12 skills · sorted by relevance
Guides implementation and maintenance of Claude applications using language-specific references and official SDKs, the libraries used to call the API. Covers tool integration, streaming responses, model migration, cost reviews, prompt audits, and evaluation workflows.
Build Claude applications using guidance for the official Anthropic SDK. Choose an integration approach, implement streaming and tool use, and follow workflows for model migration, prompt audits, evaluation, and cost optimization.
Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).
The English summary is being prepared. Showing the original description.
Think and work like an expert Information Retrieval Scientist. Use when a task calls for Information Retrieval Scientist judgment. Reasons from the Probability Ranking Principle, Saracevic's relevance layers, and Cranfield pooling through tuned BM25, SPLADE/ColBERT/dense and LLM reranking, trec_eval/ir_measures with paired topic-level tests, interleaving, and nugget-based RAG evaluation while treating unjudged-as-nonrelevant pools, position-biased clicks, LLM-judge circularity, single-vector embedding limits, and benchmark contamination as first-class failure modes.
The English summary is being prepared. Showing the original description.
Use when comparing two or more code implementations against a spec or requirements doc. Triggers on "which repo is better", "compare these implementations", "evaluate both solutions", "rank these codebases", or "judge which approach wins". Also covers choosing between competing PRs or vendor submissions solving the same problem. Does NOT review a single codebase for quality — use code review skills instead. Does NOT evaluate strategy docs — use strategy-review. Requires a spec file and 2+ repo paths.
The English summary is being prepared. Showing the original description.
Use when the user wants a cited, structured read of local documents and project knowledge. Triggers on: "analyze these docs", "scan my project for context", "read the docs folder", "summarize what's in .beagle/concepts/", "extract context from docs/", "what's in this folder", "go read everything in X and tell me what's there". Also invoked programmatically by other beagle skills (prfaq-beagle Ignition, brainstorm-beagle reference points, strategy-interview context grounding) via the companion contract. Does NOT trigger on codebase lookups ("find this function", "search the repo"), web research (use web-research), LLM-as-judge evaluation (use llm-judge), or document editing (use humanize-beagle). Produces a written scan plan, parallel-subagent findings, and a cited synthesis report on disk — never inline prose, never unsourced claims.
The English summary is being prepared. Showing the original description.
LLM-as-judge evaluation framework with 5-dimension rubric (accuracy, groundedness, coherence, completeness, helpfulness) for scoring AI-generated content quality with weighted composite scores and evidence citations
The English summary is being prepared. Showing the original description.
LLM-judge stage of an alpha-mining pipeline: take a miner's deterministic candidate list, keep only real signal, drop noise, and propose additions to that miner's curated note. Use after any zero-token detector run. Reads only the candidate digest and the home note, never the corpus. Triggers: "/alpha-judge <miner>", "judge the candidates".
The English summary is being prepared. Showing the original description.
Review board (Busabase App-in-Skill) that runs a fixed suite of mock test cases against a baseline vs candidate agent version and surfaces rubric-scored regressions before a release. Use when the user invokes $kelly-agent-eval or /kelly-agent-eval, wants to review agent-version regressions, compare baseline vs candidate quality, triage a release, or record a release approve/block decision. Deterministic mock rubric scores only — not a real LLM-judge call, and it never deploys anything.
The English summary is being prepared. Showing the original description.
Set up an LLM-judge evaluation that extracts canonical use cases for a PostHog feature at scale and streams the results to a Slack channel as a live feed. Use when someone wants to understand how users are actually using a specific AI/LLM-powered feature in production — what they're investigating, what questions they're trying to answer, and what patterns surface — without manually reading hundreds of traces. Assumes the feature emits `$ai_generation` and `$ai_evaluation` events with `$session_id` linkage to the trigger user's recording (the standard setup post the session-summary linkage PRs).
The English summary is being prepared. Showing the original description.
Investigate AI observability evaluations of both types — `hog` (deterministic code-based) and `llm_judge` (LLM-prompt-based). Find existing evaluations, inspect their configuration, run them against specific generations, query individual pass/fail results, and generate AI-powered summaries of patterns across many runs. Use when the user asks to debug why an evaluation is failing, surface common failure modes, compare results across filters, dry-run a Hog evaluator, prototype a new LLM-judge prompt, or manage the evaluation lifecycle (create, update, enable/disable, delete).
The English summary is being prepared. Showing the original description.
Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).
The English summary is being prepared. Showing the original description.