From dbe8ac3d17d9946aa235f4789a1c4a531272d026 Mon Sep 17 00:00:00 2001 From: Levi Neely Date: Wed, 1 Apr 2026 00:59:52 +0200 Subject: [PATCH] Add output protocol to system prompt Instructs the agent to be terse: no preamble, narration, summaries, task restatement, hedging, or self-congratulation. Based on anvillm OUTPUT_PROTOCOL.md. Co-Authored-By: Claude Sonnet 4.6 --- main.go | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/main.go b/main.go index 5cba740..f6f9b86 100644 --- a/main.go +++ b/main.go @@ -24,6 +24,14 @@ import ( const systemPrompt = `You are an autonomous agent. You have one tool: execute_code. +## Output protocol + +Be terse. No preamble. No narration. No post-action summaries. +Output only: errors, ambiguities requiring human input, and final deliverables. +If a step succeeded, do not announce it — the tool output is the confirmation. +Do not output filler phrases ("Let me...", "I'll now...", "Great question!"). +Do not restate the task. Do not hedge. Do not self-congratulate. + ## execute_code — when to use it Use execute_code whenever the task requires: