Remove Anthropic API call from transcribe.py - agent generates Whisper prompt itself

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

graphify/skill-aider.md

@@ -115,30 +115,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-claw.md

@@ -115,30 +115,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-codex.md

@@ -114,29 +114,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model — write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` → `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` → `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-copilot.md

@@ -117,30 +117,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-droid.md

@@ -115,30 +115,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-opencode.md

@@ -115,30 +115,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-trae.md

@@ -114,30 +114,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `GRAPHIFY_WHISPER_PROMPT` in the environment before running the transcription command.
+
+**Step 2 - Transcribe:**
 
 ```bash
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill-windows.md

@@ -107,29 +107,30 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from the detect output or analysis file. You are already a language model - write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command (PowerShell):**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` -> `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` -> `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `$env:GRAPHIFY_WHISPER_PROMPT` before running the transcription command.
+
+**Step 2 - Transcribe (PowerShell):**
 
 ```powershell
 & (Get-Content graphify-out\.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/skill.md

@@ -119,30 +119,31 @@ Skip this step entirely if `detect` returned zero `video` files.
 
 Video and audio files cannot be read directly. Transcribe them to text first, then treat the transcripts as doc files in Step 3.
 
-**Strategy:** Run non-video semantic extraction first (Step 3B) to get god nodes, use those to build a domain hint for Whisper, then transcribe. This keeps the prompt relevant without guessing the corpus topic from filenames.
+**Strategy:** Read the god nodes from `graphify-out/.graphify_detect.json` (or the analysis file if it exists from a previous run). You are already a language model — write a one-sentence domain hint yourself from those labels. Then pass it to Whisper as the initial prompt. No separate API call needed.
 
-**However**, if the corpus has *only* video files and no other docs/code, skip the god-node step and transcribe with the generic fallback prompt immediately.
+**However**, if the corpus has *only* video files and no other docs/code, use the generic fallback prompt: `"Use proper punctuation and paragraph breaks."`
 
