/slice (Python)
Use /api/topics to identify relevant verse coordinates,
then /api/slice to retrieve a canonical Scripture chunk
for LLM context — no reference parsing required.
/api/topics/api/slicePrerequisites: Python 3.9+, requests and groq libraries, a free BibleBridge API key, a free Groq API key.
pip install requests groq/slice for RAG?/api/slice retrieves canonical verse coordinates for any contiguous verse_index range — spanning chapter and book boundaries without any reference parsing. When another endpoint (such as /topics or /cross-references) returns a verse_index, you can build a precise context window around it and retrieve the full coordinate set before fetching text. This separates chunk discovery from text retrieval, which is useful for caching, deduplication, and building pre-indexed RAG stores.
Store both API keys as environment variables — never hardcode them.
export BIBLEBRIDGE_API_KEY=
export GROQ_API_KEY=your_groq_key_here/topicsQuery a topic to get ranked verse coordinates. Each result includes a verse_index — the entry point for building your RAG chunk.
curl --get https://holybible.dev/api/topics \
-H "Authorization: Bearer YOUR_API_KEY" \
--data-urlencode topic=grace \
--data limit=3/sliceTake the top result’s verse_index and expand a window around it. /slice returns canonical coordinates for every verse in the range — across any chapter or book boundary.
curl --get https://holybible.dev/api/slice \
-H "Authorization: Bearer YOUR_API_KEY" \
--data start_index=29853 \
--data end_index=29861The complete script: topic → slice → passage text → grounded LLM response.
import os
import requests
from groq import Groq
BIBLEBRIDGE_API_KEY = os.environ.get("BIBLEBRIDGE_API_KEY")
GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
if not BIBLEBRIDGE_API_KEY:
raise RuntimeError("Missing BIBLEBRIDGE_API_KEY.")
if not GROQ_API_KEY:
raise RuntimeError("Missing GROQ_API_KEY.")
BASE = "https://holybible.dev/api"
headers = {"Authorization": f"Bearer {BIBLEBRIDGE_API_KEY}"}
# Step 1: Find relevant verse coordinates for a topic
topics_resp = requests.get(
f"{BASE}/topics",
params={"topic": "grace", "limit": 5},
headers=headers,
timeout=10
)
topics_resp.raise_for_status()
top_verses = topics_resp.json()["data"]
# Step 2: Build a canonical window around the top result
top_index = top_verses[0]["verse_index"]
window = 4
slice_resp = requests.get(
f"{BASE}/slice",
params={"start_index": top_index - window, "end_index": top_index + window},
headers=headers,
timeout=10
)
slice_resp.raise_for_status()
chunk = slice_resp.json()
start_index = chunk["start_index"]
end_index = chunk["end_index"]
osis_id = chunk["osis_id"]
# Step 3: Fetch verse text for the canonical window
passage_resp = requests.get(
f"{BASE}/passage",
params={"start_index": start_index, "end_index": end_index, "version": "KJV"},
headers=headers,
timeout=10
)
passage_resp.raise_for_status()
verses = passage_resp.json()["data"]
# Step 4: Build RAG context string
context = "\n".join(
f"{v['book']['name']} {v['chapter']}:{v['verse']} — {v['text']}"
for v in verses
)
print(f"RAG chunk: {osis_id} ({chunk['verse_count']} verses)")
print()
print(context)
# Step 5: Send to LLM
client = Groq(api_key=GROQ_API_KEY)
response = client.chat.completions.create(
model="llama-3.3-70b-versatile",
messages=[
{
"role": "system",
"content": (
"You answer Bible questions using only the provided Scripture. "
"Do not cite verses not given to you."
)
},
{
"role": "user",
"content": (
f"Based only on this Scripture:\n\n{context}\n\n"
"What does the Bible teach about grace?"
)
}
]
)
print("\nLLM Response:")
print(response.choices[0].message.content)Topic: grace — top result: Ephesians 2:8 (verse_index 29857) — window ±4 verses:
RAG chunk — Eph.2.4-Eph.2.12
LLM response — grounded in the chunk above
The LLM is constrained to the provided chunk — no hallucinated verses, no unsourced citations.
/slice from text retrieval?/slice returns canonical coordinates only — no text. This lets you inspect what is in a range, deduplicate across overlapping topic results, or pre-index chunk metadata before ever making a text call. Once you have the coordinates, /passage fetches text for any translation in one call.
Because /slice operates on the global verse_index, it traverses chapter and book boundaries naturally — a window around verse 29857 spans into adjacent chapters without any client-side boundary logic.
A free API key gives you 500 calls per day — no credit card required.
Get a free API key View API docs