198 skills · sorted by relevance
Summarize Claude Code token usage and estimated spending from the local ECC metrics log. Review costs by model, session, or date, compare spending with budgets, and export records to CSV while avoiding duplicate counting of session totals.
Audit Claude Code agents, skills, rules, and MCP tool configurations to estimate how much space they consume in the model's context. Identifies oversized and overlapping components, then ranks recommendations by potential token savings.
Set up vLLM to serve large language models through an OpenAI-compatible API or process prompts in batches. The skill covers deployment, performance monitoring, and quantization, which reduces model memory requirements.
Read and update Shopify store data through GraphQL, an API format for requesting specific fields, using curl. The skill covers products, orders, customers, inventory adjustments, custom fields, and bulk data exports, with checks for API errors.
Set up reinforcement learning, training guided by response rewards, for large language models. Use Megatron-LM and SGLang to prepare data, configure training and response generation, and build asynchronous or multi-turn experiments.
Break down a Claude Code session's costs by assistant message to find expensive responses. Compare input, output, cache writes, cache reads, and models, with message cost percentiles providing context for investigating unusually costly activity.
Review recent Claude Code chats to identify development strengths, recurring difficulties, and learning priorities. Receive a personalized report with practical recommendations and relevant HackerNews articles, delivered to your own Slack direct messages.
Create and audit JSON-LD, structured data that describes page content to search engines. Check JSON syntax, schema types, and required fields for articles, products, FAQs, and breadcrumbs, then guide validation for enhanced search results.
Get started with omo through a guided feature tour and workflow introduction. Optional setup migration and past-session analysis connect its features to existing work habits, while memory tools record useful preferences for future sessions.
Execute a written work plan by assigning implementation and verification to subagents. Track progress, isolate work in Git worktrees, record test evidence, and require independent verification before marking tasks complete, with optional pull request delivery.
Manage Obsidian notes through its Local REST API. Search a vault, create or update notes, append content, and execute Obsidian commands from the command line, with JSON output for agent workflows.
Retrieve U.S. federal financial data from the Treasury’s public API, including debt, revenue, spending, interest rates, and exchange rates. Filter dates and fields, retrieve multiple pages, and organize results into tables for analysis.
Build Python AI applications on Microsoft Foundry with azure-ai-projects. Provides examples for creating agents, connecting tools, managing conversations and project resources, and running evaluations through Foundry and OpenAI-compatible clients.
Inspect local AI coding session logs with agenttrace to identify cost spikes, delays, failed tools, and retry loops. Produce readable reports, compare attempts, and configure automated checks for unusual or unhealthy runs.
Turn graded execution traces and production logs into model training examples and preferred-versus-rejected response pairs. Select useful records, remove sensitive information, exclude evaluation examples, and preserve links to the original runs.
Export an approved fine-tuned model into a format suited to its hardware and runtime. Choose weight packaging and quantization, which reduces numerical precision, then compare outputs before and after export to check for regressions.
Investigate incorrect, missing, or unexpectedly changed OpenViking memories by tracing memory files back to archived change records and, when needed, source messages. Return an evidence trail, identify gaps, and suggest next steps without modifying data.
Create and manage Arize datasets for evaluation and experiments using the ax CLI. Import records from files, append examples, inspect dataset versions, and export a chosen version for offline analysis, with checks for field consistency and export completeness.
Build a shared incident timeline in Timesketch from endpoint, server, and cloud logs. Search events, add tags and investigator notes, reconstruct the attack sequence, and organize findings into a story for incident reporting.
Reconstruct activity on a compromised computer by combining file timestamps, event logs, browser history, and other evidence into one timeline. Extract and filter events with Plaso, then review and annotate them collaboratively in Timesketch.
Deploy language models with vLLM through an OpenAI-compatible API or offline batch processing. Tune response speed, throughput, and GPU memory use with quantization, which reduces numerical precision, and execution across multiple GPUs.
Prepare training datasets for large language models with GPU-assisted cleaning. Covers duplicate removal, quality filtering, and personal information redaction, with workflows for text, images, video, and audio using NeMo Curator.
Review completed AI work by scoring task difficulty and execution quality separately. A fixed scoring matrix, arguments for higher and lower ratings, and saved score history help examine weaknesses and detect repetitive scoring patterns across sessions.
Summarize WeChat group conversations into organized digests with discussion highlights, verified speaker attribution, and message statistics. Keep group history, participant profiles, and factual notes across runs, or summarize only messages received since the previous digest.