add OCR script, use direct API call instead of built-in image tool

log-work-expense/SKILL.md

@@ -16,7 +16,13 @@ Log receipt images and expense details into the `wip.work_expenses` collection i
 
 ## Steps
 
-1. **Extract receipt info** — If the user sent an image, use the image tool to read the date, vendor, and amount. If the image tool fails, read the image with the `read` tool and try to extract the info visually. If you cannot confidently read the date, **ask the user** — never guess. If they only provided text, use that.
+1. **Extract receipt info** — If the user sent an image, run the OCR script to read the date, vendor, and amount:
+
+```bash
+python3 scripts/ocr_receipt.py /path/to/image.jpg
+```
+
+The script is located at `~/notes/skills/log-work-expense/scripts/ocr_receipt.py`. It returns JSON with `date`, `vendor`, and `amount` fields. If the output doesn't contain a confident date, **ask the user** — never guess. If they only provided text, use that.
 
 2. **Upload the receipt image to S2** — If the user provided an image, upload it to S2 and get the public URL. If they only provided a file or no image, skip this step.
 

log-work-expense/scripts/ocr_receipt.py

@@ -0,0 +1,60 @@
+#!/usr/bin/env python3
+"""OCR a receipt image via OpenRouter and return structured info."""
+import requests, json, sys, base64
+
+API_KEY = "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"
+MODEL = "google/gemini-2.5-flash-lite"
+IMAGE_PATH = sys.argv[1]
+
+with open(IMAGE_PATH, "rb") as f:
+    b64 = base64.b64encode(f.read()).decode()
+
+resp = requests.post(
+    "https://openrouter.ai/api/v1/chat/completions",
+    headers={
+        "Authorization": f"Bearer {API_KEY}",
+        "Content-Type": "application/json",
+    },
+    json={
+        "model": MODEL,
+        "messages": [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "text",
+                        "text": (
+                            "Extract from this receipt: date, vendor/restaurant name, and total amount. "
+                            "Return ONLY a JSON object with keys: date (YYYY-MM-DD format), vendor, amount. "
+                            "No other text."
+                        ),
+                    },
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": f"data:image/jpeg;base64,{b64}"},
+                    },
+                ],
+            }
+        ],
+        "max_tokens": 200,
+    },
+)
+
+data = resp.json()
+if "error" in data:
+    print(json.dumps({"error": data["error"]}), file=sys.stderr)
+    sys.exit(1)
+
+text = data["choices"][0]["message"]["content"].strip()
+# Strip markdown code fences if present
+if text.startswith("```"):
+    text = text.split("\n", 1)[-1]
+    if text.endswith("```"):
+        text = text[:-3]
+    text = text.strip()
+
+try:
+    result = json.loads(text)
+    print(json.dumps(result))
+except json.JSONDecodeError:
+    print(json.dumps({"raw": text}))