）；写入后自我验证
+6. 完全保留 `models.json` 原有访问权限，不主动收紧或放宽（例如 `0600` 保持 `0600`，`0644` 保持 `0644`）
+7. 备份名唯一且不会覆盖已有备份；如果 JSONC 扫描器无法置信定位目标，则回退到手动修改指引
 
 已有的 `modelOverrides[modelId]` 具有 Pi 的最高优先级，因此 `fix` 会直接修复该 entry。对于没有自定义 `models[]` entry 的内置模型或 API-login 模型，`fix` 会创建仅含 compat 的 `modelOverrides` entry，而不会凭空添加自定义模型定义。自检会验证三层合并后的有效 compat，因此被更高层 override 遮蔽的低层修改会被拒绝。
 
@@ -282,10 +283,10 @@ Pi 0.79+ 已内置 footer `CH` 标记，用于显示最近一次 prompt cache hi
 示例 footer：
 
 ```text
-· OpenAI cache 3/10 · 0.002M/0.005M tok (40%) ⚠️ compat
+· OpenAI cache 3/10·0.002M/0.005M 40.0% ⚠️ compat
 ```
 
-开头的 `· ` 由本扩展负责，用于把本扩展状态与同一 footer 中其他扩展发布的状态隔开。普通、disabled、router 恢复以及带 warning 的状态都会保留此前缀。统计格式仍是 `<label> <命中请求数>/<总请求数> · <cached input tokens>/<total input tokens> tok (<token 命中率>)`；`/cache-optimizer stats` 的输出不变。部分 adapter 还可能追加 `· write <tokens> tok`，运行时诊断可能追加 `⚠️ compat` 或 `⚠️ integrity`。
+开头的 `· ` 由本扩展负责，用于把本扩展状态与同一 footer 中其他扩展发布的状态隔开。普通、disabled、router 恢复以及带 warning 的状态都会保留此前缀。紧凑 footer 格式为 `<label> <命中请求数>/<总请求数>·<cached input tokens>/<total input tokens> <token 命中率>`；token 命中率保留一位小数，并去掉多余的 `tok` 后缀。`/cache-optimizer stats` 的输出不变。部分 adapter 还可能追加 `·write <tokens>`，运行时诊断可能追加 `⚠️ compat` 或 `⚠️ integrity`。
 
 支持的 footer label 包括：DS、Claude、OpenAI、Gemini、Kimi、Qwen、GLM、MiniMax、Mimo、Hunyuan、Mistral、Grok、Llama、Nemotron、Cohere、Yi、Doubao、ERNIE、Baichuan、StepFun、Spark、InternLM、Gemma、Phi、Jamba、Solar、Sonar、Nova、Reka、Falcon、DBRX、MPT、StableLM、Aquila、EXAONE、HyperCLOVA、Luminous、Hermes、Granite、Arctic、Pangu、SenseNova、Zhinao、MiniCPM、XVERSE、Orion、OpenChat、Vicuna、Wizard、Zephyr、Dolphin、OpenOrca、Starling、BLOOM、RWKV、Aya。
 

package/index.ts

@@ -1,9 +1,15 @@
 import { createHash } from "node:crypto";
-import { copyFile, mkdir, readFile, rename, unlink, writeFile } from "node:fs/promises";
-import { readFileSync } from "node:fs";
+import { chmod, copyFile, mkdir, readFile, rename, stat, unlink, writeFile } from "node:fs/promises";
+import { constants as fsConstants, readFileSync } from "node:fs";
 import { homedir } from "node:os";
 import { dirname, join } from "node:path";
-import { getAgentDir, type BuildSystemPromptOptions, type ExtensionAPI, type ExtensionContext } from "@earendil-works/pi-coding-agent";
+import {
+  getAgentDir,
+  type BuildSystemPromptOptions,
+  type ExtensionAPI,
+  type ExtensionCommandContext,
+  type ExtensionContext,
+} from "@earendil-works/pi-coding-agent";
 
 type MutableEnv = Record<string, string | undefined>;
 
@@ -85,6 +91,7 @@ const STARTUP_CACHE_RETENTION_ENV = getOrCaptureCacheRetentionBaseline();
 requestLongCacheRetention();
 
 type PiModel = NonNullable<ExtensionContext["model"]>;
+type ModelIdentity = Pick<PiModel, "provider" | "id">;
 type UnknownRecord = Record<string, unknown>;
 type CacheProviderId = "deepseek" | "openai" | "claude" | "gemini";
 
@@ -2088,7 +2095,7 @@ function isAyaLikeAssistantMessage(message: unknown, model: PiModel | undefined)
 
