diff --git a/docs/sprint-runbook/sprint-3/sprint-3-day-2.sh b/docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
new file mode 100755
index 0000000..abf731f
--- /dev/null
+++ b/docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
@@ -0,0 +1,307 @@
+#!/usr/bin/env bash
+#
+# Sprint 3 Day 2: Keyword extraction and text normalisation verification.
+#
+# Purpose:
+#   Console-only verification for deterministic NLP text processing and keyword
+#   extraction. This runbook uses the Docker/PostgreSQL stack and the current
+#   consolidated NLP test files.
+#
+# Execution:
+#   chmod +x docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
+#   ./docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
+#
+# Optional overrides:
+#   PROJECT_ROOT=/path/to/StudyBuddy-Study-Planner-Project ./docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
+#   TEST_SETTINGS_MODULE=config.settings.test ./docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
+#   LOCAL_SETTINGS_MODULE=config.settings.local ./docs/sprint-runbook/sprint-3/sprint-3-day-2.sh
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+DEFAULT_PROJECT_ROOT="$(cd "$SCRIPT_DIR/../../.." && pwd)"
+
+PROJECT_ROOT="${PROJECT_ROOT:-$DEFAULT_PROJECT_ROOT}"
+TEST_SETTINGS_MODULE="${TEST_SETTINGS_MODULE:-config.settings.test}"
+LOCAL_SETTINGS_MODULE="${LOCAL_SETTINGS_MODULE:-config.settings.local}"
+
+print_step() {
+  printf "\n==> %s\n\n" "$1"
+}
+
+run() {
+  printf '$ %s\n' "$*"
+  "$@"
+}
+
+print_step "Verify repository root"
+run cd "$PROJECT_ROOT"
+printf "Repository root: %s\n" "$(pwd)"
+
+print_step "Confirm Sprint 3 Day 2 files exist"
+
+required_files=(
+  "apps/insights/nlp/__init__.py"
+  "apps/insights/nlp/text_processing.py"
+  "apps/insights/nlp/keyword_extraction.py"
+  "apps/insights/tests/test_nlp_keyword_extraction.py"
+  "apps/insights/tests/test_nlp_text_processing.py"
+  "docs/ai-nlp-contract.md"
+)
+
+for file_path in "${required_files[@]}"; do
+  if [[ ! -f "$file_path" ]]; then
+    printf "MISSING: %s\n" "$file_path"
+    exit 1
+  fi
+
+  printf "FOUND: %s\n" "$file_path"
+done
+
+print_step "Confirm obsolete mixed NLP test file is absent"
+
+obsolete_files=(
+  "apps/insights/tests/test_keyword_extraction.py"
+)
+
+for file_path in "${obsolete_files[@]}"; do
+  if [[ -e "$file_path" ]]; then
+    printf "OBSOLETE FILE PRESENT: %s\n" "$file_path"
+    exit 1
+  fi
+
+  printf "ABSENT: %s\n" "$file_path"
+done
+
+print_step "Build and start Docker/PostgreSQL stack"
+run docker compose up -d --build
+run docker compose ps
+
+print_step "Run Django system check"
+run docker compose exec -T web python manage.py check --settings="$LOCAL_SETTINGS_MODULE"
+
+print_step "Confirm NLP modules import correctly"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+from apps.insights.nlp.text_processing import (
+    meaningful_tokens,
+    normalise_text,
+    source_text_hash,
+    tokenize,
+)
+
+print("text_processing imports verified")
+print("keyword_extraction imports verified")
+print("NLP module import verification complete")
+PY
+
+print_step "Verify text normalisation behaviour"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.text_processing import normalise_text
+
+result = normalise_text("  Django    Testing\nWorkflow  ")
+
+assert result == "django testing workflow", result
+
+print("Normalised text:", result)
+print("Text normalisation verified")
+PY
+
+print_step "Verify tokenisation removes punctuation"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.text_processing import tokenize
+
+result = tokenize("Django, pytest, and PostgreSQL!")
+
+assert result == ["django", "pytest", "and", "postgresql"], result
+
+print("Tokens:", result)
+print("Tokenisation verified")
+PY
+
+print_step "Verify stop-word and short-token filtering"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.text_processing import meaningful_tokens
+
+result = meaningful_tokens("The API is on and the UI is in sync.")
