Function startAgentSummarization

src/services/AgentSummary/agentSummary.ts:46–179

(
  taskId: string,
  agentId: AgentId,
  cacheSafeParams: CacheSafeParams,
  setAppState: TaskContext['setAppState'],
)

}

export function startAgentSummarization(
  taskId: string,
  agentId: AgentId,
  cacheSafeParams: CacheSafeParams,
  setAppState: TaskContext['setAppState'],
): { stop: () => void } {
  // Drop forkContextMessages from the closure — runSummary rebuilds it each
  // tick from getAgentTranscript(). Without this, the original fork messages
  // (passed from AgentTool.tsx) are pinned for the lifetime of the timer.
  const { forkContextMessages: _drop, ...baseParams } = cacheSafeParams
  let summaryAbortController: AbortController | null = null
  let timeoutId: ReturnType<typeof setTimeout> | null = null
  let stopped = false
  let previousSummary: string | null = null

  async function runSummary(): Promise<void> {
    if (stopped) return

    logForDebugging(`[AgentSummary] Timer fired for agent ${agentId}`)

    try {
      // Read current messages from transcript
      const transcript = await getAgentTranscript(agentId)
      if (!transcript || transcript.messages.length < 3) {
        // Not enough context yet — finally block will schedule next attempt
        logForDebugging(
          `[AgentSummary] Skipping summary for ${taskId}: not enough messages (${transcript?.messages.length ?? 0})`,
        )
        return
      }

      // Filter to clean message state
      const cleanMessages = filterIncompleteToolCalls(transcript.messages)

      // Build fork params with current messages
      const forkParams: CacheSafeParams = {
        ...baseParams,
        forkContextMessages: cleanMessages,
      }

      logForDebugging(
        `[AgentSummary] Forking for summary, ${cleanMessages.length} messages in context`,
      )

      // Create abort controller for this summary
      summaryAbortController = new AbortController()

      // Deny tools via callback, NOT by passing tools:[] - that busts cache
      const canUseTool = async () => ({
        behavior: 'deny' as const,
        message: 'No tools needed for summary',
        decisionReason: { type: 'other' as const, reason: 'summary only' },
      })

      // DO NOT set maxOutputTokens here. The fork piggybacks on the main
      // thread's prompt cache by sending identical cache-key params (system,
      // tools, model, messages prefix, thinking config). Setting maxOutputTokens
      // would clamp budget_tokens, creating a thinking config mismatch that