let5see/clawdbot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(compaction): apply staged pruning

Browse Source
This commit is contained in:
Peter Steinberger
2026-01-23 22:23:05 +00:00
parent ae0741a346
commit 022aa10063
1 changed files with 51 additions and 186 deletions
Download Patch File Download Diff File Expand all files Collapse all files

233
src/agents/pi-extensions/compaction-safeguard.ts
View File

@@ -1,12 +1,16 @@
import type { AgentMessage } from "@mariozechner/pi-agent-core"; import type { AgentMessage } from "@mariozechner/pi-agent-core";
import type { ExtensionAPI, ExtensionContext, FileOperations } from "@mariozechner/pi-coding-agent"; import type { ExtensionAPI, ExtensionContext, FileOperations } from "@mariozechner/pi-coding-agent";
import { estimateTokens, generateSummary } from "@mariozechner/pi-coding-agent"; import {
BASE_CHUNK_RATIO,
import { DEFAULT_CONTEXT_TOKENS } from "../defaults.js"; MIN_CHUNK_RATIO,
SAFETY_MARGIN,
const BASE_CHUNK_RATIO = 0.4; computeAdaptiveChunkRatio,
const MIN_CHUNK_RATIO = 0.15; estimateMessagesTokens,
const SAFETY_MARGIN = 1.2; // 20% buffer for estimateTokens() inaccuracy isOversizedForSummary,
pruneHistoryForContextShare,
resolveContextWindowTokens,
summarizeInStages,
} from "../compaction.js";
const FALLBACK_SUMMARY = const FALLBACK_SUMMARY =
"Summary unavailable due to context limits. Older messages were truncated."; "Summary unavailable due to context limits. Older messages were truncated.";
const TURN_PREFIX_INSTRUCTIONS = const TURN_PREFIX_INSTRUCTIONS =
@@ -129,175 +133,6 @@ function formatFileOperations(readFiles: string[], modifiedFiles: string[]): str
return `\n\n${sections.join("\n\n")}`; return `\n\n${sections.join("\n\n")}`;
} }
function chunkMessages(messages: AgentMessage[], maxTokens: number): AgentMessage[][] {
if (messages.length === 0) return [];
const chunks: AgentMessage[][] = [];
let currentChunk: AgentMessage[] = [];
let currentTokens = 0;
for (const message of messages) {
const messageTokens = estimateTokens(message);
if (currentChunk.length > 0 && currentTokens + messageTokens > maxTokens) {
chunks.push(currentChunk);
currentChunk = [];
currentTokens = 0;
}
currentChunk.push(message);
currentTokens += messageTokens;
if (messageTokens > maxTokens) {
// Split oversized messages to avoid unbounded chunk growth.
chunks.push(currentChunk);
currentChunk = [];
currentTokens = 0;
}
}
if (currentChunk.length > 0) {
chunks.push(currentChunk);
}
return chunks;
}
/**
* Compute adaptive chunk ratio based on average message size.
* When messages are large, we use smaller chunks to avoid exceeding model limits.
*/
function computeAdaptiveChunkRatio(messages: AgentMessage[], contextWindow: number): number {
if (messages.length === 0) return BASE_CHUNK_RATIO;
const totalTokens = messages.reduce((sum, m) => sum + estimateTokens(m), 0);
const avgTokens = totalTokens / messages.length;
// Apply safety margin to account for estimation inaccuracy
const safeAvgTokens = avgTokens * SAFETY_MARGIN;
const avgRatio = safeAvgTokens / contextWindow;
// If average message is > 10% of context, reduce chunk ratio
if (avgRatio > 0.1) {
const reduction = Math.min(avgRatio * 2, BASE_CHUNK_RATIO - MIN_CHUNK_RATIO);
return Math.max(MIN_CHUNK_RATIO, BASE_CHUNK_RATIO - reduction);
}
return BASE_CHUNK_RATIO;
}
/**
* Check if a single message is too large to summarize.
* If single message > 50% of context, it can't be summarized safely.
*/
function isOversizedForSummary(msg: AgentMessage, contextWindow: number): boolean {
const tokens = estimateTokens(msg) * SAFETY_MARGIN;
return tokens > contextWindow * 0.5;
}
async function summarizeChunks(params: {
messages: AgentMessage[];
model: NonNullable<ExtensionContext["model"]>;
apiKey: string;
signal: AbortSignal;
reserveTokens: number;
maxChunkTokens: number;
customInstructions?: string;
previousSummary?: string;
}): Promise<string> {
if (params.messages.length === 0) {
return params.previousSummary ?? "No prior history.";
}
const chunks = chunkMessages(params.messages, params.maxChunkTokens);
let summary = params.previousSummary;
for (const chunk of chunks) {
summary = await generateSummary(
chunk,
params.model,
params.reserveTokens,
params.apiKey,
params.signal,
params.customInstructions,
summary,
);
}
return summary ?? "No prior history.";
}
/**
* Summarize with progressive fallback for handling oversized messages.
* If full summarization fails, tries partial summarization excluding oversized messages.
*/
async function summarizeWithFallback(params: {
messages: AgentMessage[];
model: NonNullable<ExtensionContext["model"]>;
apiKey: string;
signal: AbortSignal;
reserveTokens: number;
maxChunkTokens: number;
contextWindow: number;
customInstructions?: string;
previousSummary?: string;
}): Promise<string> {
const { messages, contextWindow } = params;
if (messages.length === 0) {
return params.previousSummary ?? "No prior history.";
}
// Try full summarization first
try {
return await summarizeChunks(params);
} catch (fullError) {
console.warn(
