fix(agents): stabilize overflow compaction retries and session context accounting (openclaw#14102) thanks @vpesh

Verified:
- CI checks for commit 86a7ecb45e
- Rebase conflict resolution for compatibility with latest main

Co-authored-by: vpesh <9496634+vpesh@users.noreply.github.com>

src/auto-reply/reply/agent-runner.ts

@@ -371,6 +371,7 @@ export async function runReplyAgent(params: {
     }
 
     const usage = runResult.meta.agentMeta?.usage;
+    const promptTokens = runResult.meta.agentMeta?.promptTokens;
     const modelUsed = runResult.meta.agentMeta?.model ?? fallbackModel ?? defaultModel;
     const providerUsed =
       runResult.meta.agentMeta?.provider ?? fallbackProvider ?? followupRun.run.provider;
@@ -388,6 +389,7 @@ export async function runReplyAgent(params: {
       sessionKey,
       usage,
       lastCallUsage: runResult.meta.agentMeta?.lastCallUsage,
+      promptTokens,
       modelUsed,
       providerUsed,
       contextTokensUsed,

src/auto-reply/reply/followup-runner.ts

@@ -194,6 +194,7 @@ export function createFollowupRunner(params: {
       }
 
       const usage = runResult.meta.agentMeta?.usage;
+      const promptTokens = runResult.meta.agentMeta?.promptTokens;
       const modelUsed = runResult.meta.agentMeta?.model ?? fallbackModel ?? defaultModel;
       const contextTokensUsed =
         agentCfgContextTokens ??
@@ -207,6 +208,7 @@ export function createFollowupRunner(params: {
           sessionKey,
           usage,
           lastCallUsage: runResult.meta.agentMeta?.lastCallUsage,
+          promptTokens,
           modelUsed,
           providerUsed: fallbackProvider,
           contextTokensUsed,

src/auto-reply/reply/session-usage.ts

@@ -25,6 +25,7 @@ export async function persistSessionUsageUpdate(params: {
   modelUsed?: string;
   providerUsed?: string;
   contextTokensUsed?: number;
+  promptTokens?: number;
   systemPromptReport?: SessionSystemPromptReport;
   cliSessionId?: string;
   logLabel?: string;
@@ -56,6 +57,7 @@ export async function persistSessionUsageUpdate(params: {
               deriveSessionTotalTokens({
                 usage: usageForContext,
                 contextTokens: resolvedContextTokens,
+                promptTokens: params.promptTokens,
               }) ?? input,
             modelProvider: params.providerUsed ?? entry.modelProvider,
             model: params.modelUsed ?? entry.model,