diff --git a/packages/cli/src/nodehost.ts b/packages/cli/src/nodehost.ts
index fa9aba0f5d..208fc4fa69 100644
--- a/packages/cli/src/nodehost.ts
+++ b/packages/cli/src/nodehost.ts
@@ -24,9 +24,6 @@ import {
     AZURE_AI_INFERENCE_TOKEN_SCOPES,
     MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
     DOT_ENV_FILENAME,
-    LARGE_MODEL_ID,
-    SMALL_MODEL_ID,
-    VISION_MODEL_ID,
 } from "../../core/src/constants"
 import { tryReadText } from "../../core/src/fs"
 import {
@@ -41,7 +38,7 @@ import {
     ModelConfiguration,
 } from "../../core/src/host"
 import { TraceOptions } from "../../core/src/trace"
-import { deleteEmptyValues, logError, logVerbose } from "../../core/src/util"
+import { logError, logVerbose } from "../../core/src/util"
 import { parseModelIdentifier } from "../../core/src/models"
 import { LanguageModel } from "../../core/src/chat"
 import { errorMessage, NotSupportedError } from "../../core/src/error"
@@ -50,7 +47,11 @@ import { shellConfirm, shellInput, shellSelect } from "./input"
 import { shellQuote } from "../../core/src/shell"
 import { uniq } from "es-toolkit"
 import { PLimitPromiseQueue } from "../../core/src/concurrency"
-import { LanguageModelConfiguration, Project, ResponseStatus } from "../../core/src/server/messages"
+import {
+    LanguageModelConfiguration,
+    Project,
+    ResponseStatus,
+} from "../../core/src/server/messages"
 import { createAzureTokenResolver } from "./azuretoken"
 import {
     createAzureContentSafetyClient,
@@ -396,15 +397,13 @@ export class NodeHost implements RuntimeHost {
         }
 
         const {
-            trace,
             label,
             cwd,
             timeout = SHELL_EXEC_TIMEOUT,
             stdin: input,
         } = options || {}
+        const trace = options?.trace?.startTraceDetails(label || command)
         try {
-            trace?.startDetails(label || command)
-
             // python3 on windows -> python
             if (command === "python3" && process.platform === "win32")
                 command = "python"
diff --git a/packages/cli/src/scripts.ts b/packages/cli/src/scripts.ts
index a13e7e04e5..dfef0044c8 100644
--- a/packages/cli/src/scripts.ts
+++ b/packages/cli/src/scripts.ts
@@ -8,7 +8,7 @@ import {
     fixPromptDefinitions,
     createScript as coreCreateScript,
 } from "../../core/src/scripts"
-import { logInfo, logVerbose } from "../../core/src/util"
+import { deleteEmptyValues, logInfo, logVerbose } from "../../core/src/util"
 import { runtimeHost } from "../../core/src/host"
 import { RUNTIME_ERROR_CODE } from "../../core/src/constants"
 import {
@@ -27,14 +27,18 @@ export async function listScripts(options?: ScriptFilterOptions) {
     const prj = await buildProject() // Build the project to get script templates
     const scripts = filterScripts(prj.scripts, options) // Filter scripts based on options
     console.log(
-        YAMLStringify(
-            scripts.map(({ id, title, group, filename, system: isSystem }) => ({
-                id,
-                title,
-                group,
-                filename,
-                isSystem,
-            }))
+        JSON.stringify(
+            scripts.map(({ id, title, group, filename, system: isSystem }) =>
+                deleteEmptyValues({
+                    id,
+                    title,
+                    group,
+                    filename,
+                    isSystem,
+                })
+            ),
+            null,
+            2
         )
     )
 }
diff --git a/packages/core/bundleprompts.js b/packages/core/bundleprompts.js
index 99d2e15adc..3e6a12970a 100644
--- a/packages/core/bundleprompts.js
+++ b/packages/core/bundleprompts.js
@@ -11,10 +11,13 @@ async function main() {
     const promptMap = {}
     const prompts = readdirSync(dir)
     for (const prompt of prompts) {
-        if (!/\.m?js$/.test(prompt)) continue
+        if (!/\.mjs$/.test(prompt)) continue
         const text = readFileSync(`${dir}/${prompt}`, "utf-8")
-        if (/\.genai\.m?js$/.test(prompt))
-            promptMap[prompt.replace(/\.genai\.m?js$/i, "")] = text
+        if (/^system\./.test(prompt)) {
+            const id = prompt.replace(/\.m?js$/i, "")
+            if (promptMap[id]) throw new Error(`duplicate prompt ${id}`)
+            promptMap[id] = text
+        }
     }
     console.log(`found ${Object.keys(promptMap).length} prompts`)
     console.debug(Object.keys(promptMap).join("\n"))
diff --git a/packages/core/src/anthropic.ts b/packages/core/src/anthropic.ts
index e94e459e4e..b3a143a027 100644
--- a/packages/core/src/anthropic.ts
+++ b/packages/core/src/anthropic.ts
@@ -30,10 +30,6 @@ import {
 
 import { deleteUndefinedValues, logError, logVerbose } from "./util"
 import { resolveHttpProxyAgent } from "./proxy"
-import {
-    ChatCompletionRequestCacheKey,
-    getChatCompletionCache,
-} from "./chatcache"
 import { HttpsProxyAgent } from "https-proxy-agent"
 import { MarkdownTrace } from "./trace"
 import { createFetch, FetchType } from "./fetch"
@@ -289,40 +285,6 @@ const completerFactory = (
         const { model } = parseModelIdentifier(req.model)
         const { encode: encoder } = await resolveTokenEncoder(model)
 
