@code-yeongyu/senpi 2026.7.9 → 2026.7.10-2

package/dist/core/cache-stats.d.ts

@@ -0,0 +1,49 @@
+import type { AssistantMessage } from "@earendil-works/pi-ai";
+import type { SessionEntry } from "./session-manager.ts";
+/**
+ * Prompt-cache TTL: idle gaps longer than this are worth mentioning as the
+ * likely cause of a miss. Anthropic's default cache TTL is 5 minutes.
+ */
+export declare const CACHE_TTL_MS: number;
+/** A counted cache miss on a single assistant message. */
+export interface CacheMiss {
+    /** Prompt tokens that were in the previous turn's prompt but not read from cache. */
+    missedTokens: number;
+    /** Extra dollars paid vs. a full cache hit; 0 when pricing is unknown. */
+    missedCost: number;
+    /** Milliseconds since the previous request (which last refreshed the cache). */
+    idleMs: number;
+    /** True when the model changed relative to the previous request. */
+    modelChanged: boolean;
+}
+export interface CacheWasteTotals {
+    missedTokens: number;
+    missedCost: number;
+    /** Number of counted misses (turns above the noise floor). */
+    missCount: number;
+}
+/** Minimal pricing lookup, satisfied by ModelRegistry. Cost is $/million tokens. */
+export interface ModelPriceSource {
+    find(provider: string, modelId: string): {
+        cost: {
+            cacheRead: number;
+        };
+    } | undefined;
+}
+/**
+ * Cumulative cache waste across a session: prompt tokens that should have been
+ * cache reads (they were in the previous turn's prompt) but were re-billed.
+ */
+export declare function computeCacheWaste(entries: SessionEntry[], models: ModelPriceSource): CacheWasteTotals;
+/**
+ * All counted cache misses across a session, keyed by the assistant message
+ * (by reference) that paid for them. Used to re-derive transcript notices when
+ * rebuilding the chat from entries (resume, post-compaction rebuild).
+ */
+export declare function collectCacheMisses(entries: SessionEntry[], models: ModelPriceSource): Map<AssistantMessage, CacheMiss>;
+/**
+ * Detect a cache miss on a just-completed assistant message.
+ * `entries` must not yet contain `message` (message_end fires before persistence).
+ */
+export declare function detectCacheMiss(entries: SessionEntry[], message: AssistantMessage, models: ModelPriceSource): CacheMiss | undefined;
+//# sourceMappingURL=cache-stats.d.ts.map

package/dist/core/cache-stats.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cache-stats.d.ts","sourceRoot":"","sources":["../../src/core/cache-stats.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAC9D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AAEzD;;;GAGG;AACH,eAAO,MAAM,YAAY,QAAgB,CAAC;AAK1C,0DAA0D;AAC1D,MAAM,WAAW,SAAS;IACzB,qFAAqF;IACrF,YAAY,EAAE,MAAM,CAAC;IACrB,0EAA0E;IAC1E,UAAU,EAAE,MAAM,CAAC;IACnB,gFAAgF;IAChF,MAAM,EAAE,MAAM,CAAC;IACf,oEAAoE;IACpE,YAAY,EAAE,OAAO,CAAC;CACtB;AAED,MAAM,WAAW,gBAAgB;IAChC,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,CAAC;IACnB,8DAA8D;IAC9D,SAAS,EAAE,MAAM,CAAC;CAClB;AAED,oFAAoF;AACpF,MAAM,WAAW,gBAAgB;IAChC,IAAI,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG;QAAE,IAAI,EAAE;YAAE,SAAS,EAAE,MAAM,CAAA;SAAE,CAAA;KAAE,GAAG,SAAS,CAAC;CACrF;AAmGD;;;GAGG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,gBAAgB,GAAG,gBAAgB,CAErG;AAED;;;;GAIG;AACH,wBAAgB,kBAAkB,CACjC,OAAO,EAAE,YAAY,EAAE,EACvB,MAAM,EAAE,gBAAgB,GACtB,GAAG,CAAC,gBAAgB,EAAE,SAAS,CAAC,CAElC;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAC9B,OAAO,EAAE,YAAY,EAAE,EACvB,OAAO,EAAE,gBAAgB,EACzB,MAAM,EAAE,gBAAgB,GACtB,SAAS,GAAG,SAAS,CAEvB"}

package/dist/core/cache-stats.js

@@ -0,0 +1,101 @@
+/**
+ * Prompt-cache TTL: idle gaps longer than this are worth mentioning as the
+ * likely cause of a miss. Anthropic's default cache TTL is 5 minutes.
+ */
+export const CACHE_TTL_MS = 5 * 60 * 1000;
+/** Per-turn misses at or below this are cache breakpoint granularity noise. */
+const NOISE_FLOOR_TOKENS = 1024;
+/**
+ * Compute the cache miss for one assistant message relative to the previous
+ * request. Returns undefined when nothing is counted: first turn, after a
+ * reset, no cache activity ever reported (provider without cache support), or
+ * miss below the noise floor.
+ */
+function detectMiss(prev, message, models) {
+    const usage = message.usage;
+    const promptTokens = usage.input + usage.cacheRead + usage.cacheWrite;
+    // A zero-cache turn only counts when cache activity was reported before:
+    // on cache-read-only providers that is a total miss, while on providers
+    // that never report caching it means nothing.