`Full summarization failed, trying partial: ${
fullError instanceof Error ? fullError.message : String(fullError)
}`,
);
}
// Fallback 1: Summarize only small messages, note oversized ones
const smallMessages: AgentMessage[] = [];
const oversizedNotes: string[] = [];
for (const msg of messages) {
if (isOversizedForSummary(msg, contextWindow)) {
const role = (msg as { role?: string }).role ?? "message";
const tokens = estimateTokens(msg);
oversizedNotes.push(
`[Large ${role} (~${Math.round(tokens / 1000)}K tokens) omitted from summary]`,
);
} else {
smallMessages.push(msg);
}
}
if (smallMessages.length > 0) {
try {
const partialSummary = await summarizeChunks({
...params,
messages: smallMessages,
});
const notes = oversizedNotes.length > 0 ? `\n\n${oversizedNotes.join("\n")}` : "";
return partialSummary + notes;
} catch (partialError) {
console.warn(
`Partial summarization also failed: ${
partialError instanceof Error ? partialError.message : String(partialError)
}`,
);
}
}
// Final fallback: Just note what was there
return (
`Context contained ${messages.length} messages (${oversizedNotes.length} oversized). ` +
`Summary unavailable due to size limits.`
);
}
export default function compactionSafeguardExtension(api: ExtensionAPI): void { export default function compactionSafeguardExtension(api: ExtensionAPI): void {
api.on("session_before_compact", async (event, ctx) => { api.on("session_before_compact", async (event, ctx) => {
const { preparation, customInstructions, signal } = event; const { preparation, customInstructions, signal } = event;
@@ -335,19 +170,48 @@ export default function compactionSafeguardExtension(api: ExtensionAPI): void {
} }
try { try {
const contextWindowTokens = Math.max( const contextWindowTokens = resolveContextWindowTokens(model);
const turnPrefixMessages = preparation.turnPrefixMessages ?? [];
let messagesToSummarize = preparation.messagesToSummarize;
const tokensBefore =
typeof preparation.tokensBefore === "number" && Number.isFinite(preparation.tokensBefore)
? preparation.tokensBefore
: undefined;
if (tokensBefore !== undefined) {
const summarizableTokens =
estimateMessagesTokens(messagesToSummarize) + estimateMessagesTokens(turnPrefixMessages);
const newContentTokens = Math.max(0, Math.floor(tokensBefore - summarizableTokens));
const maxHistoryTokens = Math.floor(contextWindowTokens * 0.5);
if (newContentTokens > maxHistoryTokens) {
const pruned = pruneHistoryForContextShare({
messages: messagesToSummarize,
maxContextTokens: contextWindowTokens,
maxHistoryShare: 0.5,
parts: 2,
});
if (pruned.droppedChunks > 0) {
const newContentRatio = (newContentTokens / contextWindowTokens) * 100;
console.warn(
`Compaction safeguard: new content uses ${newContentRatio.toFixed(
1, 1,
Math.floor(model.contextWindow ?? DEFAULT_CONTEXT_TOKENS), )}% of context; dropped ${pruned.droppedChunks} older chunk(s) ` +
`(${pruned.droppedMessages} messages) to fit history budget.`,
); );
messagesToSummarize = pruned.messages;
}
}
}
// Use adaptive chunk ratio based on message sizes // Use adaptive chunk ratio based on message sizes
const allMessages = [...preparation.messagesToSummarize, ...preparation.turnPrefixMessages]; const allMessages = [...messagesToSummarize, ...turnPrefixMessages];
const adaptiveRatio = computeAdaptiveChunkRatio(allMessages, contextWindowTokens); const adaptiveRatio = computeAdaptiveChunkRatio(allMessages, contextWindowTokens);
const maxChunkTokens = Math.max(1, Math.floor(contextWindowTokens * adaptiveRatio)); const maxChunkTokens = Math.max(1, Math.floor(contextWindowTokens * adaptiveRatio));
const reserveTokens = Math.max(1, Math.floor(preparation.settings.reserveTokens)); const reserveTokens = Math.max(1, Math.floor(preparation.settings.reserveTokens));
const historySummary = await summarizeWithFallback({ const historySummary = await summarizeInStages({
messages: preparation.messagesToSummarize, messages: messagesToSummarize,
model, model,
apiKey, apiKey,
signal, signal,
@@ -359,9 +223,9 @@ export default function compactionSafeguardExtension(api: ExtensionAPI): void {
}); });
let summary = historySummary; let summary = historySummary;
if (preparation.isSplitTurn && preparation.turnPrefixMessages.length > 0) { if (preparation.isSplitTurn && turnPrefixMessages.length > 0) {
const prefixSummary = await summarizeWithFallback({ const prefixSummary = await summarizeInStages({
messages: preparation.turnPrefixMessages, messages: turnPrefixMessages,
model, model,
apiKey, apiKey,
signal, signal,
@@ -369,6 +233,7 @@ export default function compactionSafeguardExtension(api: ExtensionAPI): void {
maxChunkTokens, maxChunkTokens,
contextWindow: contextWindowTokens, contextWindow: contextWindowTokens,
customInstructions: TURN_PREFIX_INSTRUCTIONS, customInstructions: TURN_PREFIX_INSTRUCTIONS,
previousSummary: undefined,
}); });
summary = `${historySummary}\n\n---\n\n**Turn Context (split turn):**\n\n${prefixSummary}`; summary = `${historySummary}\n\n---\n\n**Turn Context (split turn):**\n\n${prefixSummary}`;
} }