-        const cache = !!cacheOrName || !!cacheName
-        const cacheStore = getChatCompletionCache(
-            typeof cacheOrName === "string" ? cacheOrName : cacheName
-        )
-        const cachedKey = cache
-            ? <ChatCompletionRequestCacheKey>{
-                  ...req,
-                  ...cfgNoToken,
-                  model: req.model,
-                  temperature: req.temperature,
-                  top_p: req.top_p,
-                  max_tokens: req.max_tokens,
-                  logit_bias: req.logit_bias,
-              }
-            : undefined
-        trace.itemValue(`caching`, cache)
-        trace.itemValue(`cache`, cacheStore?.name)
-        const { text: cached, finishReason: cachedFinishReason } =
-            (await cacheStore.get(cachedKey)) || {}
-        if (cached !== undefined) {
-            partialCb?.({
-                tokensSoFar: estimateTokens(cached, encoder),
-                responseSoFar: cached,
-                responseChunk: cached,
-                inner,
-            })
-            trace.itemValue(`cache hit`, await cacheStore.getKeySHA(cachedKey))
-            return {
-                text: cached,
-                finishReason: cachedFinishReason,
-                cached: true,
-            }
-        }
-
         const fetch = await createFetch({
             trace,
             retries: retry,
@@ -441,9 +403,6 @@ const completerFactory = (
                 `${usage.total_tokens} total, ${usage.prompt_tokens} prompt, ${usage.completion_tokens} completion`
             )
         }
-
-        if (finishReason === "stop")
-            await cacheStore.set(cachedKey, { text: chatResp, finishReason })
         return {
             text: chatResp,
             finishReason,
diff --git a/packages/core/src/cache.test.ts b/packages/core/src/cache.test.ts
index 086ae0ee17..45bb180981 100644
--- a/packages/core/src/cache.test.ts
+++ b/packages/core/src/cache.test.ts
@@ -36,4 +36,26 @@ describe("Cache", () => {
         assert.ok(sha)
         assert.strictEqual(typeof sha, "string")
     })
+    test("JSONLineCache getOrUpdate retrieves existing value", async () => {
+        const cache = JSONLineCache.byName<string, number>("testCache")
+        await cache.set("existingKey", 42)
+        const value = await cache.getOrUpdate(
+            "existingKey",
+            async () => 99,
+            () => true
+        )
+        assert.strictEqual(value.value, 42)
+    })
+
+    test("JSONLineCache getOrUpdate updates with new value if key does not exist", async () => {
+        const cache = JSONLineCache.byName<string, number>("testCache")
+        const value = await cache.getOrUpdate(
+            "newKey",
+            async () => 99,
+            () => true
+        )
+        assert.strictEqual(value.value, 99)
+        const cachedValue = await cache.get("newKey")
+        assert.strictEqual(cachedValue, 99)
+    })
 })
diff --git a/packages/core/src/cache.ts b/packages/core/src/cache.ts
index f0542e5b6a..55d471878f 100644
--- a/packages/core/src/cache.ts
+++ b/packages/core/src/cache.ts
@@ -25,6 +25,7 @@ export class MemoryCache<K, V>
     implements WorkspaceFileCache<any, any>
 {
     protected _entries: Record<string, CacheEntry<K, V>>
+    private _pending: Record<string, Promise<V>>
 
     // Constructor is private to enforce the use of byName factory method
     protected constructor(public readonly name: string) {
@@ -53,6 +54,7 @@ export class MemoryCache<K, V>
     protected async initialize() {
         if (this._entries) return
         this._entries = {}
+        this._pending = {}
     }
 
     /**
@@ -104,6 +106,29 @@ export class MemoryCache<K, V>
         return this._entries[sha]?.val
     }
 
+    async getOrUpdate(
+        key: K,
+        updater: () => Promise<V>,
+        validator: (val: V) => boolean
+    ): Promise<{ key: string; value: V; cached?: boolean }> {
+        await this.initialize()
+        const sha = await keySHA(key)
+        if (this._entries[sha])
+            return { key: sha, value: this._entries[sha].val, cached: true }
+        if (this._pending[sha])
+            return { key: sha, value: await this._pending[sha], cached: true }
+
+        try {
+            const p = updater()
+            this._pending[sha] = p
+            const value = await p
+            if (validator(value)) await this.set(key, value)
+            return { key: sha, value, cached: false }
+        } finally {
+            delete this._pending[sha]
+        }
+    }
+
     protected async appendEntry(entry: CacheEntry<K, V>) {}
 
     /**
@@ -177,7 +202,7 @@ export class JSONLineCache<K, V> extends MemoryCache<K, V> {
      */
     override async initialize() {
         if (this._entries) return
-        this._entries = {}
+        super.initialize()
         await host.createDirectory(this.folder()) // Ensure directory exists
         const content = await tryReadText(this.path())
         const objs: CacheEntry<K, V>[] = (await JSONLTryParse(content)) ?? []
@@ -201,7 +226,7 @@ export class JSONLineCache<K, V> extends MemoryCache<K, V> {
 }
 
 /**
- * Compute the SHA1 hash of a key for uniqueness.
+ * Compute the hash of a key for uniqueness.
  * Normalizes the key by converting it to a string and appending the core version.
  * @param key - The key to hash
  * @returns A promise resolving to the SHA256 hash string
diff --git a/packages/core/src/chat.ts b/packages/core/src/chat.ts
index 4293677a50..2e48922550 100644
--- a/packages/core/src/chat.ts
+++ b/packages/core/src/chat.ts
@@ -86,6 +86,10 @@ import {
 } from "./server/messages"
 import { unfence } from "./unwrappers"
 import { fenceMD } from "./mkmd"
+import {
+    ChatCompletionRequestCacheKey,
+    getChatCompletionCache,
+} from "./chatcache"
 
 export function toChatCompletionUserMessage(
     expanded: string,
@@ -674,13 +678,16 @@ async function processChatMessage(
     if (chatParticipants?.length) {
         let needsNewTurn = false
         for (const participant of chatParticipants) {
+            const { generator, options: participantOptions } = participant || {}
+            const { label } = participantOptions || {}
+            const participantTrace = trace.startTraceDetails(
+                `🙋 participant ${label || ""}`
+            )
             try {
-                const { generator, options: participantOptions } =
-                    participant || {}
-                const { label } = participantOptions || {}
-                trace.startDetails(`🙋 participant ${label || ""}`)
-
-                const ctx = createChatTurnGenerationContext(options, trace)
+                const ctx = createChatTurnGenerationContext(
+                    options,
+                    participantTrace
+                )
                 await generator(ctx, structuredClone(messages))
                 const node = ctx.node
                 checkCancelled(cancellationToken)
@@ -689,7 +696,7 @@ async function processChatMessage(
                     await renderPromptNode(options.model, node, {
                         flexTokens: options.flexTokens,
                         fenceFormat: options.fenceFormat,
-                        trace,
+                        trace: participantTrace,
                     })
                 if (participantMessages?.length) {
                     if (
@@ -704,7 +711,7 @@ async function processChatMessage(
                         user: true,
                         assistant: true,
                     })
-                    trace.details(
+                    participantTrace.details(
                         `💬 messages (${participantMessages.length})`,
                         renderMessagesToMarkdown(participantMessages, {
                             user: true,
@@ -714,21 +721,22 @@ async function processChatMessage(
                     )
                     messages.push(...participantMessages)
                     needsNewTurn = true
-                } else trace.item("no message")
+                } else participantTrace.item("no message")
                 if (errors?.length) {
                     err = errors[0]
-                    for (const error of errors) trace.error(undefined, error)
+                    for (const error of errors)
+                        participantTrace.error(undefined, error)
                     needsNewTurn = false
                     break
                 }
             } catch (e) {
                 err = e
                 logError(e)
-                trace.error(`participant error`, e)
+                participantTrace.error(`participant error`, e)
                 needsNewTurn = false
                 break
             } finally {
-                trace?.endDetails()
+                participantTrace.endDetails()
             }
         }
         if (needsNewTurn) return undefined
@@ -873,8 +881,12 @@ export async function executeChatSession(
         fallbackTools,
         choices,
         topLogprobs,
+        cache: cacheOrName,
+        cacheName,
     } = genOptions
     assert(!!model, "model is required")
+
+    const { token, source, ...cfgNoToken } = connectionToken
     const top_logprobs = genOptions.topLogprobs > 0 ? topLogprobs : undefined
     const logprobs = genOptions.logprobs || top_logprobs > 0 ? true : undefined
     traceLanguageModelConnection(trace, genOptions, connectionToken)
@@ -891,11 +903,17 @@ export async function executeChatSession(
                   }
           )
         : undefined
+    const cache = !!cacheOrName || !!cacheName
+    const cacheStore = cache
+        ? getChatCompletionCache(
+              typeof cacheOrName === "string" ? cacheOrName : cacheName
+          )
+        : undefined
 