 // ── Model key ──────────────────────────────────────────────────────
 
-function modelKey(model: PiModel): string {
+function modelKey(model: ModelIdentity): string {
   return `${model.provider}/${model.id}`;
 }
 
@@ -2461,12 +2468,8 @@ function getPromptCacheRetentionUnsupportedHint(): string {
   return "If this channel returns `400 Unsupported parameter: prompt_cache_retention`, remove/avoid `supportsLongCacheRetention`; this extension does not write that field directly, but Pi may send it when long retention is requested and compat says the proxy supports it.";
 }
 
-function hasPromptCacheRetentionUnsupportedSignal(headers: Record<string, string> | undefined): boolean {
-  if (!headers) return false;
-
-  const normalized = Object.entries(headers)
-    .map(([key, value]) => `${lower(key)}: ${lower(value)}`)
-    .join("\n");
+function hasPromptCacheRetentionUnsupportedText(value: unknown): boolean {
+  const normalized = lower(value);
   if (!normalized.includes("prompt_cache_retention")) return false;
 
   return [
@@ -2478,10 +2481,24 @@ function hasPromptCacheRetentionUnsupportedSignal(headers: Record<string, string
     "extra inputs",
     "not permitted",
     "unrecognized",
-    "bad request",
   ].some((needle) => normalized.includes(needle));
 }
 
+function hasPromptCacheRetentionUnsupportedSignal(headers: Record<string, string> | undefined): boolean {
+  if (!headers) return false;
+  return hasPromptCacheRetentionUnsupportedText(
+    Object.entries(headers)
+      .map(([key, value]) => `${key}: ${value}`)
+      .join("\n"),
+  );
+}
+
+function hasPromptCacheRetentionUnsupportedErrorMessage(message: unknown): boolean {
+  const record = getAssistantRecord(message);
+  return record?.stopReason === "error" &&
+    hasPromptCacheRetentionUnsupportedText(record.errorMessage);
+}
+
 type CompatAdvicePlacement = {
   providerLabel?: string;
   modelId?: string;
@@ -3733,13 +3750,13 @@ function formatTokenCount(value: number): string {
 
 function formatCacheStats(adapter: CacheProviderAdapter, stats: CacheStats): string {
   const percent = stats.totalInputTokens > 0
-    ? ` (${Math.round((stats.cachedInputTokens / stats.totalInputTokens) * 100)}%)`
-    : "";
+    ? (stats.cachedInputTokens / stats.totalInputTokens) * 100
+    : 0;
   const writeText = adapter.showCacheWrite && stats.cacheWriteInputTokens > 0
-    ? ` · write ${formatTokenCount(stats.cacheWriteInputTokens)} tok`
+    ? `·write ${formatTokenCount(stats.cacheWriteInputTokens)}`
     : "";
 
-  return `${adapter.label} ${stats.hitRequests}/${stats.totalRequests} · ${formatTokenCount(stats.cachedInputTokens)}/${formatTokenCount(stats.totalInputTokens)} tok${percent}${writeText}`;
+  return `${adapter.label} ${stats.hitRequests}/${stats.totalRequests}·${formatTokenCount(stats.cachedInputTokens)}/${formatTokenCount(stats.totalInputTokens)} ${percent.toFixed(1)}%${writeText}`;
 }
 
 function prefixFooterStatus(statusText: string | undefined): string | undefined {
@@ -4405,6 +4422,16 @@ function isPromptCacheRetention400Applicable(model: PiModel): boolean {
     getCompat(model).supportsLongCacheRetention === true;
 }
 
+function isExplicitPromptCacheRetentionUnsupportedApplicable(model: PiModel): boolean {
+  // A finalized assistant error with an explicit unsupported-parameter signal
+  // proves that prompt_cache_retention reached this provider/model. Do not
+  // require compat inherited from the active fallback model: router shells may
+  // have no upstream compat metadata when no live routing registry is present.
+  return isOpenAICompatibleApi(model.api) &&
+    !isOfficialOpenAIBaseUrl(model) &&
+    !isPiBuiltInLlamaCppModel(model);
+}
+
 /**
  * Whether the 403 sendSessionAffinityHeaders diagnostic applies to a model.
  *
@@ -6528,18 +6555,121 @@ function formatCompatKeysForInsertion(compatKeys: Record<string, unknown>): stri
 }
 