+    if (!prev || promptTokens <= 0 || (usage.cacheRead + usage.cacheWrite === 0 && !prev.reportedCache)) {
+        return undefined;
+    }
+    const missedTokens = Math.min(prev.promptTokens, promptTokens) - usage.cacheRead;
+    if (missedTokens <= NOISE_FLOOR_TOKENS)
+        return undefined;
+    // Extra cost = missed tokens billed at the actual paid rate (input/cacheWrite,
+    // incl. write premium) instead of the cache-read rate. Missed tokens can only
+    // land in the input or cacheWrite buckets, so the paid rate comes straight
+    // from this message's own cost breakdown.
+    const paidTokens = usage.input + usage.cacheWrite;
+    const paidPerToken = paidTokens > 0 ? (usage.cost.input + usage.cost.cacheWrite) / paidTokens : 0;
+    const readPerToken = usage.cacheRead > 0
+        ? usage.cost.cacheRead / usage.cacheRead
+        : (models.find(message.provider, message.model)?.cost.cacheRead ?? 0) / 1_000_000;
+    return {
+        missedTokens,
+        missedCost: missedTokens * Math.max(0, paidPerToken - readPerToken),
+        idleMs: Math.max(0, message.timestamp - prev.timestamp),
+        modelChanged: `${message.provider}/${message.model}` !== prev.modelKey,
+    };
+}
+function asPreviousRequest(message, reportedCache) {
+    const usage = message.usage;
+    const promptTokens = usage.input + usage.cacheRead + usage.cacheWrite;
+    if (promptTokens <= 0)
+        return undefined;
+    return {
+        promptTokens,
+        modelKey: `${message.provider}/${message.model}`,
+        timestamp: message.timestamp,
+        reportedCache: reportedCache || usage.cacheRead + usage.cacheWrite > 0,
+    };
+}
+function scan(entries, models) {
+    let prev;
+    const totals = { missedTokens: 0, missedCost: 0, missCount: 0 };
+    const misses = new Map();
+    for (const entry of entries) {
+        if (entry.type === "compaction" || entry.type === "branch_summary") {
+            // The context legitimately changed; the next turn's prompt is new content,
+            // not re-billed content. Model switches are NOT exempt: they re-bill the
+            // full prompt and should be counted.
+            prev = undefined;
+            continue;
+        }
+        if (entry.type === "message" && entry.message.role === "assistant") {
+            const miss = detectMiss(prev, entry.message, models);
+            if (miss) {
+                totals.missedTokens += miss.missedTokens;
+                totals.missedCost += miss.missedCost;
+                totals.missCount += 1;
+                misses.set(entry.message, miss);
+            }
+            prev = asPreviousRequest(entry.message, prev?.reportedCache ?? false) ?? prev;
+        }
+    }
+    return { prev, totals, misses };
+}
+/**
+ * Cumulative cache waste across a session: prompt tokens that should have been
+ * cache reads (they were in the previous turn's prompt) but were re-billed.
+ */
+export function computeCacheWaste(entries, models) {
+    return scan(entries, models).totals;
+}
+/**
+ * All counted cache misses across a session, keyed by the assistant message
+ * (by reference) that paid for them. Used to re-derive transcript notices when
+ * rebuilding the chat from entries (resume, post-compaction rebuild).
+ */
+export function collectCacheMisses(entries, models) {
+    return scan(entries, models).misses;
+}
+/**
+ * Detect a cache miss on a just-completed assistant message.
+ * `entries` must not yet contain `message` (message_end fires before persistence).
+ */
+export function detectCacheMiss(entries, message, models) {
+    return detectMiss(scan(entries, models).prev, message, models);
+}
+//# sourceMappingURL=cache-stats.js.map

package/dist/core/cache-stats.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"cache-stats.js","sourceRoot":"","sources":["../../src/core/cache-stats.ts"],"names":[],"mappings":"AAGA;;;GAGG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC;AAE1C,+EAA+E;AAC/E,MAAM,kBAAkB,GAAG,IAAI,CAAC;AAuChC;;;;;GAKG;AACH,SAAS,UAAU,CAClB,IAAiC,EACjC,OAAyB,EACzB,MAAwB;IAExB,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;IAC5B,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,SAAS,GAAG,KAAK,CAAC,UAAU,CAAC;IACtE,yEAAyE;IACzE,wEAAwE;IACxE,8CAA8C;IAC9C,IAAI,CAAC,IAAI,IAAI,YAAY,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,GAAG,KAAK,CAAC,UAAU,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;QACrG,OAAO,SAAS,CAAC;IAClB,CAAC;IAED,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,YAAY,EAAE,YAAY,CAAC,GAAG,KAAK,CAAC,SAAS,CAAC;IACjF,IAAI,YAAY,IAAI,kBAAkB;QAAE,OAAO,SAAS,CAAC;IAEzD,+EAA+E;IAC/E,8EAA8E;IAC9E,2EAA2E;IAC3E,0CAA0C;IAC1C,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,UAAU,CAAC;IAClD,MAAM,YAAY,GAAG,