-**Transcription command:**
+**Step 1 - Write the Whisper prompt yourself.**
+
+Read the top god node labels from detect output or analysis, then compose a short domain hint sentence, for example:
+
+- Labels: `transformer, attention, encoder, decoder` → `"Machine learning research on transformer architectures and attention mechanisms. Use proper punctuation and paragraph breaks."`
+- Labels: `kubernetes, deployment, pod, helm` → `"DevOps discussion about Kubernetes deployments and Helm charts. Use proper punctuation and paragraph breaks."`
+
+Set it as `WHISPER_PROMPT` to use in the next command.
+
+**Step 2 - Transcribe:**
 
 ```bash
+GRAPHIFY_WHISPER_MODEL=base  # or whatever --whisper-model the user passed
 $(cat graphify-out/.graphify_python) -c "
-import json
+import json, os
 from pathlib import Path
-from graphify.transcribe import build_whisper_prompt, transcribe_all
+from graphify.transcribe import transcribe_all
 
 detect = json.loads(Path('graphify-out/.graphify_detect.json').read_text())
 video_files = detect.get('files', {}).get('video', [])
-
+prompt = os.environ.get('GRAPHIFY_WHISPER_PROMPT', 'Use proper punctuation and paragraph breaks.')
-# Try to load god nodes from a previous partial run or pass [] if not yet available
-try:
-    analysis = json.loads(Path('graphify-out/.graphify_analysis.json').read_text())
-    god_nodes = analysis.get('god_nodes', [])
-except Exception:
-    god_nodes = []
-
-prompt = build_whisper_prompt(god_nodes)
-print(f'Whisper prompt: {prompt}')
 
 transcript_paths = transcribe_all(video_files, initial_prompt=prompt)
 print(json.dumps(transcript_paths))

graphify/transcribe.py

@@ -91,14 +91,14 @@ def download_audio(url: str, output_dir: Path) -> Path:
 def build_whisper_prompt(god_nodes: list[dict]) -> str:
     """Build a domain hint for Whisper from god nodes extracted from the corpus.
 
-    Takes the top god nodes (most connected concepts) already extracted from
+    Formats the top god node labels into a topic string for Whisper.
-    non-video files and asks the LLM to summarise them into a one-sentence
+    The coding agent (Claude Code, Codex, etc.) generates the actual one-sentence
-    speech-to-text hint. Falls back to a generic prompt if no nodes available.
+    domain hint from these labels and passes it via GRAPHIFY_WHISPER_PROMPT or
+    as initial_prompt — no separate API call needed here.
     """
     if not god_nodes:
         return _FALLBACK_PROMPT
 
-    # Use env override if set
     override = os.environ.get("GRAPHIFY_WHISPER_PROMPT")
     if override:
         return override
@@ -107,26 +107,6 @@ def build_whisper_prompt(god_nodes: list[dict]) -> str:
     if not labels:
         return _FALLBACK_PROMPT
 
-    try:
-        import anthropic
-        client = anthropic.Anthropic()
-        msg = client.messages.create(
-            model="claude-haiku-4-5-20251001",
-            max_tokens=60,
-            messages=[{
-                "role": "user",
-                "content": (
-                    f"These are the key concepts from a document corpus: {', '.join(labels)}. "
-                    "Write a single short sentence (under 20 words) that describes the domain "
-                    "for a speech-to-text model. Start with 'Technical' or the domain name. "
-                    "No explanation, just the sentence."
-                ),
-            }],
-        )
-        prompt = msg.content[0].text.strip().strip('"')
-        return prompt + " Use proper punctuation and paragraph breaks."
-    except Exception:
-        # If LLM call fails for any reason, fall back gracefully
     topics = ", ".join(labels[:5])
     return f"Technical discussion about {topics}. Use proper punctuation and paragraph breaks."
 

tests/test_transcribe.py

@@ -1,13 +1,10 @@
 """Tests for graphify.transcribe — video/audio transcription support."""
 from __future__ import annotations
 
-import json
 import os
 from pathlib import Path
 from unittest.mock import MagicMock, patch
 
-import sys
-
 import pytest
 
 from graphify.transcribe import (
@@ -47,38 +44,13 @@ def test_build_whisper_prompt_env_override(monkeypatch):
     assert prompt == "Custom domain hint."
 
 
-def test_build_whisper_prompt_llm_success():
+def test_build_whisper_prompt_returns_topic_string():
-    """Successful LLM call returns generated prompt with punctuation suffix."""
+    """Returns a topic-based prompt from god node labels — no LLM call."""
     god_nodes = [{"label": "neural networks"}, {"label": "transformers"}, {"label": "attention"}]
-
-    fake_response = MagicMock()
-    fake_response.content = [MagicMock(text="Machine learning and deep learning research")]
-
-    mock_anthropic = MagicMock()
-    mock_anthropic.Anthropic.return_value.messages.create.return_value = fake_response
-
     with patch.dict(os.environ, {}, clear=False):
         os.environ.pop("GRAPHIFY_WHISPER_PROMPT", None)
-        with patch.dict(sys.modules, {"anthropic": mock_anthropic}):
         prompt = build_whisper_prompt(god_nodes)
-
+    assert "neural networks" in prompt.lower() or "transformers" in prompt.lower()
-    assert "Machine learning" in prompt
-    assert "punctuation" in prompt.lower()
-
-
-def test_build_whisper_prompt_llm_failure_fallback():
-    """If LLM call raises, falls back to topic-based prompt."""
-    god_nodes = [{"label": "kubernetes"}, {"label": "docker"}, {"label": "helm"}]
-
-    mock_anthropic = MagicMock()
-    mock_anthropic.Anthropic.return_value.messages.create.side_effect = Exception("API error")
-
-    with patch.dict(os.environ, {}, clear=False):
-        os.environ.pop("GRAPHIFY_WHISPER_PROMPT", None)
-        with patch.dict(sys.modules, {"anthropic": mock_anthropic}):
-            prompt = build_whisper_prompt(god_nodes)
-
-    assert "kubernetes" in prompt.lower() or "docker" in prompt.lower()
     assert "punctuation" in prompt.lower()