 /**
- * Generate the timestamp string for backup filename.
- * Format: YYYYMMDDTHHMMSSZ (UTC)
+ * Generate a unique UTC timestamp component for backup filenames.
+ * Milliseconds, process id, and an in-process sequence prevent collisions.
  */
-function backupTimestamp(): string {
-  const now = new Date();
+let backupSequence = 0;
+
+function backupTimestamp(now: Date = new Date()): string {
   const y = now.getUTCFullYear();
-  const m = String(now.getUTCMonth() + 1).padStart(2, '0');
-  const d = String(now.getUTCDate()).padStart(2, '0');
-  const h = String(now.getUTCHours()).padStart(2, '0');
-  const min = String(now.getUTCMinutes()).padStart(2, '0');
-  const s = String(now.getUTCSeconds()).padStart(2, '0');
-  return `${y}${m}${d}T${h}${min}${s}Z`;
+  const m = String(now.getUTCMonth() + 1).padStart(2, "0");
+  const d = String(now.getUTCDate()).padStart(2, "0");
+  const h = String(now.getUTCHours()).padStart(2, "0");
+  const min = String(now.getUTCMinutes()).padStart(2, "0");
+  const s = String(now.getUTCSeconds()).padStart(2, "0");
+  const ms = String(now.getUTCMilliseconds()).padStart(3, "0");
+  return `${y}${m}${d}T${h}${min}${s}${ms}Z-${process.pid}-${backupSequence++}`;
+}
+
+function uniqueTempPath(targetPath: string, purpose: string): string {
+  return `${targetPath}.${process.pid}.${Date.now()}.${backupSequence++}.${purpose}.tmp`;
+}
+
+async function atomicReplaceTextFilePreservingMode(
+  targetPath: string,
+  content: string,
+  mode: number,
+  purpose: string,
+): Promise<void> {
+  const tempPath = uniqueTempPath(targetPath, purpose);
+  try {
+    await writeFile(tempPath, content, { encoding: "utf8", mode });
+    await chmod(tempPath, mode);
+    await rename(tempPath, targetPath);
+  } catch (error) {
+    try {
+      await unlink(tempPath);
+    } catch (cleanupError) {
+      if (getErrorCode(cleanupError) !== "ENOENT") {
+        console.warn(`${LOG_PREFIX}: failed to remove temporary models.json file`, cleanupError);
+      }
+    }
+    throw error;
+  }
+}
+
+async function atomicRestoreFileFromBackup(
+  backupPath: string,
+  targetPath: string,
+  mode: number,
+): Promise<void> {
+  const tempPath = uniqueTempPath(targetPath, "restore");
+  try {
+    await copyFile(backupPath, tempPath, fsConstants.COPYFILE_EXCL);
+    await chmod(tempPath, mode);
+    await rename(tempPath, targetPath);
+  } catch (error) {
+    try {
+      await unlink(tempPath);
+    } catch (cleanupError) {
+      if (getErrorCode(cleanupError) !== "ENOENT") {
+        console.warn(`${LOG_PREFIX}: failed to remove temporary restore file`, cleanupError);
+      }
+    }
+    throw error;
+  }
+}
+
+type ModelsJsonFixTransactionResult =
+  | { ok: true }
+  | { ok: false; postCheckError: string };
+
+async function applyModelsJsonFixTransaction(
+  modifiedText: string,
+  backupPath: string,
+  validateWrittenText: (writtenText: string) => string | null,
+): Promise<ModelsJsonFixTransactionResult> {
+  const originalMode = (await stat(MODELS_JSON_PATH)).mode & 0o7777;
+  await copyFile(MODELS_JSON_PATH, backupPath, fsConstants.COPYFILE_EXCL);
+  await chmod(backupPath, originalMode);
+
+  let targetReplaced = false;
+  try {
+    await atomicReplaceTextFilePreservingMode(
+      MODELS_JSON_PATH,
+      modifiedText,
+      originalMode,
+      "fix",
+    );
+    targetReplaced = true;
+
+    const writtenText = await readFile(MODELS_JSON_PATH, "utf8");
+    const postCheckError = validateWrittenText(writtenText);
+    const writtenMode = (await stat(MODELS_JSON_PATH)).mode & 0o7777;
+    const effectiveError = postCheckError ?? (
+      writtenMode === originalMode
+        ? null
+        : `models.json access mode changed from ${originalMode.toString(8)} to ${writtenMode.toString(8)}`
+    );
+    if (effectiveError !== null) {
+      await atomicRestoreFileFromBackup(backupPath, MODELS_JSON_PATH, originalMode);
+      return { ok: false, postCheckError: effectiveError };
+    }
+
+    return { ok: true };
+  } catch (error) {
+    if (targetReplaced) {
+      try {
+        await atomicRestoreFileFromBackup(backupPath, MODELS_JSON_PATH, originalMode);
+      } catch (restoreError) {
+        throw new AggregateError(
+          [error, restoreError],
+          "models.json fix failed and the atomic backup restore also failed",
+        );
+      }
+    }
+    throw error;
+  }
 }
 