+
+assert result == ["api", "sync"], result
+
+print("Meaningful tokens:", result)
+print("Stop-word filtering verified")
+PY
+
+print_step "Verify source hash stability"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.text_processing import source_text_hash
+
+first_hash = source_text_hash("Django    testing workflow")
+second_hash = source_text_hash(" django testing   workflow ")
+
+assert first_hash == second_hash, (first_hash, second_hash)
+assert len(first_hash) == 64, first_hash
+
+print("Source hash:", first_hash)
+print("Source hash stability verified")
+PY
+
+print_step "Verify keyword extraction ranks repeated terms first"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+
+result = extract_keywords(
+    "Django testing testing pytest Django testing database.",
+    limit=3,
+)
+
+assert result == ["testing", "django", "database"], result
+
+print("Ranked keywords:", result)
+print("Repeated-term ranking verified")
+PY
+
+print_step "Verify deterministic keyword tie-breaking"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+
+result = extract_keywords("zebra alpha beta zebra alpha beta", limit=3)
+
+assert result == ["alpha", "beta", "zebra"], result
+
+print("Tie-break keywords:", result)
+print("Alphabetical tie-breaking verified")
+PY
+
+print_step "Verify keyword limit handling"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+
+result = extract_keywords(
+    "django pytest postgres docker templates bootstrap views models",
+    limit=4,
+)
+
+assert len(result) == 4, result
+assert result == ["bootstrap", "django", "docker", "models"], result
+
+print("Limited keywords:", result)
+print("Keyword limit verified")
+PY
+
+print_step "Verify empty and low-information input handling"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+
+none_result = extract_keywords(None)
+empty_result = extract_keywords("")
+stop_word_result = extract_keywords("and the of to")
+short_token_result = extract_keywords("AI ML UX")
+zero_limit_result = extract_keywords("biology biology chemistry", limit=0)
+negative_limit_result = extract_keywords("biology biology chemistry", limit=-1)
+
+assert none_result == [], none_result
+assert empty_result == [], empty_result
+assert stop_word_result == [], stop_word_result
+assert short_token_result == [], short_token_result
+assert zero_limit_result == [], zero_limit_result
+assert negative_limit_result == [], negative_limit_result
+
+print("None input keywords:", none_result)
+print("Empty input keywords:", empty_result)
+print("Stop-word-only keywords:", stop_word_result)
+print("Short-token-only keywords:", short_token_result)
+print("Zero-limit keywords:", zero_limit_result)
+print("Negative-limit keywords:", negative_limit_result)
+print("Low-information keyword handling verified")
+PY
+
+print_step "Verify keyword extraction is stable across repeated runs"
+
+docker compose exec -T web python manage.py shell --settings="$LOCAL_SETTINGS_MODULE" <<'PY'
+from apps.insights.nlp.keyword_extraction import extract_keywords
+
+text = (
+    "Django testing protects the workflow. "
+    "Django testing confirms database behaviour. "
+    "Pytest testing keeps changes safe."
+)
+
+first = extract_keywords(text, limit=5)
+second = extract_keywords(text, limit=5)
+third = extract_keywords(text, limit=5)
+
+assert first == second == third, (first, second, third)
+
+print("Repeated run one:", first)
+print("Repeated run two:", second)
+print("Repeated run three:", third)
+print("Deterministic keyword extraction verified")
+PY
+
+print_step "Confirm AI/NLP contract document sections"
+
+required_sections=(
+  "## Current Scope"
+  "## Deterministic Contract"
+  "## Text Normalisation"
+  "## Keyword Extraction"
+  "## Testing Contract"
+)
+
+for section in "${required_sections[@]}"; do
+  if ! grep -Fq "$section" docs/ai-nlp-contract.md; then
+    printf "MISSING SECTION: %s\n" "$section"
+    exit 1
+  fi
+
+  printf "FOUND: %s\n" "$section"
+done
+
+printf "\nAI/NLP contract section check is a smoke check for expected headings.\n"
+
+print_step "Run Sprint 3 Day 2 NLP keyword and text-processing tests"
+run docker compose exec -T web env DJANGO_SETTINGS_MODULE="$TEST_SETTINGS_MODULE" \
+  pytest \
+  apps/insights/tests/test_nlp_keyword_extraction.py \
+  apps/insights/tests/test_nlp_text_processing.py \
+  -q
+
+print_step "Run all current insights tests"
+run docker compose exec -T web env DJANGO_SETTINGS_MODULE="$TEST_SETTINGS_MODULE" pytest apps/insights -q
+
+print_step "Run full project test suite"
+run docker compose exec -T web env DJANGO_SETTINGS_MODULE="$TEST_SETTINGS_MODULE" pytest -q
+
+print_step "Final receipt"
+
+cat <<'RECEIPT'
+Repository root verified.
+Sprint 3 Day 2 files verified.
+Obsolete mixed NLP test file is absent.
+Docker/PostgreSQL stack is running.
+Django system check passed.
+NLP modules import correctly.
+Text normalisation verified.
+Tokenisation verified.
+Stop-word filtering verified.
+Source hash stability verified.
+Repeated-term keyword ranking verified.
+Alphabetical tie-breaking verified.
+Keyword limit handling verified.
+Empty and low-information input handling verified.
+Deterministic repeated-run behaviour verified.
+AI/NLP contract document headings verified.
+Sprint 3 Day 2 NLP keyword and text-processing tests pass.
+Current insights tests pass.
+Full project regression suite passes.
+Sprint 3 Day 2 verification complete.
+RECEIPT