Agent skill · data analytics · jeremylongshore
gamma-performance-tuning
Optimize Gamma API performance and reduce latency.
Why this skill is useful
Adds executable scripts for optimizing Gamma API performance, including caching and adaptive polling strategies that are not commonly found in public documentation.
What it needs
Requires node-cache, p-limit installed locally. About 4k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Gamma Performance Tuning Output Publish a tuning receipt with baseline/post-change aggregate metrics, change owner, canary result, and rollback state. Do not include source content, viewer data, or tokens. Examples Optimize a fictional staging deck, compare aggregate load/render metrics, and roll back if a synthetic sharing or accessibility check regresses. Overview Optimize Gamma API integration performance. Gamma's generate-poll-retrieve pattern means most latency is in generation time (10-60s), not API call overhead. Optimize by: reducing poll overhead, parallelizing batch operations, caching results, and choosing the right generation parameters. Prerequisites Working Gamma integration (see gamma-sdk-patterns) Understanding of async patterns Redis or in-memory cache (recommended) Performance Characteristics Operation Typical Latency Notes ----------- ---------------- ------- POST /generations 200-500ms Just starts the generation GET /generations/{id} (poll) 100-300ms Per poll request Full generation (poll to completion) 10-60s Depends on content + cards GET /themes 100-200ms Cacheable GET /folders 100-200ms Cacheable Instructions Step 1: Optimize Poll Strategy Step 2: Cache Static Data Step 3: Parallel Batch Generation Step 4: Reduce Generation Time Step 5: Preload Data at Startup Step 6: Connection Keep-Alive Performance Targets Operation Target Action if Exceeded ----------- -------- ------------------- Theme/folder lookup < 50ms (cached) Verify cache hit Generation start < 500ms Check network latency Full generation (5 cards) < 30s Use textAmount: "brief" Full generation (10+ cards) < 60s Split into smaller decks Batch of 10 presentations < 3 min Use concurrency limit of 3 Error Handling Issue Cause Solution ------- ------- ---------- High latency on first request Cold TCP connection Use keep-alive agent Cache miss storm Cache expired simultaneously Stagger TTLs Batch rate limiting Too many concurrent requests Reduce p-limit concurrency Poll timeout Complex ge …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/gamma-performance-tuning--7e5044