 // Internal helpers exported only so the task verification script
@@ -6593,6 +6723,7 @@ export const __internals_for_tests = {
   isSessionAffinity403Applicable,
   isOpenAISdkHeader403Applicable,
   hasPromptCacheRetentionUnsupportedSignal,
+  hasPromptCacheRetentionUnsupportedErrorMessage,
   // Non-GPT OpenAI-compatible model detection
   isKimiLikeModel,
   isKimiLikeAssistantMessage,
@@ -6816,6 +6947,9 @@ export const __internals_for_tests = {
   deepEqualIgnoringKeys,
   formatCompatKeysForInsertion,
   backupTimestamp,
+  atomicReplaceTextFilePreservingMode,
+  atomicRestoreFileFromBackup,
+  applyModelsJsonFixTransaction,
   // Fix suggestion builder
   buildFixSuggestion,
   // Adaptive thinking compat helpers
@@ -7615,6 +7749,31 @@ export default function (pi: ExtensionAPI) {
     syncSessionHash(ctx);
     const msgRecord = asRecord(event.message);
 
+    // Some providers expose an HTTP 400 error body only through the finalized
+    // assistant error message; after_provider_response may contain the status
+    // and no diagnostic response headers. Record the same model-scoped
+    // prompt_cache_retention fallback from that authoritative message identity.
+    if (runtimeOptimizerEnabled && hasPromptCacheRetentionUnsupportedErrorMessage(event.message)) {
+      const fallbackModel = resolveRouteModel(ctx.model, ctx) ?? ctx.model;
+      const messageModel = modelFromAssistantMessage(event.message, fallbackModel) ?? fallbackModel;
+      const errorModel = messageModel
+        ? findModelInRegistry(ctx.modelRegistry, messageModel.provider, messageModel.id) ?? messageModel
+        : undefined;
+      if (errorModel && isExplicitPromptCacheRetentionUnsupportedApplicable(errorModel)) {
+        const key = modelKey(errorModel);
+        promptCacheRetention400Models.add(key);
+        if (!warnedPromptCacheRetention400Models.has(key)) {
+          warnedPromptCacheRetention400Models.add(key);
+          ctx.ui.notify(
+            `⚠️ ${LOG_PREFIX}: ${key} rejected prompt_cache_retention. ` +
+            getPromptCacheRetentionUnsupportedHint() +
+            ` Run /cache-optimizer doctor for the exact edit location.`,
+            "warning",
+          );
+        }
+      }
+    }
+
     // Record only Anthropic's explicit mixed-TTL ordering error. This is a
     // non-retryable 400 in Pi 0.82.1, so the fallback applies to the next
     // subsequent request (and to a retry only if another layer initiates one).
@@ -7729,27 +7888,27 @@ export default function (pi: ExtensionAPI) {
   //   reset   — reset current provider/model footer stats bucket (local only)
   //   (no args) — interactive menu (with UI) or help summary
   // ────────────────────────────────────────────────────────────────
-  pi.registerCommand("cache-optimizer", {
-    description: "Configure and diagnose Pi cache behavior",
-    getArgumentCompletions: getCacheOptimizerArgumentCompletions,
-    handler: async (args: string, cmdCtx) => {
-      syncSessionHash(cmdCtx);
-      const selectedModel = cmdCtx.model;
-      const model = resolveRouteModel(selectedModel, cmdCtx as unknown as ExtensionContext) ?? selectedModel;
-      const commandParts = args.trim().toLowerCase().split(/\s+/).filter(Boolean);
-      const subcommand = commandParts[0] || "help";
+  async function handleCacheOptimizerCommand(
+    args: string,
+    cmdCtx: ExtensionCommandContext,
+  ): Promise<void> {
+    syncSessionHash(cmdCtx);
+    const selectedModel = cmdCtx.model;
+    const model = resolveRouteModel(selectedModel, cmdCtx) ?? selectedModel;
+    const commandParts = args.trim().toLowerCase().split(/\s+/).filter(Boolean);
+    const subcommand = commandParts[0] || "help";
 