+    const chatTrace = trace.startTraceDetails(`🧠 llm chat`, { expanded: true })
     try {
-        trace.startDetails(`🧠 llm chat`, { expanded: true })
         if (toolDefinitions?.length) {
-            trace.detailsFenced(`🛠️ tools`, tools, "yaml")
+            chatTrace.detailsFenced(`🛠️ tools`, tools, "yaml")
             const toolNames = toolDefinitions.map(({ spec }) => spec.name)
             const duplicates = uniq(toolNames).filter(
                 (name, index) => toolNames.lastIndexOf(name) !== index
@@ -909,7 +927,7 @@ export async function executeChatSession(
             collapseChatMessages(messages)
             const tokens = estimateChatTokens(model, messages)
             if (messages)
-                trace.details(
+                chatTrace.details(
                     `💬 messages (${messages.length})`,
                     renderMessagesToMarkdown(messages, {
                         user: true,
@@ -923,10 +941,10 @@ export async function executeChatSession(
             let resp: ChatCompletionResponse
             try {
                 checkCancelled(cancellationToken)
+                const reqTrace = chatTrace.startTraceDetails(`📤 llm request`)
                 try {
-                    trace.startDetails(`📤 llm request`)
                     const logit_bias = await choicesToLogitBias(
-                        trace,
+                        reqTrace,
                         model,
                         choices
                     )
@@ -962,21 +980,50 @@ export async function executeChatSession(
                                   : undefined,
                         messages,
                     }
-                    updateChatFeatures(trace, req)
+                    updateChatFeatures(reqTrace, req)
                     logVerbose(
                         `chat: sending ${messages.length} messages to ${model} (~${tokens ?? "?"} tokens)\n`
                     )
-                    resp = await completer(
-                        req,
-                        connectionToken,
-                        genOptions,
-                        trace
-                    )
+
+                    const infer = async () =>
+                        await completer(
+                            req,
+                            connectionToken,
+                            genOptions,
+                            reqTrace
+                        )
+                    if (cacheStore) {
+                        const cachedKey = deleteUndefinedValues({
+                            ...req,
+                            ...cfgNoToken,
+                        }) satisfies ChatCompletionRequestCacheKey
+                        const validator = (value: ChatCompletionResponse) => {
+                            const ok = value?.finishReason === "stop"
+                            return ok
+                        }
+                        const cacheRes = await cacheStore.getOrUpdate(
+                            cachedKey,
+                            infer,
+                            validator
+                        )
+                        resp = cacheRes.value
+                        resp.cached = cacheRes.cached
+                        if (resp.cached) {
+                            reqTrace.itemValue("cache", cacheStore.name)
+                            reqTrace.itemValue("cache_key", cacheRes.key)
+                            logVerbose(
+                                `chat: cache hit (${cacheStore.name}/${cacheRes.key.slice(0, 7)})`
+                            )
+                        }
+                    } else {
+                        resp = await infer()
+                    }
+
                     if (resp.variables)
                         genVars = { ...(genVars || {}), ...resp.variables }
                 } finally {
                     logVerbose("\n")
-                    trace.endDetails()
+                    reqTrace.endDetails()
                 }
 
                 const output = await processChatMessage(
@@ -1008,9 +1055,9 @@ export async function executeChatSession(
             }
         }
     } finally {
-        await dispose(disposables, { trace })
-        stats.trace(trace)
-        trace.endDetails()
+        await dispose(disposables, { trace: chatTrace })
+        stats.trace(chatTrace)
+        chatTrace.endDetails()
     }
 }
 
diff --git a/packages/core/src/chatcache.ts b/packages/core/src/chatcache.ts
index 1cc01020e8..bf1ecc9b24 100644
--- a/packages/core/src/chatcache.ts
+++ b/packages/core/src/chatcache.ts
@@ -9,21 +9,13 @@ import { LanguageModelConfiguration } from "./server/messages"
 // Define the type for a cache key, which combines chat completion request
 // with additional model options, excluding "token" and "source" from the language model configuration.
 export type ChatCompletionRequestCacheKey = CreateChatCompletionRequest &
-    ModelOptions &
     Omit<LanguageModelConfiguration, "token" | "source">
 
-// Define the type for a cache value, containing the response text
-// and the reason for completion.
-export type ChatCompletationRequestCacheValue = {
-    text: string
-    finishReason: ChatCompletionResponse["finishReason"]
-}
-
 // Define a JSON line cache type that maps cache keys to cache values.
 // This cache stores chat completion requests and their associated responses.
 export type ChatCompletationRequestCache = JSONLineCache<
     ChatCompletionRequestCacheKey,
-    ChatCompletationRequestCacheValue
+    ChatCompletionResponse
 >
 
 // Function to retrieve a chat completion cache.
@@ -34,6 +26,6 @@ export function getChatCompletionCache(
 ): ChatCompletationRequestCache {
     return JSONLineCache.byName<
         ChatCompletionRequestCacheKey,
-        ChatCompletationRequestCacheValue
+        ChatCompletionResponse
     >(name || CHAT_CACHE)
 }
diff --git a/packages/core/src/crypto.ts b/packages/core/src/crypto.ts
index 96c5610b14..9a5827355d 100644
--- a/packages/core/src/crypto.ts
+++ b/packages/core/src/crypto.ts
@@ -3,6 +3,7 @@ import { getRandomValues as cryptoGetRandomValues } from "crypto"
 
 // Importing the toHex function from the util module to convert byte arrays to hexadecimal strings
 import { concatBuffers, toHex, utf8Encode } from "./util"
+import { CORE_VERSION } from "./version"
 
 function getRandomValues(bytes: Uint8Array) {
     if (typeof self !== "undefined" && self.crypto) {
@@ -40,7 +41,7 @@ export function randomHex(size: number) {
 }
 
 export async function hash(value: any, options?: HashOptions) {
-    const { algorithm = "sha-1", length, ...rest } = options || {}
+    const { algorithm = "sha-256", version, length, ...rest } = options || {}
 
     const sep = utf8Encode("|")
     const h: Uint8Array[] = []
@@ -68,6 +69,10 @@ export async function hash(value: any, options?: HashOptions) {
             }
         else h.push(utf8Encode(JSON.stringify(v)))
     }
+    if (version) {
+        await append(CORE_VERSION)
+        await append(sep)
+    }
     await append(value)
     await append(sep)
     await append(rest)
diff --git a/packages/core/src/fileedits.ts b/packages/core/src/fileedits.ts
index 94a14bfb7d..487bebbbff 100644
--- a/packages/core/src/fileedits.ts
+++ b/packages/core/src/fileedits.ts
@@ -121,8 +121,8 @@ export async function computeFileEdits(
 