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC;IAClG,MAAM,YAAY,GACjB,KAAK,CAAC,SAAS,GAAG,CAAC;QAClB,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC,SAAS;QACxC,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,EAAE,OAAO,CAAC,KAAK,CAAC,EAAE,IAAI,CAAC,SAAS,IAAI,CAAC,CAAC,GAAG,SAAS,CAAC;IAEpF,OAAO;QACN,YAAY;QACZ,UAAU,EAAE,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,YAAY,GAAG,YAAY,CAAC;QACnE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC;QACvD,YAAY,EAAE,GAAG,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,KAAK,EAAE,KAAK,IAAI,CAAC,QAAQ;KACtE,CAAC;AACH,CAAC;AAED,SAAS,iBAAiB,CAAC,OAAyB,EAAE,aAAsB;IAC3E,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC;IAC5B,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,SAAS,GAAG,KAAK,CAAC,UAAU,CAAC;IACtE,IAAI,YAAY,IAAI,CAAC;QAAE,OAAO,SAAS,CAAC;IACxC,OAAO;QACN,YAAY;QACZ,QAAQ,EAAE,GAAG,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,KAAK,EAAE;QAChD,SAAS,EAAE,OAAO,CAAC,SAAS;QAC5B,aAAa,EAAE,aAAa,IAAI,KAAK,CAAC,SAAS,GAAG,KAAK,CAAC,UAAU,GAAG,CAAC;KACtE,CAAC;AACH,CAAC;AAED,SAAS,IAAI,CACZ,OAAuB,EACvB,MAAwB;IAExB,IAAI,IAAiC,CAAC;IACtC,MAAM,MAAM,GAAqB,EAAE,YAAY,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,CAAC,EAAE,CAAC;IAClF,MAAM,MAAM,GAAG,IAAI,GAAG,EAA+B,CAAC;IAEtD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC7B,IAAI,KAAK,CAAC,IAAI,KAAK,YAAY,IAAI,KAAK,CAAC,IAAI,KAAK,gBAAgB,EAAE,CAAC;YACpE,2EAA2E;YAC3E,yEAAyE;YACzE,qCAAqC;YACrC,IAAI,GAAG,SAAS,CAAC;YACjB,SAAS;QACV,CAAC;QACD,IAAI,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,OAAO,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;YACpE,MAAM,IAAI,GAAG,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;YACrD,IAAI,IAAI,EAAE,CAAC;gBACV,MAAM,CAAC,YAAY,IAAI,IAAI,CAAC,YAAY,CAAC;gBACzC,MAAM,CAAC,UAAU,IAAI,IAAI,CAAC,UAAU,CAAC;gBACrC,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;gBACtB,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;YACjC,CAAC;YACD,IAAI,GAAG,iBAAiB,CAAC,KAAK,CAAC,OAAO,EAAE,IAAI,EAAE,aAAa,IAAI,KAAK,CAAC,IAAI,IAAI,CAAC;QAC/E,CAAC;IACF,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC;AACjC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAuB,EAAE,MAAwB;IAClF,OAAO,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,MAAM,CAAC;AACrC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,kBAAkB,CACjC,OAAuB,EACvB,MAAwB;IAExB,OAAO,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,MAAM,CAAC;AACrC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,eAAe,CAC9B,OAAuB,EACvB,OAAyB,EACzB,MAAwB;IAExB,OAAO,UAAU,CAAC,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,IAAI,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;AAChE,CAAC","sourcesContent":["import type { AssistantMessage } from \"@earendil-works/pi-ai\";\nimport type { SessionEntry } from \"./session-manager.ts\";\n\n/**\n * Prompt-cache TTL: idle gaps longer than this are worth mentioning as the\n * likely cause of a miss. Anthropic's default cache TTL is 5 minutes.\n */\nexport const CACHE_TTL_MS = 5 * 60 * 1000;\n\n/** Per-turn misses at or below this are cache breakpoint granularity noise. */\nconst NOISE_FLOOR_TOKENS = 1024;\n\n/** A counted cache miss on a single assistant message. */\nexport interface CacheMiss {\n\t/** Prompt tokens that were in the previous turn's prompt but not read from cache. */\n\tmissedTokens: number;\n\t/** Extra dollars paid vs. a full cache hit; 0 when pricing is unknown. */\n\tmissedCost: number;\n\t/** Milliseconds since the previous request (which last refreshed the cache). */\n\tidleMs: number;\n\t/** True when the model changed relative to the previous request. */\n\tmodelChanged: boolean;\n}\n\nexport interface CacheWasteTotals {\n\tmissedTokens: number;\n\tmissedCost: number;\n\t/** Number of counted misses (turns above the noise floor). */\n\tmissCount: number;\n}\n\n/** Minimal pricing lookup, satisfied by ModelRegistry. Cost is $/million tokens. */\nexport interface ModelPriceSource {\n\tfind(provider: string, modelId: string): { cost: { cacheRead: number } } | undefined;\n}\n\n/** The last request seen by the scan; everything in its prompt should be cached. */\ninterface PreviousRequest {\n\tpromptTokens: number;\n\tmodelKey: string;\n\ttimestamp: number;\n\t/**\n\t * Sticky: some earlier request in this scan segment reported cache activity.\n\t * Distinguishes a total miss on a cache-read-only provider (OpenAI-style,\n\t * writes unreported) from a provider that never reports caching at all.\n\t */\n\treportedCache: boolean;\n}\n\n/**\n * Compute the cache miss for one assistant message relative to the previous\n * request. Returns undefined when nothing is counted: first turn, after a\n * reset, no cache activity ever reported (provider without cache support), or\n * miss below the noise floor.