       if (subcommand === "enable") {
         setRuntimeOptimizerEnabled(true);
         resetCurrentSessionStats();
-        await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
-        await publishStatus(cmdCtx as unknown as ExtensionContext, model);
+        await flushPersistCacheStats(cmdCtx);
+        await publishStatus(cmdCtx, model);
         cmdCtx.ui.notify(`✅ Pi Cache Optimizer enabled for this Pi process. Local footer stats were reset for before/after comparison.\n${formatOptimizerRuntimeMode()}`, "info");
       } else if (subcommand === "disable") {
         setRuntimeOptimizerEnabled(false);
         resetCurrentSessionStats();
-        await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
-        await publishStatus(cmdCtx as unknown as ExtensionContext, model);
+        await flushPersistCacheStats(cmdCtx);
+        await publishStatus(cmdCtx, model);
         cmdCtx.ui.notify(`⏸️ Pi Cache Optimizer disabled for this Pi process. Local footer stats were reset and will keep collecting while disabled for comparison.\n${formatOptimizerRuntimeMode()}`, "warning");
       } else if (subcommand === "doctor") {
         if (!model) {
@@ -7795,7 +7954,7 @@ export default function (pi: ExtensionAPI) {
           await writePersistedFooterMode(nextMode);
           persistedFooterStatsMode = nextMode;
           lastStatusText = undefined;
-          await publishStatus(cmdCtx as unknown as ExtensionContext, model);
+          await publishStatus(cmdCtx, model);
           const resolved = resolveFooterStatsMode(persistedFooterStatsMode);
           cmdCtx.ui.notify(
             `✅ Footer mode set to ${resolved.mode}. Persistent config overrides ${FOOTER_MODE_ENV}.`,
@@ -7842,10 +8001,10 @@ export default function (pi: ExtensionAPI) {
         resetStatsForModel(model);
 
         // Persist immediately.
-        await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
+        await flushPersistCacheStats(cmdCtx);
 
         // Update footer to show 0/0.
-        await publishStatus(cmdCtx as unknown as ExtensionContext, model);
+        await publishStatus(cmdCtx, model);
 
         cmdCtx.ui.notify(
           `✅ Reset local footer cache stats for "${displayKey}". ` +
@@ -7998,20 +8157,20 @@ export default function (pi: ExtensionAPI) {
               );
               if (confirmed) {
                 try {
-                  await copyFile(MODELS_JSON_PATH, backupPath);
-                  const tempPath = `${MODELS_JSON_PATH}.${process.pid}.${Date.now()}.fix.tmp`;
-                  await writeFile(tempPath, plan.modifiedText, "utf8");
-                  await rename(tempPath, MODELS_JSON_PATH);
-
-                  const writtenText = await readFile(MODELS_JSON_PATH, "utf8");
-                  const postErr = selfCheckMissingEntryInsertion(
-                    originalText, writtenText,
-                    suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys,
+                  const result = await applyModelsJsonFixTransaction(
+                    plan.modifiedText,
+                    backupPath,
+                    (writtenText) => selfCheckMissingEntryInsertion(
+                      originalText,
+                      writtenText,
+                      suggestion.providerLabel,
+                      suggestion.modelId,
+                      suggestion.compatKeys,
+                    ),
                   );
-                  if (postErr !== null) {
-                    await copyFile(backupPath, MODELS_JSON_PATH);
+                  if ("postCheckError" in result) {
                     cmdCtx.ui.notify(
-                      `❌ Post-write self-check failed: ${postErr}\n` +
+                      `❌ Post-write self-check failed: ${result.postCheckError}\n` +
                       `The backup at ${backupPath} has been restored. No changes applied.`,
                       "error",
                     );
@@ -8025,7 +8184,8 @@ export default function (pi: ExtensionAPI) {
                   );
                 } catch (e) {
                   cmdCtx.ui.notify(
-                    `❌ Write failed: ${e instanceof Error ? e.message : String(e)}`,
+                    `❌ Write failed: ${e instanceof Error ? e.message : String(e)}\n` +
+                    `Backup may be at: ${backupPath}`,
                     "error",
                   );
                 }
@@ -8147,22 +8307,21 @@ export default function (pi: ExtensionAPI) {
 