     // Apply user-defined output processors
     if (outputProcessors?.length) {
+        const outputTrace = trace.startTraceDetails("🖨️ output processors")
         try {
-            trace.startDetails("🖨️ output processors")
             for (const outputProcessor of outputProcessors) {
                 const {
                     text: newText,
@@ -141,7 +141,7 @@ export async function computeFileEdits(
 
                 if (newText !== undefined) {
                     text = newText
-                    trace.detailsFenced(`📝 text`, text)
+                    outputTrace.detailsFenced(`📝 text`, text)
                 }
 
                 if (files)
@@ -149,7 +149,7 @@ export async function computeFileEdits(
                         const fn = runtimeHost.path.isAbsolute(n)
                             ? n
                             : runtimeHost.resolvePath(projFolder, n)
-                        trace.detailsFenced(`📁 file ${fn}`, content)
+                        outputTrace.detailsFenced(`📁 file ${fn}`, content)
                         const fileEdit = await getFileEdit(fn)
                         fileEdit.after = content
                         fileEdit.validation = { pathValid: true }
@@ -158,9 +158,9 @@ export async function computeFileEdits(
             }
         } catch (e) {
             logError(e)
-            trace.error(`output processor failed`, e)
+            outputTrace.error(`output processor failed`, e)
         } finally {
-            trace.endDetails()
+            outputTrace.endDetails()
         }
     }
 
@@ -225,52 +225,55 @@ function validateFileOutputs(
 ) {
     if (fileOutputs?.length && Object.keys(fileEdits || {}).length) {
         trace.startDetails("🗂 file outputs")
-        for (const fileEditName of Object.keys(fileEdits)) {
-            const fe = fileEdits[fileEditName]
-            for (const fileOutput of fileOutputs) {
-                const { pattern, options } = fileOutput
-                if (isGlobMatch(fileEditName, pattern)) {
-                    try {
-                        trace.startDetails(`📁 ${fileEditName}`)
-                        trace.itemValue(`pattern`, pattern)
-                        const { schema: schemaId } = options || {}
-                        if (/\.(json|yaml)$/i.test(fileEditName)) {
-                            const { after } = fileEdits[fileEditName]
-                            const data = /\.json$/i.test(fileEditName)
-                                ? JSON5parse(after)
-                                : YAMLParse(after)
-                            trace.detailsFenced("📝 data", data)
-                            if (schemaId) {
-                                const schema = schemas[schemaId]
-                                if (!schema)
-                                    fe.validation = {
-                                        schemaError: `schema ${schemaId} not found`,
-                                    }
-                                else
-                                    fe.validation = validateJSONWithSchema(
-                                        data,
-                                        schema,
-                                        {
-                                            trace,
+        try {
+            for (const fileEditName of Object.keys(fileEdits)) {
+                const fe = fileEdits[fileEditName]
+                for (const fileOutput of fileOutputs) {
+                    const { pattern, options } = fileOutput
+                    if (isGlobMatch(fileEditName, pattern)) {
+                        try {
+                            trace.startDetails(`📁 ${fileEditName}`)
+                            trace.itemValue(`pattern`, pattern)
+                            const { schema: schemaId } = options || {}
+                            if (/\.(json|yaml)$/i.test(fileEditName)) {
+                                const { after } = fileEdits[fileEditName]
+                                const data = /\.json$/i.test(fileEditName)
+                                    ? JSON5parse(after)
+                                    : YAMLParse(after)
+                                trace.detailsFenced("📝 data", data)
+                                if (schemaId) {
+                                    const schema = schemas[schemaId]
+                                    if (!schema)
+                                        fe.validation = {
+                                            schemaError: `schema ${schemaId} not found`,
                                         }
-                                    )
+                                    else
+                                        fe.validation = validateJSONWithSchema(
+                                            data,
+                                            schema,
+                                            {
+                                                trace,
+                                            }
+                                        )
+                                }
+                            } else {
+                                fe.validation = { pathValid: true }
                             }
-                        } else {
-                            fe.validation = { pathValid: true }
-                        }
-                    } catch (e) {
-                        trace.error(errorMessage(e))
-                        fe.validation = {
-                            schemaError: errorMessage(e),
+                        } catch (e) {
+                            trace.error(errorMessage(e))
+                            fe.validation = {
+                                schemaError: errorMessage(e),
+                            }
+                        } finally {
+                            trace.endDetails()
                         }
-                    } finally {
-                        trace.endDetails()
+                        break
                     }
-                    break
                 }
             }
+        } finally {
+            trace.endDetails()
         }
-        trace.endDetails()
     }
 }
 
