fix(compress): reject ambiguous summary matches

Author: Tarquinen

lib/tools/compress.ts

@@ -6,7 +6,12 @@ import { saveSessionState } from "../state/persistence"
 import { COMPRESS_TOOL_SPEC } from "../prompts"
 import { getCurrentParams, countAllMessageTokens, countTokens } from "../strategies/utils"
 import type { AssistantMessage } from "@opencode-ai/sdk/v2"
-import { findStringInMessages, collectToolIdsInRange, collectMessageIdsInRange } from "./utils"
+import {
+    findStringInMessages,
+    collectToolIdsInRange,
+    collectMessageIdsInRange,
+    findSummaryAnchorForBoundary,
+} from "./utils"
 import { sendCompressNotification } from "../ui/notification"
 import { buildCompressionGraphData, cacheSystemPromptTokens } from "../ui/utils"
 import { prune as applyPruneTransforms } from "../messages/prune"
@@ -183,17 +188,19 @@ export function createCompressTool(ctx: ToolContext): ReturnType<typeof tool> {
                             summaries: state.compressSummaries.length,
                         },
                     )
-                    // TODO: This takes the first summary text match and does not error on
-                    // multiple matching summaries (ambiguous fallback).
-                    const s = state.compressSummaries.find((s) => s.summary.includes(startString))
+                    const s = findSummaryAnchorForBoundary(
+                        state.compressSummaries,
+                        startString,
+                        "startString",
+                    )
                     if (s) {
                         rawStartIndex = messages.findIndex((m) => m.info.id === s.anchorMessageId)
                         clog.info(C.COMPRESS, `✓ Start resolved via summary anchor`, {
                             anchorMessageId: s.anchorMessageId,
                             rawStartIndex,
                         })
                     } else {
-                        clog.error(
+                        clog.warn(
                             C.COMPRESS,
                             `✗ Start not found in any summary either\nCannot resolve boundary`,
                         )
@@ -208,17 +215,19 @@ export function createCompressTool(ctx: ToolContext): ReturnType<typeof tool> {
                             summaries: state.compressSummaries.length,
                         },
                     )
-                    // TODO: This takes the first summary text match and does not error on
-                    // multiple matching summaries (ambiguous fallback).
-                    const s = state.compressSummaries.find((s) => s.summary.includes(endString))
+                    const s = findSummaryAnchorForBoundary(
+                        state.compressSummaries,
+                        endString,
+                        "endString",
+                    )
                     if (s) {
                         rawEndIndex = messages.findIndex((m) => m.info.id === s.anchorMessageId)
                         clog.info(C.COMPRESS, `✓ End resolved via summary anchor`, {
                             anchorMessageId: s.anchorMessageId,
                             rawEndIndex,
                         })
                     } else {
-                        clog.error(
+                        clog.warn(
                             C.COMPRESS,
                             `✗ End not found in any summary either\nCannot resolve boundary`,
                         )
@@ -398,16 +407,6 @@ export function createCompressTool(ctx: ToolContext): ReturnType<typeof tool> {
 
                 state.stats.pruneTokenCounter += estimatedCompressedTokens
 
-                const rawStartResult = {
-                    messageId: anchorMessageId,
-                    messageIndex: rawStartIndex,
-                }
-                const rawEndMessageId = messages[rawEndIndex]?.info.id || endResult.messageId
-                const rawEndResult = {
-                    messageId: rawEndMessageId,
-                    messageIndex: rawEndIndex,
-                }
-
                 const currentParams = getCurrentParams(state, messages, logger)
                 const summaryTokens = countTokens(args.content.summary)
 

lib/tools/utils.ts

@@ -1,5 +1,5 @@
 import { partial_ratio } from "fuzzball"
-import type { WithParts } from "../state"
+import type { CompressSummary, WithParts } from "../state"
 import type { Logger } from "../logger"
 import { isIgnoredUserMessage } from "../messages/utils"
 import { clog, C } from "../compress-logger"
@@ -21,6 +21,36 @@ interface MatchResult {
     matchType: "exact" | "fuzzy"
 }
 
+export function findSummaryAnchorForBoundary(
+    summaries: CompressSummary[],
+    searchString: string,
+    stringType: "startString" | "endString",
+): CompressSummary | undefined {
+    const matches = summaries.filter((s) => s.summary.includes(searchString))
+
+    if (matches.length > 1) {
+        const sample = matches.slice(0, 8).map((s) => ({
+            anchorMessageId: s.anchorMessageId,
+            preview: s.summary.substring(0, 120),
+        }))
+
+        clog.error(C.BOUNDARY, `✗ Multiple Summary Matches (ambiguous)`, {
+            type: stringType,
+            count: matches.length,
+            matches: sample,
+            omitted: Math.max(0, matches.length - sample.length),
+            searchPreview: searchString.substring(0, 150),
+        })
+
+        throw new Error(
+            `Found multiple matches for ${stringType}. ` +
+                `Provide more surrounding context to uniquely identify the intended match.`,
+        )
+    }
+
+    return matches[0]
+}
+
 function summarizeMatches(
     matches: MatchResult[],
     limit = 8,