\n */\nfunction detectMiss(\n\tprev: PreviousRequest | undefined,\n\tmessage: AssistantMessage,\n\tmodels: ModelPriceSource,\n): CacheMiss | undefined {\n\tconst usage = message.usage;\n\tconst promptTokens = usage.input + usage.cacheRead + usage.cacheWrite;\n\t// A zero-cache turn only counts when cache activity was reported before:\n\t// on cache-read-only providers that is a total miss, while on providers\n\t// that never report caching it means nothing.\n\tif (!prev || promptTokens <= 0 || (usage.cacheRead + usage.cacheWrite === 0 && !prev.reportedCache)) {\n\t\treturn undefined;\n\t}\n\n\tconst missedTokens = Math.min(prev.promptTokens, promptTokens) - usage.cacheRead;\n\tif (missedTokens <= NOISE_FLOOR_TOKENS) return undefined;\n\n\t// Extra cost = missed tokens billed at the actual paid rate (input/cacheWrite,\n\t// incl. write premium) instead of the cache-read rate. Missed tokens can only\n\t// land in the input or cacheWrite buckets, so the paid rate comes straight\n\t// from this message's own cost breakdown.\n\tconst paidTokens = usage.input + usage.cacheWrite;\n\tconst paidPerToken = paidTokens > 0 ? (usage.cost.input + usage.cost.cacheWrite) / paidTokens : 0;\n\tconst readPerToken =\n\t\tusage.cacheRead > 0\n\t\t\t? usage.cost.cacheRead / usage.cacheRead\n\t\t\t: (models.find(message.provider, message.model)?.cost.cacheRead ?? 0) / 1_000_000;\n\n\treturn {\n\t\tmissedTokens,\n\t\tmissedCost: missedTokens * Math.max(0, paidPerToken - readPerToken),\n\t\tidleMs: Math.max(0, message.timestamp - prev.timestamp),\n\t\tmodelChanged: `${message.provider}/${message.model}` !== prev.modelKey,\n\t};\n}\n\nfunction asPreviousRequest(message: AssistantMessage, reportedCache: boolean): PreviousRequest | undefined {\n\tconst usage = message.usage;\n\tconst promptTokens = usage.input + usage.cacheRead + usage.cacheWrite;\n\tif (promptTokens <= 0) return undefined;\n\treturn {\n\t\tpromptTokens,\n\t\tmodelKey: `${message.provider}/${message.model}`,\n\t\ttimestamp: message.timestamp,\n\t\treportedCache: reportedCache || usage.cacheRead + usage.cacheWrite > 0,\n\t};\n}\n\nfunction scan(\n\tentries: SessionEntry[],\n\tmodels: ModelPriceSource,\n): { prev: PreviousRequest | undefined; totals: CacheWasteTotals; misses: Map<AssistantMessage, CacheMiss> } {\n\tlet prev: PreviousRequest | undefined;\n\tconst totals: CacheWasteTotals = { missedTokens: 0, missedCost: 0, missCount: 0 };\n\tconst misses = new Map<AssistantMessage, CacheMiss>();\n\n\tfor (const entry of entries) {\n\t\tif (entry.type === \"compaction\" || entry.type === \"branch_summary\") {\n\t\t\t// The context legitimately changed; the next turn's prompt is new content,\n\t\t\t// not re-billed content. Model switches are NOT exempt: they re-bill the\n\t\t\t// full prompt and should be counted.\n\t\t\tprev = undefined;\n\t\t\tcontinue;\n\t\t}\n\t\tif (entry.type === \"message\" && entry.message.role === \"assistant\") {\n\t\t\tconst miss = detectMiss(prev, entry.message, models);\n\t\t\tif (miss) {\n\t\t\t\ttotals.missedTokens += miss.missedTokens;\n\t\t\t\ttotals.missedCost += miss.missedCost;\n\t\t\t\ttotals.missCount += 1;\n\t\t\t\tmisses.set(entry.message, miss);\n\t\t\t}\n\t\t\tprev = asPreviousRequest(entry.message, prev?.reportedCache ?? false) ?? prev;\n\t\t}\n\t}\n\treturn { prev, totals, misses };\n}\n\n/**\n * Cumulative cache waste across a session: prompt tokens that should have been\n * cache reads (they were in the previous turn's prompt) but were re-billed.\n */\nexport function computeCacheWaste(entries: SessionEntry[], models: ModelPriceSource): CacheWasteTotals {\n\treturn scan(entries, models).totals;\n}\n\n/**\n * All counted cache misses across a session, keyed by the assistant message\n * (by reference) that paid for them. Used to re-derive transcript notices when\n * rebuilding the chat from entries (resume, post-compaction rebuild).\n */\nexport function collectCacheMisses(\n\tentries: SessionEntry[],\n\tmodels: ModelPriceSource,\n): Map<AssistantMessage, CacheMiss> {\n\treturn scan(entries, models).misses;\n}\n\n/**\n * Detect a cache miss on a just-completed assistant message.\n * `entries` must not yet contain `message` (message_end fires before persistence).\n */\nexport function detectCacheMiss(\n\tentries: SessionEntry[],\n\tmessage: AssistantMessage,\n\tmodels: ModelPriceSource,\n): CacheMiss | undefined {\n\treturn detectMiss(scan(entries, models).prev, message, models);\n}\n"]}

package/dist/core/compaction/compaction.d.ts

@@ -66,9 +66,8 @@ export declare function shouldCompact(contextTokens: number, contextWindow: numb
  */
 export declare function estimateTokens(message: AgentMessage): number;
 /**
- * Find the user message (or bashExecution) that starts the turn containing the given entry index.
