From 2e22f0de099e096f7e3a2c06f4c7d6b20b194f27 Mon Sep 17 00:00:00 2001
From: milkomida77 <145509032+milkomida77@users.noreply.github.com>
Date: Mon, 6 Apr 2026 16:01:54 +0200
Subject: [PATCH] fix: qualify performance claims and comparison table per
 review feedback

---
 Open Source prompts/Guardian AI/README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/Open Source prompts/Guardian AI/README.md b/Open Source prompts/Guardian AI/README.md
index ea309277..6ae460df 100644
--- a/Open Source prompts/Guardian AI/README.md	
+++ b/Open Source prompts/Guardian AI/README.md	
@@ -23,7 +23,7 @@ Guardian uses a hub-and-spoke model:
 
 ## Key Patterns
 
-1. **Identity + NOT-block**: Each agent defines what it IS and what it IS NOT (reduces task drift ~35%)
+1. **Identity + NOT-block**: Each agent defines what it IS and what it IS NOT (~35% reduction observed in production testing)
 2. **Task Registry**: SQLite-based anti-duplication with similarity matching
 3. **Setup Master**: Every task gets a blueprint before delegation (agents, tools, risks, order)
 4. **Quality Gate**: Agent output is a CLAIM; test output is EVIDENCE
@@ -31,7 +31,7 @@ Guardian uses a hub-and-spoke model:
 
 ## Differences from other agent systems
 
-| Feature | Guardian | Typical Agent Frameworks |
+| Feature | Guardian | Common Open-Source Agent Frameworks |
 |---------|----------|------------------------|
 | Prompt length | 200-800 lines | 20-50 lines |
 | Constraint ratio | 20-30% | <5% |