diff --git a/packages/core/src/genaisrc/system.agent_docs.genai.mjs b/packages/core/src/genaisrc/system.agent_docs.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_docs.genai.mjs
rename to packages/core/src/genaisrc/system.agent_docs.mjs
diff --git a/packages/core/src/genaisrc/system.agent_fs.genai.mjs b/packages/core/src/genaisrc/system.agent_fs.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_fs.genai.mjs
rename to packages/core/src/genaisrc/system.agent_fs.mjs
diff --git a/packages/core/src/genaisrc/system.agent_git.genai.mjs b/packages/core/src/genaisrc/system.agent_git.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_git.genai.mjs
rename to packages/core/src/genaisrc/system.agent_git.mjs
diff --git a/packages/core/src/genaisrc/system.agent_github.genai.mjs b/packages/core/src/genaisrc/system.agent_github.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_github.genai.mjs
rename to packages/core/src/genaisrc/system.agent_github.mjs
diff --git a/packages/core/src/genaisrc/system.agent_interpreter.genai.mjs b/packages/core/src/genaisrc/system.agent_interpreter.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_interpreter.genai.mjs
rename to packages/core/src/genaisrc/system.agent_interpreter.mjs
diff --git a/packages/core/src/genaisrc/system.agent_planner.genai.mjs b/packages/core/src/genaisrc/system.agent_planner.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_planner.genai.mjs
rename to packages/core/src/genaisrc/system.agent_planner.mjs
diff --git a/packages/core/src/genaisrc/system.agent_user_input.genai.mjs b/packages/core/src/genaisrc/system.agent_user_input.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_user_input.genai.mjs
rename to packages/core/src/genaisrc/system.agent_user_input.mjs
diff --git a/packages/core/src/genaisrc/system.agent_web.genai.mjs b/packages/core/src/genaisrc/system.agent_web.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.agent_web.genai.mjs
rename to packages/core/src/genaisrc/system.agent_web.mjs
diff --git a/packages/core/src/genaisrc/system.annotations.genai.mjs b/packages/core/src/genaisrc/system.annotations.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.annotations.genai.mjs
rename to packages/core/src/genaisrc/system.annotations.mjs
diff --git a/packages/core/src/genaisrc/system.assistant.genai.mjs b/packages/core/src/genaisrc/system.assistant.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.assistant.genai.mjs
rename to packages/core/src/genaisrc/system.assistant.mjs
diff --git a/packages/core/src/genaisrc/system.changelog.genai.mjs b/packages/core/src/genaisrc/system.changelog.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.changelog.genai.mjs
rename to packages/core/src/genaisrc/system.changelog.mjs
diff --git a/packages/core/src/genaisrc/system.diagrams.genai.mjs b/packages/core/src/genaisrc/system.diagrams.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.diagrams.genai.mjs
rename to packages/core/src/genaisrc/system.diagrams.mjs
diff --git a/packages/core/src/genaisrc/system.diff.genai.mjs b/packages/core/src/genaisrc/system.diff.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.diff.genai.mjs
rename to packages/core/src/genaisrc/system.diff.mjs
diff --git a/packages/core/src/genaisrc/system.explanations.genai.mjs b/packages/core/src/genaisrc/system.explanations.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.explanations.genai.mjs
rename to packages/core/src/genaisrc/system.explanations.mjs
diff --git a/packages/core/src/genaisrc/system.files.genai.mjs b/packages/core/src/genaisrc/system.files.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.files.genai.mjs
rename to packages/core/src/genaisrc/system.files.mjs
diff --git a/packages/core/src/genaisrc/system.files_schema.genai.mjs b/packages/core/src/genaisrc/system.files_schema.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.files_schema.genai.mjs
rename to packages/core/src/genaisrc/system.files_schema.mjs
diff --git a/packages/core/src/genaisrc/system.fs_ask_file.genai.mjs b/packages/core/src/genaisrc/system.fs_ask_file.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.fs_ask_file.genai.mjs
rename to packages/core/src/genaisrc/system.fs_ask_file.mjs
diff --git a/packages/core/src/genaisrc/system.fs_diff_files.genai.mjs b/packages/core/src/genaisrc/system.fs_diff_files.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.fs_diff_files.genai.mjs
rename to packages/core/src/genaisrc/system.fs_diff_files.mjs
diff --git a/packages/core/src/genaisrc/system.fs_find_files.genai.mjs b/packages/core/src/genaisrc/system.fs_find_files.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.fs_find_files.genai.mjs
rename to packages/core/src/genaisrc/system.fs_find_files.mjs
diff --git a/packages/core/src/genaisrc/system.fs_read_file.genai.mjs b/packages/core/src/genaisrc/system.fs_read_file.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.fs_read_file.genai.mjs
rename to packages/core/src/genaisrc/system.fs_read_file.mjs
diff --git a/packages/core/src/genaisrc/system.git.genai.mjs b/packages/core/src/genaisrc/system.git.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.git.genai.mjs
rename to packages/core/src/genaisrc/system.git.mjs
diff --git a/packages/core/src/genaisrc/system.git_diff.genai.mjs b/packages/core/src/genaisrc/system.git_diff.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.git_diff.genai.mjs
rename to packages/core/src/genaisrc/system.git_diff.mjs
diff --git a/packages/core/src/genaisrc/system.git_info.genai.mjs b/packages/core/src/genaisrc/system.git_info.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.git_info.genai.mjs
rename to packages/core/src/genaisrc/system.git_info.mjs
diff --git a/packages/core/src/genaisrc/system.github_actions.genai.mjs b/packages/core/src/genaisrc/system.github_actions.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.github_actions.genai.mjs
rename to packages/core/src/genaisrc/system.github_actions.mjs
diff --git a/packages/core/src/genaisrc/system.github_files.genai.mjs b/packages/core/src/genaisrc/system.github_files.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.github_files.genai.mjs
rename to packages/core/src/genaisrc/system.github_files.mjs
diff --git a/packages/core/src/genaisrc/system.github_info.genai.mjs b/packages/core/src/genaisrc/system.github_info.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.github_info.genai.mjs
rename to packages/core/src/genaisrc/system.github_info.mjs
diff --git a/packages/core/src/genaisrc/system.github_issues.genai.mjs b/packages/core/src/genaisrc/system.github_issues.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.github_issues.genai.mjs
rename to packages/core/src/genaisrc/system.github_issues.mjs
diff --git a/packages/core/src/genaisrc/system.github_pulls.genai.mjs b/packages/core/src/genaisrc/system.github_pulls.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.github_pulls.genai.mjs
rename to packages/core/src/genaisrc/system.github_pulls.mjs
diff --git a/packages/core/src/genaisrc/system.math.genai.mjs b/packages/core/src/genaisrc/system.math.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.math.genai.mjs
rename to packages/core/src/genaisrc/system.math.mjs
diff --git a/packages/core/src/genaisrc/system.md_find_files.genai.mjs b/packages/core/src/genaisrc/system.md_find_files.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.md_find_files.genai.mjs
rename to packages/core/src/genaisrc/system.md_find_files.mjs
diff --git a/packages/core/src/genaisrc/system.md_frontmatter.genai.mjs b/packages/core/src/genaisrc/system.md_frontmatter.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.md_frontmatter.genai.mjs
rename to packages/core/src/genaisrc/system.md_frontmatter.mjs
diff --git a/packages/core/src/genaisrc/system.meta_prompt.genai.mjs b/packages/core/src/genaisrc/system.meta_prompt.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.meta_prompt.genai.mjs
rename to packages/core/src/genaisrc/system.meta_prompt.mjs
diff --git a/packages/core/src/genaisrc/system.meta_schema.genai.mjs b/packages/core/src/genaisrc/system.meta_schema.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.meta_schema.genai.mjs
rename to packages/core/src/genaisrc/system.meta_schema.mjs
diff --git a/packages/core/src/genaisrc/system.genai.mjs b/packages/core/src/genaisrc/system.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.genai.mjs
rename to packages/core/src/genaisrc/system.mjs
diff --git a/packages/core/src/genaisrc/system.node_info.genai.mjs b/packages/core/src/genaisrc/system.node_info.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.node_info.genai.mjs
rename to packages/core/src/genaisrc/system.node_info.mjs
diff --git a/packages/core/src/genaisrc/system.node_test.genai.mjs b/packages/core/src/genaisrc/system.node_test.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.node_test.genai.mjs
rename to packages/core/src/genaisrc/system.node_test.mjs
diff --git a/packages/core/src/genaisrc/system.output_markdown.genai.mjs b/packages/core/src/genaisrc/system.output_markdown.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.output_markdown.genai.mjs
rename to packages/core/src/genaisrc/system.output_markdown.mjs
diff --git a/packages/core/src/genaisrc/system.output_plaintext.genai.mjs b/packages/core/src/genaisrc/system.output_plaintext.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.output_plaintext.genai.mjs
rename to packages/core/src/genaisrc/system.output_plaintext.mjs
diff --git a/packages/core/src/genaisrc/system.planner.genai.mjs b/packages/core/src/genaisrc/system.planner.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.planner.genai.mjs
rename to packages/core/src/genaisrc/system.planner.mjs
diff --git a/packages/core/src/genaisrc/system.python.genai.mjs b/packages/core/src/genaisrc/system.python.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.python.genai.mjs
rename to packages/core/src/genaisrc/system.python.mjs
diff --git a/packages/core/src/genaisrc/system.python_code_interpreter.genai.mjs b/packages/core/src/genaisrc/system.python_code_interpreter.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.python_code_interpreter.genai.mjs
rename to packages/core/src/genaisrc/system.python_code_interpreter.mjs
diff --git a/packages/core/src/genaisrc/system.python_types.genai.mjs b/packages/core/src/genaisrc/system.python_types.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.python_types.genai.mjs
rename to packages/core/src/genaisrc/system.python_types.mjs
diff --git a/packages/core/src/genaisrc/system.retrieval_fuzz_search.genai.mjs b/packages/core/src/genaisrc/system.retrieval_fuzz_search.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.retrieval_fuzz_search.genai.mjs
rename to packages/core/src/genaisrc/system.retrieval_fuzz_search.mjs
diff --git a/packages/core/src/genaisrc/system.retrieval_vector_search.genai.mjs b/packages/core/src/genaisrc/system.retrieval_vector_search.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.retrieval_vector_search.genai.mjs
rename to packages/core/src/genaisrc/system.retrieval_vector_search.mjs
diff --git a/packages/core/src/genaisrc/system.retrieval_web_search.genai.mjs b/packages/core/src/genaisrc/system.retrieval_web_search.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.retrieval_web_search.genai.mjs
rename to packages/core/src/genaisrc/system.retrieval_web_search.mjs
diff --git a/packages/core/src/genaisrc/system.safety_canary_word.genai.mjs b/packages/core/src/genaisrc/system.safety_canary_word.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_canary_word.genai.mjs
rename to packages/core/src/genaisrc/system.safety_canary_word.mjs
diff --git a/packages/core/src/genaisrc/system.safety_harmful_content.genai.mjs b/packages/core/src/genaisrc/system.safety_harmful_content.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_harmful_content.genai.mjs
rename to packages/core/src/genaisrc/system.safety_harmful_content.mjs
diff --git a/packages/core/src/genaisrc/system.safety_jailbreak.genai.mjs b/packages/core/src/genaisrc/system.safety_jailbreak.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_jailbreak.genai.mjs
rename to packages/core/src/genaisrc/system.safety_jailbreak.mjs
diff --git a/packages/core/src/genaisrc/system.safety_protected_material.genai.mjs b/packages/core/src/genaisrc/system.safety_protected_material.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_protected_material.genai.mjs
rename to packages/core/src/genaisrc/system.safety_protected_material.mjs
diff --git a/packages/core/src/genaisrc/system.safety_ungrounded_content_summarization.genai.mjs b/packages/core/src/genaisrc/system.safety_ungrounded_content_summarization.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_ungrounded_content_summarization.genai.mjs
rename to packages/core/src/genaisrc/system.safety_ungrounded_content_summarization.mjs
diff --git a/packages/core/src/genaisrc/system.safety_validate_harmful_content.genai.mjs b/packages/core/src/genaisrc/system.safety_validate_harmful_content.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.safety_validate_harmful_content.genai.mjs
rename to packages/core/src/genaisrc/system.safety_validate_harmful_content.mjs
diff --git a/packages/core/src/genaisrc/system.schema.genai.mjs b/packages/core/src/genaisrc/system.schema.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.schema.genai.mjs
rename to packages/core/src/genaisrc/system.schema.mjs
diff --git a/packages/core/src/genaisrc/system.tasks.genai.mjs b/packages/core/src/genaisrc/system.tasks.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.tasks.genai.mjs
rename to packages/core/src/genaisrc/system.tasks.mjs
diff --git a/packages/core/src/genaisrc/system.technical.genai.mjs b/packages/core/src/genaisrc/system.technical.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.technical.genai.mjs
rename to packages/core/src/genaisrc/system.technical.mjs
diff --git a/packages/core/src/genaisrc/system.tool_calls.genai.mjs b/packages/core/src/genaisrc/system.tool_calls.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.tool_calls.genai.mjs
rename to packages/core/src/genaisrc/system.tool_calls.mjs
diff --git a/packages/core/src/genaisrc/system.tools.genai.mjs b/packages/core/src/genaisrc/system.tools.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.tools.genai.mjs
rename to packages/core/src/genaisrc/system.tools.mjs
diff --git a/packages/core/src/genaisrc/system.typescript.genai.mjs b/packages/core/src/genaisrc/system.typescript.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.typescript.genai.mjs
rename to packages/core/src/genaisrc/system.typescript.mjs
diff --git a/packages/core/src/genaisrc/system.user_input.genai.mjs b/packages/core/src/genaisrc/system.user_input.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.user_input.genai.mjs
rename to packages/core/src/genaisrc/system.user_input.mjs
diff --git a/packages/core/src/genaisrc/system.vision_ask_image.genai.mjs b/packages/core/src/genaisrc/system.vision_ask_image.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.vision_ask_image.genai.mjs
rename to packages/core/src/genaisrc/system.vision_ask_image.mjs
diff --git a/packages/core/src/genaisrc/system.zero_shot_cot.genai.mjs b/packages/core/src/genaisrc/system.zero_shot_cot.mjs
similarity index 100%
rename from packages/core/src/genaisrc/system.zero_shot_cot.genai.mjs
rename to packages/core/src/genaisrc/system.zero_shot_cot.mjs
diff --git a/packages/core/src/openai.ts b/packages/core/src/openai.ts
index 97751dfcfd..d2d23bebb2 100644
--- a/packages/core/src/openai.ts
+++ b/packages/core/src/openai.ts
@@ -24,10 +24,6 @@ import { RequestError, errorMessage, serializeError } from "./error"
 import { createFetch, iterateBody, traceFetchPost } from "./fetch"
 import { parseModelIdentifier } from "./models"
 import { JSON5TryParse } from "./json5"
-import {
-    ChatCompletionRequestCacheKey,
-    getChatCompletionCache,
-} from "./chatcache"
 import {
     ChatCompletionToolCall,
     ChatCompletionResponse,
@@ -99,36 +95,6 @@ export const OpenAIChatCompletion: ChatCompletionHandler = async (
     const { provider, model } = parseModelIdentifier(req.model)
     const { encode: encoder } = await resolveTokenEncoder(model)
 