+ * Find the context-visible user-role message that starts the turn containing the given entry index.
  * Returns -1 if no turn start found before the index.
- * BashExecutionMessage is treated like a user message for turn boundaries.
  */
 export declare function findTurnStartIndex(entries: SessionEntry[], entryIndex: number, startIndex: number): number;
 export interface CutPointResult {

package/dist/core/compaction/compaction.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"compaction.d.ts","sourceRoot":"","sources":["../../../src/core/compaction/compaction.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,+BAA+B,CAAC;AAC3F,OAAO,KAAK,EAA6B,KAAK,EAAuB,KAAK,EAAE,MAAM,8BAA8B,CAAC;AAUjH,OAAO,EAA6C,KAAK,YAAY,EAAE,MAAM,uBAAuB,CAAC;AACrG,OAAO,EAIN,KAAK,cAAc,EAInB,MAAM,YAAY,CAAC;AAWpB,kEAAkE;AAClE,MAAM,WAAW,iBAAiB;IACjC,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,aAAa,EAAE,MAAM,EAAE,CAAC;CACxB;AA0ED,8EAA8E;AAC9E,MAAM,WAAW,gBAAgB,CAAC,CAAC,GAAG,OAAO;IAC5C,OAAO,EAAE,MAAM,CAAC;IAChB,gBAAgB,EAAE,MAAM,CAAC;IACzB,YAAY,EAAE,MAAM,CAAC;IACrB,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B,+FAA+F;IAC/F,OAAO,CAAC,EAAE,CAAC,CAAC;CACZ;AAMD,MAAM,WAAW,kBAAkB;IAClC,OAAO,EAAE,OAAO,CAAC;IACjB,aAAa,EAAE,MAAM,CAAC;IACtB,gBAAgB,EAAE,MAAM,CAAC;IACzB,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,qBAAqB,CAAC,EAAE,MAAM,CAAC;IAC/B,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,2BAA2B,CAAC,EAAE,MAAM,CAAC;IACrC,yBAAyB,CAAC,EAAE,MAAM,CAAC;IACnC,uBAAuB,CAAC,EAAE,MAAM,CAAC;CACjC;AAED,eAAO,MAAM,2BAA2B,EAAE,kBAYzC,CAAC;AAMF;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,KAAK,EAAE,KAAK,GAAG,MAAM,CAE3D;AAqBD;;GAEG;AACH,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,YAAY,EAAE,GAAG,KAAK,GAAG,SAAS,CAShF;AAED,MAAM,WAAW,oBAAoB;IACpC,MAAM,EAAE,MAAM,CAAC;IACf,WAAW,EAAE,MAAM,CAAC;IACpB,cAAc,EAAE,MAAM,CAAC;IACvB,cAAc,EAAE,MAAM,GAAG,IAAI,CAAC;CAC9B;AAUD;;;GAGG;AACH,wBAAgB,qBAAqB,CAAC,QAAQ,EAAE,YAAY,EAAE,GAAG,oBAAoB,CA4BpF;AAED;;GAEG;AACH,wBAAgB,aAAa,CAAC,aAAa,EAAE,MAAM,EAAE,aAAa,EAAE,MAAM,EAAE,QAAQ,EAAE,kBAAkB,GAAG,OAAO,CAGjH;AAwBD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,MAAM,CAsC5D;AAiDD;;;;GAIG;AACH,wBAAgB,kBAAkB,CAAC,OAAO,EAAE,YAAY,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,MAAM,CAe1G;AAED,MAAM,WAAW,cAAc;IAC9B,mCAAmC;IACnC,mBAAmB,EAAE,MAAM,CAAC;IAC5B,qFAAqF;IACrF,cAAc,EAAE,MAAM,CAAC;IACvB,uEAAuE;IACvE,WAAW,EAAE,OAAO,CAAC;CACrB;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,YAAY,CAC3B,OAAO,EAAE,YAAY,EAAE,EACvB,UAAU,EAAE,MAAM,EAClB,QAAQ,EAAE,MAAM,EAChB,gBAAgB,EAAE,MAAM,GACtB,cAAc,CAiEhB;AA4GD;;;GAGG;AACH,wBAAsB,eAAe,CACpC,eAAe,EAAE,YAAY,EAAE,EAC/B,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EACjB,aAAa,EAAE,MAAM,EACrB,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAChC,MAAM,CAAC,EAAE,WAAW,EACpB,kBAAkB,CAAC,EAAE,MAAM,EAC3B,eAAe,CAAC,EAAE,MAAM,EACxB,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACnC,aAAa,CAAC,EAAE,aAAa,EAC7B,QAAQ,CAAC,EAAE,QAAQ,EACnB,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAC1B,OAAO,CAAC,MAAM,CAAC,CA2DjB;AAMD,MAAM,WAAW,qBAAqB;IACrC,kCAAkC;IAClC,gBAAgB,EAAE,MAAM,CAAC;IACzB,qDAAqD;IACrD,mBAAmB,EAAE,YAAY,EAAE,CAAC;IACpC,2EAA2E;IAC3E,kBAAkB,EAAE,YAAY,EAAE,CAAC;IACnC,iEAAiE;IACjE,WAAW,EAAE,OAAO,CAAC;IACrB,YAAY,EAAE,MAAM,CAAC;IACrB,6DAA6D;IAC7D,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,yDAAyD;IACzD,OAAO,EAAE,cAAc,CAAC;IACxB,8CAA8C;IAC9C,QAAQ,EAAE,kBAAkB,CAAC;CAC7B;AAED,wBAAgB,iBAAiB,CAChC,WAAW,EAAE,YAAY,EAAE,EAC3B,QAAQ,EAAE,kBAAkB,GAC1B,qBAAqB,GAAG,SAAS,CA8EnC;AAqBD;;;;;;GAMG;AACH,wBAAsB,OAAO,CAC5B,WAAW,EAAE,qBAAqB,EAClC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EACjB,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAChC,kBAAkB,CAAC,EAAE,MAAM,EAC3B,MAAM,CAAC,EAAE,WAAW,EACpB,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACnC,aAAa,CAAC,EAAE,aAAa,EAC7B,QAAQ,CAAC,EAAE,QAAQ,EACnB,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAC1B,OAAO,CAAC,gBAAgB,CAAC,CA+E3B"}
