diff --git a/src/ietf_analyzer/search.py b/src/ietf_analyzer/search.py
index cbff081..bb1f84a 100644
--- a/src/ietf_analyzer/search.py
+++ b/src/ietf_analyzer/search.py
@@ -200,11 +200,55 @@ class HybridSearch:
return results
+ def _build_sources(self, search_results: list[dict]) -> tuple[list[dict], str]:
+ """Build source list and context block from search results."""
+ sources_block = ""
+ sources = []
+ for r in search_results:
+ draft = self.db.get_draft(r["name"])
+ if draft is None:
+ continue
+ text_preview = draft.full_text[:500] if draft.full_text else ""
+ sources_block += f"\n---\n**{draft.name}** — {draft.title}\n"
+ sources_block += f"Abstract: {draft.abstract[:500]}\n"
+ if text_preview:
+ sources_block += f"Content excerpt: {text_preview}\n"
+ sources.append({
+ "name": draft.name,
+ "title": draft.title,
+ "similarity": r.get("similarity", r.get("score", 0)),
+ "excerpt": r.get("excerpt", ""),
+ "match_type": r.get("match_type", ""),
+ })
+ return sources, sources_block
+
+ def search_only(self, question: str, top_k: int = 5) -> dict:
+ """Search without Claude synthesis (free — only FTS5 + Ollama).
+
+ Returns {sources: [...], has_cached_answer: bool, answer: str|None}.
+ """
+ search_results = self.search(question, top_k=top_k)
+ if not search_results:
+ return {"sources": [], "has_cached_answer": False, "answer": None}
+
+ sources, sources_block = self._build_sources(search_results)
+
+ # Check if we already have a cached answer for this question
+ prompt = ASK_PROMPT.format(question=question, sources_block=sources_block)
+ phash = _prompt_hash(prompt)
+ cached = self.db.get_cached_response("_ask_", phash)
+
+ return {
+ "sources": sources,
+ "has_cached_answer": cached is not None,
+ "answer": cached,
+ }
+
def ask(self, question: str, top_k: int = 5, cheap: bool = True) -> dict:
"""Answer a natural language question using search + Claude synthesis.
Returns {answer: str, sources: [{name, title, similarity, excerpt}]}.
- Caches Claude responses via llm_cache.
+ Caches Claude responses via llm_cache permanently.
"""
search_results = self.search(question, top_k=top_k)
@@ -214,31 +258,7 @@ class HybridSearch:
"sources": [],
}
- # Build context from top results
- sources_block = ""
- sources = []
- for r in search_results:
- draft = self.db.get_draft(r["name"])
- if draft is None:
- continue
-
- # Title + abstract + first 500 chars of full text
- text_preview = ""
- if draft.full_text:
- text_preview = draft.full_text[:500]
-
- sources_block += f"\n---\n**{draft.name}** — {draft.title}\n"
- sources_block += f"Abstract: {draft.abstract[:500]}\n"
- if text_preview:
- sources_block += f"Content excerpt: {text_preview}\n"
-
- sources.append({
- "name": draft.name,
- "title": draft.title,
- "similarity": r.get("similarity", r.get("score", 0)),
- "excerpt": r.get("excerpt", ""),
- "match_type": r.get("match_type", ""),
- })
+ sources, sources_block = self._build_sources(search_results)
prompt = ASK_PROMPT.format(
question=question,
@@ -246,7 +266,7 @@ class HybridSearch:
)
phash = _prompt_hash(prompt)
- # Check cache
+ # Check cache — cached answers are served forever (no TTL)
cached = self.db.get_cached_response("_ask_", phash)
if cached:
return {
@@ -262,7 +282,7 @@ class HybridSearch:
prompt, max_tokens=1024, cheap=cheap
)
- # Cache the response
+ # Cache permanently
self.db.cache_response(
"_ask_", phash,
self.config.claude_model_cheap if cheap else self.config.claude_model,
diff --git a/src/webui/app.py b/src/webui/app.py
index 25d77a7..9b7e21a 100644
--- a/src/webui/app.py
+++ b/src/webui/app.py
@@ -45,7 +45,8 @@ from webui.data import (
get_author_network_full,
get_citation_graph,
get_comparison_data,
- get_ask_data,
+ get_ask_search,
+ get_ask_synthesize,
global_search,
)
@@ -325,20 +326,32 @@ def ask_page():
result = None
if question:
top_k = request.args.get("top", 5, type=int)
- result = get_ask_data(db(), question, top_k=top_k)
+ # Search only (free) — returns sources + cached answer if available
+ result = get_ask_search(db(), question, top_k=top_k)
return render_template("ask.html", question=question, result=result)
-@app.route("/api/ask", methods=["POST"])
-def api_ask():
- """Answer a question via hybrid search + Claude. Returns JSON."""
