From 9d9472c63028d216071c32895a76b001f30084e4 Mon Sep 17 00:00:00 2001
From: rustfuture <121799572+rustfuture@users.noreply.github.com>
Date: Mon, 28 Sep 2026 22:29:01 +0300
Subject: [PATCH] State that quoted lines are verbatim but not guaranteed
 relevant

Say plainly what the answer contract guarantees.

- Quotes are verbatim source lines; the README's own measured results (12 false accepts in 42 questions) show relevance is not guaranteed, so the opening and the second bullet now say so.
- Mark `evaluation/evaluate_modes.py` as legacy: it overwrites `evaluation/results_modes.json` in place, unlike the documented `evaluation/v3/evaluate.py`. Docstring only, no behaviour change.
---
 README.md                    | 4 ++--
 evaluation/evaluate_modes.py | 6 +++++-
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 80ff7a7..f584f4d 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # Repository Intelligence
 
-Repository Intelligence searches local code and returns exact source lines to answer questions instead of generating prose.
+Repository Intelligence searches local code and returns verbatim source lines for a question instead of generating prose; relevance is not guaranteed.
 
 [![CI](https://github.com/rustfuture/repository-intelligence/actions/workflows/ci.yml/badge.svg)](https://github.com/rustfuture/repository-intelligence/actions/workflows/ci.yml)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
@@ -8,7 +8,7 @@ Repository Intelligence searches local code and returns exact source lines to an
 **Status:** Experimental research prototype (v0.1.0) for local code inspection.
 
 - Searches code using text matching, offline hash embeddings, or local neural models.
-- Answers questions by extracting and quoting exact source lines instead of generating text.
+- Responds to questions by quoting exact source lines instead of generating text; quotes match the source, but relevance is not guaranteed.
 - Stores portable repository index snapshots without external database dependencies.
 - Synchronizes file additions, changes, and deletions incrementally through Git diffs.
 - Exposes local search and index reload endpoints over a lightweight HTTP server.
diff --git a/evaluation/evaluate_modes.py b/evaluation/evaluate_modes.py
index 60a9f71..3001137 100644
--- a/evaluation/evaluate_modes.py
+++ b/evaluation/evaluate_modes.py
@@ -1,5 +1,9 @@
 #!/usr/bin/env python3
-"""Compare lexical, semantic, and RRF hybrid retrieval modes across splits."""
+"""Compare lexical, semantic, and RRF hybrid retrieval modes across splits.
+
+Legacy script: it overwrites evaluation/results_modes.json in place. The documented
+reproduction path is evaluation/v3/evaluate.py, which writes to a new directory.
+"""
 import json
 import os
 import re