+{"version":3,"file":"compaction.d.ts","sourceRoot":"","sources":["../../../src/core/compaction/compaction.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,+BAA+B,CAAC;AAC3F,OAAO,KAAK,EAA6B,KAAK,EAAuB,KAAK,EAAE,MAAM,8BAA8B,CAAC;AAGjH,OAAO,EAGN,KAAK,YAAY,EAEjB,MAAM,uBAAuB,CAAC;AAC/B,OAAO,EAIN,KAAK,cAAc,EAInB,MAAM,YAAY,CAAC;AAWpB,kEAAkE;AAClE,MAAM,WAAW,iBAAiB;IACjC,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,aAAa,EAAE,MAAM,EAAE,CAAC;CACxB;AAwDD,8EAA8E;AAC9E,MAAM,WAAW,gBAAgB,CAAC,CAAC,GAAG,OAAO;IAC5C,OAAO,EAAE,MAAM,CAAC;IAChB,gBAAgB,EAAE,MAAM,CAAC;IACzB,YAAY,EAAE,MAAM,CAAC;IACrB,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B,+FAA+F;IAC/F,OAAO,CAAC,EAAE,CAAC,CAAC;CACZ;AAMD,MAAM,WAAW,kBAAkB;IAClC,OAAO,EAAE,OAAO,CAAC;IACjB,aAAa,EAAE,MAAM,CAAC;IACtB,gBAAgB,EAAE,MAAM,CAAC;IACzB,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,qBAAqB,CAAC,EAAE,MAAM,CAAC;IAC/B,kBAAkB,CAAC,EAAE,OAAO,CAAC;IAC7B,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,2BAA2B,CAAC,EAAE,MAAM,CAAC;IACrC,yBAAyB,CAAC,EAAE,MAAM,CAAC;IACnC,uBAAuB,CAAC,EAAE,MAAM,CAAC;CACjC;AAED,eAAO,MAAM,2BAA2B,EAAE,kBAYzC,CAAC;AAMF;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,KAAK,EAAE,KAAK,GAAG,MAAM,CAE3D;AAqBD;;GAEG;AACH,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,YAAY,EAAE,GAAG,KAAK,GAAG,SAAS,CAShF;AAED,MAAM,WAAW,oBAAoB;IACpC,MAAM,EAAE,MAAM,CAAC;IACf,WAAW,EAAE,MAAM,CAAC;IACpB,cAAc,EAAE,MAAM,CAAC;IACvB,cAAc,EAAE,MAAM,GAAG,IAAI,CAAC;CAC9B;AAUD;;;GAGG;AACH,wBAAgB,qBAAqB,CAAC,QAAQ,EAAE,YAAY,EAAE,GAAG,oBAAoB,CA4BpF;AAED;;GAEG;AACH,wBAAgB,aAAa,CAAC,aAAa,EAAE,MAAM,EAAE,aAAa,EAAE,MAAM,EAAE,QAAQ,EAAE,kBAAkB,GAAG,OAAO,CAGjH;AAwBD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,MAAM,CAsC5D;AA2DD;;;GAGG;AACH,wBAAgB,kBAAkB,CAAC,OAAO,EAAE,YAAY,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,MAAM,CAO1G;AAED,MAAM,WAAW,cAAc;IAC9B,mCAAmC;IACnC,mBAAmB,EAAE,MAAM,CAAC;IAC5B,qFAAqF;IACrF,cAAc,EAAE,MAAM,CAAC;IACvB,uEAAuE;IACvE,WAAW,EAAE,OAAO,CAAC;CACrB;AAED;;;;;;;;;;;;;;;GAeG;AACH,wBAAgB,YAAY,CAC3B,OAAO,EAAE,YAAY,EAAE,EACvB,UAAU,EAAE,MAAM,EAClB,QAAQ,EAAE,MAAM,EAChB,gBAAgB,EAAE,MAAM,GACtB,cAAc,CA6DhB;AA4GD;;;GAGG;AACH,wBAAsB,eAAe,CACpC,eAAe,EAAE,YAAY,EAAE,EAC/B,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EACjB,aAAa,EAAE,MAAM,EACrB,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAChC,MAAM,CAAC,EAAE,WAAW,EACpB,kBAAkB,CAAC,EAAE,MAAM,EAC3B,eAAe,CAAC,EAAE,MAAM,EACxB,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACnC,aAAa,CAAC,EAAE,aAAa,EAC7B,QAAQ,CAAC,EAAE,QAAQ,EACnB,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAC1B,OAAO,CAAC,MAAM,CAAC,CA2DjB;AAMD,MAAM,WAAW,qBAAqB;IACrC,kCAAkC;IAClC,gBAAgB,EAAE,MAAM,CAAC;IACzB,qDAAqD;IACrD,mBAAmB,EAAE,YAAY,EAAE,CAAC;IACpC,2EAA2E;IAC3E,kBAAkB,EAAE,YAAY,EAAE,CAAC;IACnC,iEAAiE;IACjE,WAAW,EAAE,OAAO,CAAC;IACrB,YAAY,EAAE,MAAM,CAAC;IACrB,6DAA6D;IAC7D,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,yDAAyD;IACzD,OAAO,EAAE,cAAc,CAAC;IACxB,8CAA8C;IAC9C,QAAQ,EAAE,kBAAkB,CAAC;CAC7B;AAED,wBAAgB,iBAAiB,CAChC,WAAW,EAAE,YAAY,EAAE,EAC3B,QAAQ,EAAE,kBAAkB,GAC1B,qBAAqB,GAAG,SAAS,CA8EnC;AAqBD;;;;;;GAMG;AACH,wBAAsB,OAAO,CAC5B,WAAW,EAAE,qBAAqB,EAClC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EACjB,MAAM,EAAE,MAAM,GAAG,SAAS,EAC1B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAChC,kBAAkB,CAAC,EAAE,MAAM,EAC3B,MAAM,CAAC,EAAE,WAAW,EACpB,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACnC,aAAa,CAAC,EAAE,aAAa,EAC7B,QAAQ,CAAC,EAAE,QAAQ,EACnB,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAC1B,OAAO,CAAC,gBAAgB,CAAC,CA+E3B"}