-    const cache = !!cacheOrName || !!cacheName
-    const cacheStore = getChatCompletionCache(
-        typeof cacheOrName === "string" ? cacheOrName : cacheName
-    )
-    const cachedKey = cache
-        ? <ChatCompletionRequestCacheKey>{
-              ...req,
-              ...cfgNoToken,
-              model: req.model,
-              temperature: req.temperature,
-              top_p: req.top_p,
-              max_tokens: req.max_tokens,
-              logit_bias: req.logit_bias,
-          }
-        : undefined
-    trace.itemValue(`caching`, cache)
-    trace.itemValue(`cache`, cacheStore?.name)
-    const { text: cached, finishReason: cachedFinishReason } =
-        (await cacheStore.get(cachedKey)) || {}
-    if (cached !== undefined) {
-        partialCb?.({
-            tokensSoFar: estimateTokens(cached, encoder),
-            responseSoFar: cached,
-            responseChunk: cached,
-            inner,
-        })
-        trace.itemValue(`cache hit`, await cacheStore.getKeySHA(cachedKey))
-        return { text: cached, finishReason: cachedFinishReason, cached: true }
-    }
-
     const postReq = structuredClone({
         ...req,
         messages: req.messages.map(({ cacheControl, ...rest }) => ({
@@ -428,8 +394,6 @@ export const OpenAIChatCompletion: ChatCompletionHandler = async (
         )
     }
 