         // Write: backup → temp + rename → self-check again
         try {
-          // Backup
-          await copyFile(MODELS_JSON_PATH, backupPath);
-
-          // Atomic write
-          const tempPath = `${MODELS_JSON_PATH}.${process.pid}.${Date.now()}.fix.tmp`;
-          await writeFile(tempPath, modifiedText, "utf8");
-          await rename(tempPath, MODELS_JSON_PATH);
-
-          // Post-write self-check (read back)
-          const writtenText = await readFile(MODELS_JSON_PATH, "utf8");
-          const postCheckError = selfCheckFix(originalText, writtenText, suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys, decision.placement);
-          if (postCheckError !== null) {
-            // Restore from backup
-            await copyFile(backupPath, MODELS_JSON_PATH);
+          const result = await applyModelsJsonFixTransaction(
+            modifiedText,
+            backupPath,
+            (writtenText) => selfCheckFix(
+              originalText,
+              writtenText,
+              suggestion.providerLabel,
+              suggestion.modelId,
+              suggestion.compatKeys,
+              decision.placement,
+            ),
+          );
+          if ("postCheckError" in result) {
             cmdCtx.ui.notify(
-              `❌ Post-write self-check failed: ${postCheckError}\n` +
+              `❌ Post-write self-check failed: ${result.postCheckError}\n` +
               `The backup at ${backupPath} has been restored. No changes applied.`,
               "error",
             );
@@ -8198,182 +8357,17 @@ export default function (pi: ExtensionAPI) {
           ];
           const choice = await cmdCtx.ui.select("Cache Optimizer", menuOptions);
           if (choice === menuOptions[0]) {
-            setRuntimeOptimizerEnabled(true);
-            resetCurrentSessionStats();
-            await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
-            await publishStatus(cmdCtx as unknown as ExtensionContext, model);
-            cmdCtx.ui.notify(`✅ Pi Cache Optimizer enabled for this Pi process. Local footer stats were reset for before/after comparison.\n${formatOptimizerRuntimeMode()}`, "info");
+            await handleCacheOptimizerCommand("enable", cmdCtx);
           } else if (choice === menuOptions[1]) {
-            setRuntimeOptimizerEnabled(false);
-            resetCurrentSessionStats();
-            await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
-            await publishStatus(cmdCtx as unknown as ExtensionContext, model);
-            cmdCtx.ui.notify(`⏸️ Pi Cache Optimizer disabled for this Pi process. Local footer stats were reset and will keep collecting while disabled for comparison.\n${formatOptimizerRuntimeMode()}`, "warning");
+            await handleCacheOptimizerCommand("disable", cmdCtx);
           } else if (choice === menuOptions[2]) {
-            if (!model) {
-              cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
-            } else {
-              const diagnosis = buildDoctorDiagnosis(model, { promptCacheRetention400: promptCacheRetention400Models.has(modelKey(model)), anthropicTtlOrderError: anthropicTtlOrderErrorModels.has(modelKey(model)), sessionAffinity403: sendSessionAffinityHeaders403Models.has(modelKey(model)), openAISdkHeader403: openAISdkHeader403Models.has(modelKey(model)) });
-              const adapter = selectAdapterForModel(model);
-              const sk = model ? sessionModelKey(model) : undefined;
-              const statsState = model ? cacheStatsTotalsByModel[modelKey(model)] : undefined;
-              const samples = sk ? getRecentSamples(sk) : [];
-              const lowHitLines = buildLowHitDiagnosis(model, adapter, statsState, samples);
-              const fullDiagnosis = lowHitLines.length > 0
-                ? diagnosis + "\n" + lowHitLines.join("\n")
-                : diagnosis;
-              cmdCtx.ui.notify(fullDiagnosis, "info");
-            }
+            await handleCacheOptimizerCommand("doctor", cmdCtx);
           } else if (choice === menuOptions[3]) {
-            if (!model) {
-              cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
-            } else {
-              const adapter = selectAdapterForModel(model);
-              const sk = model ? sessionModelKey(model) : undefined;
-              const statsState = model ? cacheStatsTotalsByModel[modelKey(model)] : undefined;
-              const samples = sk ? getRecentSamples(sk) : [];
-              const output = buildStatsOutput(model, adapter, statsState, samples);
-              cmdCtx.ui.notify(output, "info");
-            }
+            await handleCacheOptimizerCommand("stats", cmdCtx);
           } else if (choice === menuOptions[4]) {
-            if (!model) {
-              cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
-            } else {
-              const compatResult = buildCompatDiagnosis(model);
-              if (compatResult) {
-                cmdCtx.ui.notify(compatResult, "warning");
-              } else {
-                cmdCtx.ui.notify(
-                  isAdaptiveThinkingCompatApplicable(model) || isDeepSeekCompatCheckApplicable(model) || isCompatCheckApplicable(model)
-                    ? "✅ Compat fully configured."
