Performance & cost: gpt-4o-mini, gzip + caching, parallel GeoJSON#4
Open
joaquinOEF wants to merge 2 commits into
Open
Performance & cost: gpt-4o-mini, gzip + caching, parallel GeoJSON#4joaquinOEF wants to merge 2 commits into
joaquinOEF wants to merge 2 commits into
Conversation
… OpenAI timeout The /api/chat/* and /api/osm/* endpoints called OpenAI / Overpass with no auth, no rate limit, no input validation, and no body-size cap — anyone could run up the OpenAI bill, use us as a free proxy, or OOM the server with a huge payload. - express-rate-limit (40/min/IP) on /api/chat and /api/osm - zod-validate /api/chat/query (string ≤500 chars + optional bbox tuple); cap index-assets (≤5000), index-geojson features (≤20000), overpass query length (≤10k) - express.json/urlencoded limit 1mb - OpenAI client timeout 15s + maxRetries 2 (a hung call can't block the server) - health check no longer leaks OPENAI_API_KEY presence - .gitignore .env* Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…, bigger embed batch - generateSummary: gpt-4o → gpt-4o-mini (~50× cheaper, fast; summary is short prose) - compression() middleware → the ~1.9MB GeoJSON ships gzipped (~200-300KB) - express.static maxAge 1d → repeat visits don't re-download the static assets/GeoJSON - osmService: load the 6 risk-zone files in parallel (Promise.all) instead of sequentially - embedding batch 50 → 500 (far fewer OpenAI round-trips → faster startup indexing) Deferred (needs a persistent vector store to be reliable on Replit cold starts, so not a low-risk one-liner): persisting embeddings so the server doesn't re-embed on every boot. Mitigated here by the larger batch size. Tracked as a follow-up. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This was referenced Jun 8, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Low-risk performance/cost wins.
compression) — the ~1.9 MB of GeoJSON ships at ~200–300 KB.express.staticmaxAge: 1dso repeat visits don't re-download assets/GeoJSON (index.html stays uncached via the SPA fallback).osmServiceloads the 6 risk-zone files withPromise.allinstead of sequentially.Deferred (intentionally)
Persisting embeddings so the server doesn't re-embed ~6k features on every boot. On Replit, a disk cache doesn't reliably survive cold starts, so doing it right needs a persistent vector store (Qdrant/pgvector) — that's an M-effort follow-up, not a low-risk one-liner. The larger batch size already cuts the re-embed time.
🤖 Generated with Claude Code