Skill UI

This runbook serves as a comprehensive guide for responding to incidents within the OpenEvidence platform. It outlines severity levels (P1-P3), step-by-step triage procedures, and mitigation actions, ensuring rapid and structured resolution of API downtime or degradation issues in a healthcare context.

This guide demonstrates how to implement robust rate limiting and retry logic for critical healthcare APIs using a token bucket pattern. It covers managing specific service limits (e.g., clinical queries, evidence synthesis) and implementing strategies like exponential backoff and batch processing to ensure graceful degradation and reliable data retrieval in clinical decision support systems.

This skill provides comprehensive, structured audit logging for OpenRouter API interactions. It automatically captures essential metadata like total cost, token usage, model details, latency, and status, while also performing PII redaction. The resulting audit trail is stored in an append-only SQLite database, making it ideal for building compliance-ready systems, detailed cost monitoring, and deep debugging of production environments.

This bundle provides a structured mechanism for capturing comprehensive debug information when interacting with the OpenRouter API. It captures the full request parameters, response content, detailed error logs, generation metadata (such as cost, token usage, and provider), and the execution environment. This tool is essential for diagnosing failed API calls, unexpected behaviors, or performance/latency issues.

A comprehensive guide on implementing failover and fallback mechanisms for Large Language Models (LLMs) when using OpenRouter. This skill covers native model fallbacks, provider routing strategies, and advanced client-side fallback chains. Use this pattern when building mission-critical or highly resilient systems that must maintain service availability even if a primary model or an entire provider goes offline. It ensures robust API integration and graceful degradation.

Covers HuggingFace PEFT workflows for LoRA/QLoRA-based LLM fine-tuning, enabling sub-1% parameter updates on limited GPUs plus adapter serving tips.

This comprehensive runbook provides step-by-step procedures for rapidly responding to Perplexity API outages and service degradations. It covers initial triage (checking models, auth, DNS), handling common errors (401, 429, 500/503), implementing fallbacks, and structured post-incident activities, including evidence collection and postmortem documentation.

This guide details the process of migrating search and LLM functionalities from legacy APIs (Google CSE, Bing, SerpAPI) to Perplexity Sonar. It demonstrates how to utilize the Strangler Fig pattern to build a robust adapter layer. The primary benefit is consolidating a complex, multi-step pipeline (Search -> Raw Results -> LLM Summarization) into a single, efficient API call, ensuring automatic citation handling and streamlined development.

This skill provides comprehensive patterns for building highly reliable applications that interact with external Large Language Model (LLM) APIs, such as Perplexity Sonar. It implements critical reliability techniques including Circuit Breakers, Model Fallback, Streaming Timeout handling, and Caching layers. Use this to ensure your application maintains high availability and graceful degradation even when external services experience intermittent failures or variable performance.

Fully managed, auto-scaling vector database tailored for production AI workloads such as RAG, recommendations, and semantic search, offering hybrid dense+sparse search, namespaces, metadata filters, and <100ms latency on serverless infrastructure.

Automatically validates Claude Code plugin repositories by checking required files, JSON schema, frontmatter, permissions, security, marketplace metadata, and README compliance whenever a user asks to validate or check plugins.

A comprehensive runbook for rapidly responding to PostHog integration failures. It provides detailed triage steps and troubleshooting guides for common API errors (401, 429, etc.), determining the root cause, and implementing graceful degradation patterns to ensure core application functionality remains stable during an outage.