-                    : getCompatCheckNotApplicableLines(model).join("\n"),
-                  "info",
-                );
-              }
-            }
+            await handleCacheOptimizerCommand("compat", cmdCtx);
           } else if (choice === menuOptions[5]) {
-            // Fix — auto-fix compat issues
-            if (!model) {
-              cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
-              return;
-            }
-            const suggestion = buildCommandFixSuggestion(model);
-            if (!suggestion) {
-              const key = modelKey(model);
-              cmdCtx.ui.notify(`✅ Nothing to fix for "${key}". Compat already configured.`, "info");
-              return;
-            }
-
-            // Read models.json
-            let originalText: string;
-            try {
-              originalText = await readFile(MODELS_JSON_PATH, "utf8");
-            } catch {
-              cmdCtx.ui.notify(`❌ Could not read ${MODELS_JSON_PATH}. File may not exist.`, "error");
-              return;
-            }
-
-            const location = locateModelInJsonc(originalText, suggestion.providerLabel, suggestion.modelId);
-            const menuDecision = location
-              ? chooseFixPlacement(
-                originalText,
-                location,
-                suggestion.compatKeys,
-                suggestion.providerLabel,
-                suggestion.forceModelLevel,
-              )
-              : undefined;
-            const overridePlan = location
-              ? undefined
-              : composeModelOverrideInsertion(
-                originalText,
-                suggestion.providerLabel,
-                suggestion.modelId,
-                suggestion.compatKeys,
-              );
-            if (!location && !overridePlan) {
-              cmdCtx.ui.notify(
-                `❌ Could not safely locate a modelOverrides insertion point in ${getModelsJsonDisplayPath()}.\n` +
-                `Manual edit required:\n${formatMissingEntryManualSnippet(suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys)}\n` +
-                `Then run /reload.`,
-                "warning",
-              );
-              return;
-            }
-            const modifiedText = location && menuDecision
-              ? composeFixInsertion(originalText, location, suggestion.compatKeys, menuDecision.placement)
-              : overridePlan!.modifiedText;
-            const checkError = location && menuDecision
-              ? selfCheckFix(originalText, modifiedText, suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys, menuDecision.placement)
-              : selfCheckMissingEntryInsertion(originalText, modifiedText, suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys);
-            if (checkError !== null) {
-              cmdCtx.ui.notify(`❌ Self-check failed: ${checkError}\nNo changes made.`, "error");
-              return;
-            }
-
-            const keysPreview = JSON.stringify(suggestion.compatKeys, null, 2);
-            const ts = backupTimestamp();
-            const backupPath = `${MODELS_JSON_PATH}.backup-cache-optimizer-${ts}`;
-
-            const menuLocationDesc = overridePlan?.placementLabel ?? (menuDecision?.placement === "provider"
-              ? `providers["${suggestion.providerLabel}"] -> compat (provider level)`
-              : menuDecision?.placement === "modelOverride"
-                ? `providers["${suggestion.providerLabel}"] -> modelOverrides -> "${suggestion.modelId}" -> compat`
-                : `providers["${suggestion.providerLabel}"] -> models -> "${suggestion.modelId}" -> compat (model level)`);
-            const menuScopeRiskLine = menuDecision?.placement === "provider"
-              ? `  1. This change applies to ALL ${location?.allModelIds.length || 1} model(s) in the "${suggestion.providerLabel}" provider, across all sessions.`
-              : `  1. This change affects ALL sessions using the "${suggestion.providerLabel}" provider/channel (scoped to model "${suggestion.modelId}").`;
-
-            const previewLines = [
-              `📝 Preview of changes to ${getModelsJsonDisplayPath()}:`,
-              `Location: ${menuLocationDesc}`,
-              `Placement: ${menuDecision ? `${menuDecision.placement} level — ${menuDecision.reason}` : "modelOverrides — built-in/API-login-safe model override"}`,
-              `Compat JSON to write:`,
-              keysPreview,
-              ``,
-              `⚠️  Risk notice:`,
-              menuScopeRiskLine,
-              `  2. A timestamped backup will be written to: ${backupPath}`,
-              `  3. You must restart Pi / run /reload for the change to take effect.`,
-              `  4. If the file contains comments, verify the result after write.`,
-              ``,
-              `Apply these changes?`,
-            ];
-
-            const confirmed = await cmdCtx.ui.confirm("Cache Optimizer — Fix", previewLines.join("\n"));
-            if (!confirmed) {
-              cmdCtx.ui.notify("No changes were made. Canceled by user.", "info");
-              return;
-            }
-
-            try {
-              await copyFile(MODELS_JSON_PATH, backupPath);
-              const tempPath = `${MODELS_JSON_PATH}.${process.pid}.${Date.now()}.fix.tmp`;
-              await writeFile(tempPath, modifiedText, "utf8");