+@app.route("/api/ask/synthesize", methods=["POST"])
+def api_ask_synthesize():
+ """Synthesize an answer via Claude (costs tokens, cached permanently). Returns JSON."""
data = request.get_json(force=True, silent=True)
if not data or "question" not in data:
return jsonify({"error": "Missing 'question' in request body"}), 400
question = data["question"]
top_k = data.get("top_k", 5)
- cheap = data.get("cheap", True)
- result = get_ask_data(db(), question, top_k=top_k, cheap=cheap)
+ result = get_ask_synthesize(db(), question, top_k=top_k, cheap=True)
+ return jsonify(result)
+
+
+@app.route("/api/ask", methods=["POST"])
+def api_ask():
+ """Search only (free). Returns JSON with sources + cached answer if available."""
+ data = request.get_json(force=True, silent=True)
+ if not data or "question" not in data:
+ return jsonify({"error": "Missing 'question' in request body"}), 400
+ question = data["question"]
+ top_k = data.get("top_k", 5)
+ result = get_ask_search(db(), question, top_k=top_k)
return jsonify(result)
diff --git a/src/webui/data.py b/src/webui/data.py
index d4dbd1c..915bd9a 100644
--- a/src/webui/data.py
+++ b/src/webui/data.py
@@ -1204,11 +1204,18 @@ def get_comparison_data(db: Database, names: list[str]) -> dict | None:
}
-def get_ask_data(db: Database, question: str, top_k: int = 5, cheap: bool = True) -> dict:
- """Run hybrid search + Claude synthesis for a question.
+def get_ask_search(db: Database, question: str, top_k: int = 5) -> dict:
+ """Search-only (free) — returns sources + cached answer if available."""
+ from ietf_analyzer.config import Config
+ from ietf_analyzer.search import HybridSearch
- Returns {answer: str, sources: [{name, title, similarity, excerpt}]}.
- """
+ config = Config.load()
+ searcher = HybridSearch(config, db)
+ return searcher.search_only(question, top_k=top_k)
+
+
+def get_ask_synthesize(db: Database, question: str, top_k: int = 5, cheap: bool = True) -> dict:
+ """Run Claude synthesis (costs tokens, result is cached permanently)."""
from ietf_analyzer.config import Config
from ietf_analyzer.search import HybridSearch
diff --git a/src/webui/templates/ask.html b/src/webui/templates/ask.html
index 9e3f571..a44e6c2 100644
--- a/src/webui/templates/ask.html
+++ b/src/webui/templates/ask.html
@@ -13,24 +13,17 @@
background: linear-gradient(135deg, rgba(30, 41, 59, 0.8), rgba(30, 41, 59, 0.4));
backdrop-filter: blur(10px);
}
- .source-row {
- transition: all 0.15s ease;
- }
- .source-row:hover {
- background: rgba(59, 130, 246, 0.05);
- }
+ .source-row { transition: all 0.15s ease; }
+ .source-row:hover { background: rgba(59, 130, 246, 0.05); }
.loading-spinner {
border: 3px solid rgba(59, 130, 246, 0.2);
border-top-color: #3b82f6;
border-radius: 50%;
- width: 24px;
- height: 24px;
+ width: 20px; height: 20px;
animation: spin 0.8s linear infinite;
display: inline-block;
}
- @keyframes spin {
- to { transform: rotate(360deg); }
- }
+ @keyframes spin { to { transform: rotate(360deg); } }
{% endblock %}
@@ -38,7 +31,7 @@
Ask the Draft Corpus
-
Ask natural language questions about IETF AI/agent drafts. Answers are synthesized from the most relevant documents.
+
Search across {{ "{:,}".format(stats.total if stats is defined and stats else 434) }} drafts using keyword + semantic similarity. AI synthesis is optional.