Agent skill · software engineering · jeremylongshore
miro-performance-tuning
Diagnose supplied Miro latency and credit evidence, then implement repository-side read/write controls for cursor behavior, bulk semantics, throughput, and freshness. Use when improving Miro throughput or freshness. Trigger with \"miro integration performance tuning\".
Why this skill is useful
Adds executable scripts for optimizing Miro API performance with caching and pagination that aren't in public docs.
What it needs
Requires lru-cache, redis installed locally. Requires miro account access. About 2k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Miro Integration Performance Tuning Overview Optimize credits, latency, payload, or reconciliation time without sacrificing correctness or tenant fairness; use the evidence produced here to make the next decision explicit and reviewable. Prerequisites Approved Miro application, tenant, and board scope Current repository and deployment evidence Named owner, success criteria, and rollback or recovery boundary Tool Discipline Use Read, Glob, and Grep to inspect the repository, configuration names, adapters, tests, and evidence. Use WebFetch only for current official Miro documentation. Use Write or Edit after confirming the requested mode, target environment, tenant, board, and approval boundary. These declared tools do not call authenticated Miro APIs or deployment CLIs; implement client, configuration, and test changes, then return exact operator commands or an approval-gated handoff for live execution. Current Contract Calls consume endpoint-specific credits under a shared user/application budget. Cursor presence, not guessed page fullness, controls collection continuation. Bulk create supports at most twenty items and charges Level 2 credits per item. Board searches without team/project filters may have indexing delay. Authentication For REST work, use OAuth 2.0 Authorization Code with the narrowest Miro scopes. Bind each encrypted token record to its user, application, authorized team, and granted scopes. Never print access tokens, refresh tokens, client secrets, authorization codes, or board content. Instructions 1. Measure latency, pages, items, bytes, credits, errors, and queue age by operation. 2. Find redundant reads, unbounded traversal, duplicate callers, and reconciliation hot spots. 3. Apply narrow filters, checkpoints, deduplication, bounded concurrency, and explicit cache freshness. 4. Use bulk create only when transactional and per-item credit semantics fit. 5. Load-test synthetic data under a fixed credit budget and verify tenant fairness. 6. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/miro-performance-tuning