package/dist/core/compaction/compaction.js

@@ -5,8 +5,8 @@
  * and after compaction the session is reloaded.
  */
 import { completeSimple } from "@earendil-works/pi-ai/compat";
-import { convertToLlm, createBranchSummaryMessage, createCompactionSummaryMessage, createCustomMessage, filterContextExcludedMessages, isContextExcludedCustomMessage, } from "../messages.js";
-import { buildSessionContext } from "../session-manager.js";
+import { convertToLlm, filterContextExcludedMessages, isContextExcludedCustomMessage } from "../messages.js";
+import { buildSessionContext, sessionEntryToContextMessages, } from "../session-manager.js";
 import { computeFileLists, createFileOps, extractFileOpsFromMessage, formatFileOperations, SUMMARIZATION_SYSTEM_PROMPT, serializeConversation, } from "./utils.js";
 /**
  * Extract file operations from messages and previous compaction entries.
@@ -42,32 +42,14 @@ function extractFileOperations(messages, entries, prevCompactionIndex) {
  * Extract AgentMessage from an entry if it produces one.
  * Returns undefined for entries that don't contribute to LLM context.
  */
-function getMessageFromEntry(entry) {
-    if (entry.type === "message") {
-        return entry.message;
-    }
-    if (entry.type === "custom_message") {
-        if (isContextExcludedCustomMessage(entry.customType)) {
-            return undefined;
-        }
-        return createCustomMessage(entry.customType, entry.content, entry.display, entry.details, entry.timestamp);
-    }
-    if (entry.type === "branch_summary") {
-        return createBranchSummaryMessage(entry.summary, entry.fromId, entry.timestamp);
-    }
+function contextMessagesForCompactionEntry(entry) {
     if (entry.type === "compaction") {
-        return createCompactionSummaryMessage(entry.summary, entry.tokensBefore, entry.timestamp);
+        return [];
     }
-    return undefined;
-}
-function isCutPointCustomMessage(entry) {
-    return entry.type === "custom_message" && !isContextExcludedCustomMessage(entry.customType);
+    return sessionEntryToContextMessages(entry).filter((message) => message.role !== "custom" || !isContextExcludedCustomMessage(message.customType));
 }
 function getMessageFromEntryForCompaction(entry) {
-    if (entry.type === "compaction") {
-        return undefined;
-    }
-    return getMessageFromEntry(entry);
+    return contextMessagesForCompactionEntry(entry)[0];
 }
 export const DEFAULT_COMPACTION_SETTINGS = {
     enabled: true,
@@ -229,69 +211,68 @@ export function estimateTokens(message) {
         }
     }
 }
+function isCutPointMessage(message) {
+    switch (message.role) {
+        case "user":
+        case "assistant":
+        case "bashExecution":
+        case "custom":
+        case "branchSummary":
+        case "compactionSummary":
+            return true;
+        case "toolResult":
+            return false;
+    }
+    return false;
+}
+function isTurnStartMessage(message) {
+    switch (message.role) {
+        case "user":
+        case "bashExecution":
+        case "custom":
+        case "branchSummary":
+        case "compactionSummary":
+            return true;
+        case "assistant":
+        case "toolResult":
+            return false;
+    }
+    return false;
+}
+function isTurnStartEntry(entry) {
+    if (entry.type === "compaction") {
+        return false;
+    }
+    return contextMessagesForCompactionEntry(entry).some(isTurnStartMessage);
+}
 /**
- * Find valid cut points: indices of user, assistant, custom, or bashExecution messages.