-    if (done && finishReason === "stop")
-        await cacheStore.set(cachedKey, { text: chatResp, finishReason })
     return {
         text: chatResp,
         toolCalls,
diff --git a/packages/core/src/parsers.test.ts b/packages/core/src/parsers.test.ts
index aaa0a5ae6e..3e42c5dc74 100644
--- a/packages/core/src/parsers.test.ts
+++ b/packages/core/src/parsers.test.ts
@@ -61,7 +61,7 @@ describe("parsers", async () => {
         )
         let i = 1
         for (const img of result.images) {
-          await writeFile(`./loremipsum.temp.${i++}.png`, img)
+            await writeFile(`./loremipsum.temp.${i++}.png`, img)
         }
         assert(result.file.content.includes("Lorem"))
     })
@@ -120,8 +120,8 @@ describe("parsers", async () => {
     test("hash", async () => {
         const result = await parsers.hash(
             { test: "test string", arr: [1, 2, "32"], v: new Uint8Array(123) },
-            { length: 20 }
+            { length: 20, version: false }
         )
-        assert.strictEqual(result, "2c34c8d7df7428c89c64") // Example hash value
+        assert.strictEqual(result, "889772de6ac65b917519") // Example hash value
     })
 })
diff --git a/packages/core/src/promptcontext.ts b/packages/core/src/promptcontext.ts
index e1ec28c811..513bd9f80f 100644
--- a/packages/core/src/promptcontext.ts
+++ b/packages/core/src/promptcontext.ts
@@ -112,20 +112,20 @@ export async function createPromptContext(
         webSearch: async (q, options) => {
             const { provider, count, ignoreMissingProvider } = options || {}
             // Conduct a web search and return the results
+            const webTrace = trace.startTraceDetails(
+                `🌐 web search <code>${HTMLEscape(q)}</code>`
+            )
             try {
-                trace.startDetails(
-                    `🌐 web search <code>${HTMLEscape(q)}</code>`
-                )
                 let files: WorkspaceFile[]
                 if (provider === "bing")
-                    files = await bingSearch(q, { trace, count })
+                    files = await bingSearch(q, { trace: webTrace, count })
                 else if (provider === "tavily")
-                    files = await tavilySearch(q, { trace, count })
+                    files = await tavilySearch(q, { trace: webTrace, count })
                 else {
                     for (const f of [bingSearch, tavilySearch]) {
                         files = await f(q, {
                             ignoreMissingApiKey: true,
-                            trace,
+                            trace: webTrace,
                             count,
                         })
                         if (files) break
@@ -133,17 +133,17 @@ export async function createPromptContext(
                 }
                 if (!files) {
                     if (ignoreMissingProvider) {
-                        trace.log(`no search provider configured`)
+                        webTrace.log(`no search provider configured`)
                         return undefined
                     }
                     throw new Error(
                         `No search provider configured. See ${DOCS_WEB_SEARCH_URL}.`
                     )
                 }
-                trace.files(files, { model, secrets: env.secrets })
+                webTrace.files(files, { model, secrets: env.secrets })
                 return files
             } finally {
-                trace.endDetails()
+                webTrace.endDetails()
             }
         },
         fuzzSearch: async (q, files_, searchOptions) => {
diff --git a/packages/core/src/trace.ts b/packages/core/src/trace.ts
index 175d804faf..c4e7912a3d 100644
--- a/packages/core/src/trace.ts
+++ b/packages/core/src/trace.ts
@@ -55,7 +55,7 @@ export class MarkdownTrace extends EventTarget implements ToolCallTrace {
             .join("")
     }
 
-    startTraceDetails(title: string) {
+    startTraceDetails(title: string, options?: { expanded?: boolean }) {
         const trace = new MarkdownTrace({ ...this.options })
         trace.addEventListener(TRACE_CHUNK, (ev) =>
             this.dispatchEvent(
@@ -65,7 +65,7 @@ export class MarkdownTrace extends EventTarget implements ToolCallTrace {
         trace.addEventListener(TRACE_DETAILS, () =>
             this.dispatchEvent(new Event(TRACE_DETAILS))
         )
-        trace.startDetails(title)
+        trace.startDetails(title, options)
         this._content.push(trace)
         return trace
     }
diff --git a/packages/core/src/transformers.ts b/packages/core/src/transformers.ts
index f9b70845c9..8ffa686a4c 100644
--- a/packages/core/src/transformers.ts
+++ b/packages/core/src/transformers.ts
@@ -15,10 +15,6 @@ import { deleteUndefinedValues, dotGenaiscriptPath, logVerbose } from "./util"
 import { parseModelIdentifier } from "./models"
 import prettyBytes from "pretty-bytes"
 import { hash } from "./crypto"
-import {
-    ChatCompletionRequestCacheKey,
-    getChatCompletionCache,
-} from "./chatcache"
 import { PLimitPromiseQueue } from "./concurrency"
 
 function progressBar(): ProgressCallback {
@@ -74,35 +70,6 @@ export const TransformersCompletion: ChatCompletionHandler = async (
 
     trace.itemValue("model", model)
 