-              await rename(tempPath, MODELS_JSON_PATH);
-
-              const writtenText = await readFile(MODELS_JSON_PATH, "utf8");
-              const postCheck = location && menuDecision
-                ? selfCheckFix(originalText, writtenText, suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys, menuDecision.placement)
-                : selfCheckMissingEntryInsertion(originalText, writtenText, suggestion.providerLabel, suggestion.modelId, suggestion.compatKeys);
-              if (postCheck !== null) {
-                await copyFile(backupPath, MODELS_JSON_PATH);
-                cmdCtx.ui.notify(`❌ Post-write check failed: ${postCheck}\nBackup restored.`, "error");
-                return;
-              }
-
-              cmdCtx.ui.notify(
-                `✅ Fix applied to ${getModelsJsonDisplayPath()}.` +
-                `\nBackup: ${backupPath}` +
-                `\nRun /reload or restart Pi for the change to take effect.`,
-                "info",
-              );
-            } catch (writeError) {
-              cmdCtx.ui.notify(
-                `❌ Write failed: ${writeError instanceof Error ? writeError.message : String(writeError)}`,
-                "error",
-              );
-            }
+            await handleCacheOptimizerCommand("fix", cmdCtx);
           } else if (choice === menuOptions[6]) {
             const modeOptions = ["total — Daily cumulative totals (default)", "session — Current Pi conversation session", "process — Current Pi process only", "Cancel"];
             const modeChoice = await cmdCtx.ui.select("Footer cache stats mode", modeOptions);
@@ -8384,42 +8378,9 @@ export default function (pi: ExtensionAPI) {
                 : modeChoice === modeOptions[2]
                   ? "process"
                   : undefined;
-            if (nextMode) {
-              try {
-                await writePersistedFooterMode(nextMode);
-                persistedFooterStatsMode = nextMode;
-                lastStatusText = undefined;
-                await publishStatus(cmdCtx as unknown as ExtensionContext, model);
-                cmdCtx.ui.notify(
-                  `✅ Footer mode set to ${nextMode}. Persistent config overrides ${FOOTER_MODE_ENV}.`,
-                  "info",
-                );
-              } catch (error) {
-                cmdCtx.ui.notify(
-                  `❌ Could not update footer mode config: ${error instanceof Error ? error.message : String(error)}`,
-                  "error",
-                );
-              }
-            }
+            if (nextMode) await handleCacheOptimizerCommand(`config footer-mode ${nextMode}`, cmdCtx);
           } else if (choice === menuOptions[7]) {
-            if (!model) {
-              cmdCtx.ui.notify("No active model selected. Select a model first with /model or pi --model.", "warning");
-            } else {
-              const adapter = selectAdapterForModel(model);
-              if (!adapter) {
-                cmdCtx.ui.notify("ℹ️ Active model does not match a cache adapter. No stats to reset.", "info");
-              } else {
-                const displayKey = modelKey(model);
-                resetStatsForModel(model);
-                await flushPersistCacheStats(cmdCtx as unknown as ExtensionContext);
-                await publishStatus(cmdCtx as unknown as ExtensionContext, model);
-                cmdCtx.ui.notify(
-                  `✅ Reset local footer cache stats for "${displayKey}". ` +
-                  "Upstream provider prompt cache was not modified.",
-                  "info",
-                );
-              }
-            }
+            await handleCacheOptimizerCommand("reset", cmdCtx);
           }
           // choice === "cancel" or undefined → no action
           return;
@@ -8459,6 +8420,11 @@ export default function (pi: ExtensionAPI) {
         }
         cmdCtx.ui.notify(diagnosis.join("\n"), "info");
       }
-    },
+  }
+
+  pi.registerCommand("cache-optimizer", {
+    description: "Configure and diagnose Pi cache behavior",
+    getArgumentCompletions: getCacheOptimizerArgumentCompletions,
+    handler: handleCacheOptimizerCommand,
   });
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-cache-optimizer",
-  "version": "2.8.1",
+  "version": "2.8.3",
   "description": "Improve Pi prompt/KV cache hit rates with stable prompts, OpenAI-compatible cache keys, proxy compat warnings, and footer cache stats.",
   "keywords": [
     "pi-package",
@@ -28,19 +28,20 @@
     "#extension": "./index.ts"
   },
   "scripts": {
-    "test": "node --import jiti/register tests/review-findings.test.ts",
+    "test": "node --import jiti/register --test tests/*.test.ts",
     "typecheck": "tsc --noEmit --pretty false",
     "check:diff": "git diff --check",
     "check:pack": "npm pack --dry-run --cache node_modules/.cache/npm",
     "check": "npm run typecheck && npm test && npm run check:diff && npm run check:pack"
   },
   "devDependencies": {
+    "@earendil-works/pi-coding-agent": "^0.84.2",
     "@types/node": "^22.0.0",
     "jiti": "^2.7.0",
     "typescript": "^5.8.3"
   },
   "peerDependencies": {
-    "@earendil-works/pi-coding-agent": "*"
+    "@earendil-works/pi-coding-agent": ">=0.82.0"
   },
   "repository": {
     "type": "git",