+ * Find valid cut points: indices of context-visible user-like or assistant messages.
  * Never cut at tool results (they must follow their tool call).
  * When we cut at an assistant message with tool calls, its tool results follow it
  * and will be kept.
- * BashExecutionMessage is treated like a user message (user-initiated context).
  */
 function findValidCutPoints(entries, startIndex, endIndex) {
     const cutPoints = [];
     for (let i = startIndex; i < endIndex; i++) {
         const entry = entries[i];
-        switch (entry.type) {
-            case "message": {
-                const role = entry.message.role;
-                switch (role) {
-                    case "bashExecution":
-                    case "custom":
-                    case "branchSummary":
-                    case "compactionSummary":
-                    case "user":
-                    case "assistant":
-                        cutPoints.push(i);
-                        break;
-                    case "toolResult":
-                        break;
-                }
-                break;
-            }
-            case "thinking_level_change":
-            case "model_change":
-            case "compaction":
-            case "branch_summary":
-            case "custom":
-            case "custom_message":
-            case "label":
-            case "session_info":
-                break;
+        if (entry.type === "compaction") {
+            continue;
         }
-        // branch_summary and non-excluded custom_message are user-role messages, valid cut points
-        if (entry.type === "branch_summary" || isCutPointCustomMessage(entry)) {
+        if (contextMessagesForCompactionEntry(entry).some(isCutPointMessage)) {
             cutPoints.push(i);
         }
     }
     return cutPoints;
 }
 /**
- * Find the user message (or bashExecution) that starts the turn containing the given entry index.
+ * Find the context-visible user-role message that starts the turn containing the given entry index.
  * Returns -1 if no turn start found before the index.
- * BashExecutionMessage is treated like a user message for turn boundaries.
  */
 export function findTurnStartIndex(entries, entryIndex, startIndex) {
     for (let i = entryIndex; i >= startIndex; i--) {
-        const entry = entries[i];
-        // branch_summary and non-excluded custom_message are user-role messages, can start a turn
-        if (entry.type === "branch_summary" || isCutPointCustomMessage(entry)) {
+        if (isTurnStartEntry(entries[i])) {
             return i;
         }
-        if (entry.type === "message") {
-            const role = entry.message.role;
-            if (role === "user" || role === "bashExecution") {
-                return i;
-            }
-        }
     }
     return -1;
 }
@@ -321,10 +302,9 @@ export function findCutPoint(entries, startIndex, endIndex, keepRecentTokens) {
     let cutIndex = cutPoints[0]; // Default: keep from first message (not header)
     for (let i = endIndex - 1; i >= startIndex; i--) {
         const entry = entries[i];
-        if (entry.type !== "message")
+        const messageTokens = contextMessagesForCompactionEntry(entry).reduce((sum, message) => sum + estimateTokens(message), 0);
+        if (messageTokens === 0)
             continue;
-        // Estimate this message's size
-        const messageTokens = estimateTokens(entry.message);
         accumulatedTokens += messageTokens;
         // Check if we've exceeded the budget
         if (accumulatedTokens >= keepRecentTokens) {
@@ -343,31 +323,26 @@ export function findCutPoint(entries, startIndex, endIndex, keepRecentTokens) {
             break;
         }
     }
-    // Scan backwards from cutIndex to include any non-message entries (bash, settings, etc.)
+    // Scan backwards from cutIndex to include adjacent metadata entries that do not affect context.
     while (cutIndex > startIndex) {
         const prevEntry = entries[cutIndex - 1];
-        // Stop at session header or compaction boundaries
-        if (prevEntry.type === "compaction") {
+        // Stop at compaction boundaries or context-visible entries.
+        if (prevEntry.type === "compaction" || contextMessagesForCompactionEntry(prevEntry).length > 0) {
             break;
         }
         if (prevEntry.type === "custom_message" && isContextExcludedCustomMessage(prevEntry.customType)) {
             break;
         }
-        if (prevEntry.type === "message") {
-            // Stop if we hit any message
-            break;
-        }
-        // Include this non-message entry (bash, settings change, etc.)
         cutIndex--;
     }
     // Determine if this is a split turn
     const cutEntry = entries[cutIndex];
-    const isUserMessage = cutEntry.type === "message" && cutEntry.message.role === "user";
-    const turnStartIndex = isUserMessage ? -1 : findTurnStartIndex(entries, cutIndex, startIndex);
+    const startsTurn = isTurnStartEntry(cutEntry);
+    const turnStartIndex = startsTurn ? -1 : findTurnStartIndex(entries, cutIndex, startIndex);
     return {
         firstKeptEntryIndex: cutIndex,
         turnStartIndex,
-        isSplitTurn: !isUserMessage && turnStartIndex !== -1,
+        isSplitTurn: !startsTurn && turnStartIndex !== -1,
     };
 }
 // ============================================================================