-    const cache = !!cacheOrName || !!cacheName
-    const cacheStore = getChatCompletionCache(
-        typeof cacheOrName === "string" ? cacheOrName : cacheName
-    )
-    const cachedKey = cache
-        ? <ChatCompletionRequestCacheKey>{
-              ...req,
-              model: req.model,
-              temperature: req.temperature,
-              top_p: req.top_p,
-              max_tokens: req.max_tokens,
-              logit_bias: req.logit_bias,
-          }
-        : undefined
-    trace.itemValue(`caching`, cache)
-    trace.itemValue(`cache`, cacheStore?.name)
-    const { text: cached, finishReason: cachedFinishReason } =
-        (await cacheStore.get(cachedKey)) || {}
-    if (cached !== undefined) {
-        partialCb?.({
-            tokensSoFar: 0, // TODO
-            responseSoFar: cached,
-            responseChunk: cached,
-            inner,
-        })
-        trace.itemValue(`cache hit`, await cacheStore.getKeySHA(cachedKey))
-        return { text: cached, finishReason: cachedFinishReason, cached: true }
-    }
-
     const device =
         process.env.HUGGINGFACE_TRANSFORMERS_DEVICE ||
         process.env.TRANSFORMERS_DEVICE ||
@@ -134,7 +101,9 @@ export const TransformersCompletion: ChatCompletionHandler = async (
                 tokensSoFar,
                 responseSoFar: chatResp,
                 responseChunk: text,
-                responseTokens:  [{ token: text, logprob: Number.NaN } satisfies Logprob],
+                responseTokens: [
+                    { token: text, logprob: Number.NaN } satisfies Logprob,
+                ],
                 inner,
             })
         },
@@ -165,9 +134,6 @@ export const TransformersCompletion: ChatCompletionHandler = async (
     })
 
     const finishReason = "stop"
-    if (finishReason === "stop")
-        await cacheStore.set(cachedKey, { text, finishReason })
-
     return {
         text,
         finishReason: "stop",
diff --git a/packages/core/src/usage.ts b/packages/core/src/usage.ts
index fe84898da1..a0f3b9ae7c 100644
--- a/packages/core/src/usage.ts
+++ b/packages/core/src/usage.ts
@@ -82,6 +82,7 @@ export class GenerationStats {
         messages: ChatCompletionMessageParam[]
         usage: ChatCompletionUsage
         model: string
+        cached: boolean
     }[] = []
 
     /**
@@ -305,31 +306,35 @@ export class GenerationStats {
         const {
             usage = { completion_tokens: 0, prompt_tokens: 0, total_tokens: 0 },
             model,
+            cached,
         } = resp
         const { messages } = req
 
-        this.usage.completion_tokens += usage.completion_tokens ?? 0
-        this.usage.prompt_tokens += usage.prompt_tokens ?? 0
-        this.usage.total_tokens += usage.total_tokens ?? 0
+        if (!cached) {
+            this.usage.completion_tokens += usage.completion_tokens ?? 0
+            this.usage.prompt_tokens += usage.prompt_tokens ?? 0
+            this.usage.total_tokens += usage.total_tokens ?? 0
 
-        this.usage.completion_tokens_details.audio_tokens +=
-            usage.completion_tokens_details?.audio_tokens ?? 0
-        this.usage.completion_tokens_details.reasoning_tokens +=
-            usage.completion_tokens_details?.reasoning_tokens ?? 0
-        this.usage.completion_tokens_details.audio_tokens +=
-            usage.prompt_tokens_details?.audio_tokens ?? 0
-        this.usage.completion_tokens_details.reasoning_tokens +=
-            usage.prompt_tokens_details?.cached_tokens ?? 0
-        this.usage.completion_tokens_details.accepted_prediction_tokens +=
-            usage.completion_tokens_details?.accepted_prediction_tokens ?? 0
-        this.usage.completion_tokens_details.rejected_prediction_tokens +=
-            usage.completion_tokens_details?.rejected_prediction_tokens ?? 0
+            this.usage.completion_tokens_details.audio_tokens +=
+                usage.completion_tokens_details?.audio_tokens ?? 0
+            this.usage.completion_tokens_details.reasoning_tokens +=
+                usage.completion_tokens_details?.reasoning_tokens ?? 0
+            this.usage.completion_tokens_details.audio_tokens +=
+                usage.prompt_tokens_details?.audio_tokens ?? 0
+            this.usage.completion_tokens_details.reasoning_tokens +=
+                usage.prompt_tokens_details?.cached_tokens ?? 0
+            this.usage.completion_tokens_details.accepted_prediction_tokens +=
+                usage.completion_tokens_details?.accepted_prediction_tokens ?? 0
+            this.usage.completion_tokens_details.rejected_prediction_tokens +=
+                usage.completion_tokens_details?.rejected_prediction_tokens ?? 0
+        }
 
         const { provider } = parseModelIdentifier(this.model)
         const chatTurn = {
             messages: structuredClone(messages),
             usage: structuredClone(usage),
             model: `${provider}:${model}`,
+            cached,
         }
         this.chatTurns.push(chatTurn)
     }
diff --git a/packages/sample/genaisrc/cache.genai.mts b/packages/sample/genaisrc/cache.genai.mts
index b85e951eca..369bcf4717 100644
--- a/packages/sample/genaisrc/cache.genai.mts
+++ b/packages/sample/genaisrc/cache.genai.mts
@@ -23,6 +23,15 @@ for (const cache of [
 
 console.log(`cache test passed`)
 
+const innerPrompt = `Generate 2 word poem. ${Math.random()}`
+await Promise.all(
+    Array(10)
+        .fill(0)
+        .map(async (_, i) => {
+            await runPrompt(innerPrompt, { cache: "inner", label: `run-${i}` })
+        })
+)
+
 $`Generate 2 word poem.`
 
 defOutputProcessor(async ({ text }) => {
diff --git a/packages/sample/src/cli.test.ts b/packages/sample/src/cli.test.ts
index 759b4b661b..113aace06f 100644
--- a/packages/sample/src/cli.test.ts
+++ b/packages/sample/src/cli.test.ts
@@ -36,7 +36,11 @@ describe("scripts", async () => {
     const cmd = "scripts"
     await test("list", async () => {
         const res = await $`node ${cli} ${cmd} list`
-        assert(res.stdout.includes("id: poem"))
+        const d = JSON.parse(res.stdout)
+        assert(d.find((s) => s.id === "poem"))
+        assert(d.find((s) => s.id === "system"))
+        assert(d.find((s) => s.id === "system.output_markdown"))
+        assert(!d.some((s) => s.system && s.filename))
     })
     await test("create foobar", async () => {
         const res = await $`node ${cli} ${cmd} create foobar`
diff --git a/packages/vscode/src/llmrequesttree.ts b/packages/vscode/src/llmrequesttree.ts
index f8487607db..60d06c770d 100644
--- a/packages/vscode/src/llmrequesttree.ts
+++ b/packages/vscode/src/llmrequesttree.ts
@@ -2,17 +2,19 @@ import * as vscode from "vscode"
 import { ExtensionState } from "./state"
 import { infoUri } from "./markdowndocumentprovider"
 import { CacheEntry } from "../../core/src/cache"
-import { CreateChatCompletionRequest } from "../../core/src/chattypes"
+import {
+    ChatCompletionResponse,
+    CreateChatCompletionRequest,
+} from "../../core/src/chattypes"
 import { CHANGE, CACHE_LLMREQUEST_PREFIX } from "../../core/src/constants"
 import {
     ChatCompletationRequestCache,
-    ChatCompletationRequestCacheValue,
     getChatCompletionCache,
 } from "../../core/src/chatcache"
 
 type LLMRequestTreeNode = CacheEntry<
     CreateChatCompletionRequest,
-    ChatCompletationRequestCacheValue
+    ChatCompletionResponse
 >
 
 class LLMRequestTreeDataProvider