Explore AI Agent Skills & Claude Prompts

Ground-truth 2026 playbook for writing posts, replies, and threads on Bluesky. Covers the three feeds (Discover, Following, Custom), why custom feeds and starter packs are the dominant 2026 growth levers, the 300-grapheme limit, quote-post culture, self-reply threads, labelers and stackable moderation, and why X-ported posts flop. Activate when drafting for Bluesky — not for X/Twitter threads.

ai-cost-finops

Operational playbook for cutting LLM application cost in 2026. Covers provider pricing (Anthropic Claude Opus/Sonnet/Haiku, OpenAI GPT-4o/o-series, Google Gemini Pro/Flash), Anthropic prompt caching (90% discount on cached tokens, up to 85% latency cut), OpenAI automatic prompt caching (50% discount), Anthropic/OpenAI/Gemini Batch APIs (50% off), model routing patterns (Martian, NotDiamond, OpenRouter Auto vs manual rules), token economics (output 3–8× input), structured output cost wins, RAG cost stack (embeddings, rerank, vector DBs), and FinOps observability (Helicone, Langfuse, Phoenix, LangSmith, Vantage). Use when projecting LLM cost, hunting waste in an existing pipeline, picking a model, or setting up per-feature attribution. Output: cost projections with cited prices and quantified optimization levers.

ai-rag-patterns

Operational playbook for designing and tuning production RAG (Retrieval-Augmented Generation) systems in 2026. Covers chunking strategies (fixed/recursive, semantic, late chunking, Anthropic Contextual Retrieval), retrieval (BM25, dense, hybrid with reciprocal rank fusion, ColBERT late-interaction), query rewriting (HyDE, multi-query, decomposition), reranking (Cohere Rerank, cross-encoders, BGE-Reranker), evaluation (RAGAS metrics, TruLens RAG Triad, DeepEval), the seven failure modes (Barnett et al.), agentic RAG patterns, multi-modal RAG (ColPali), and the long-context-vs-RAG trade-off. Use when designing a new RAG pipeline, diagnosing why an existing one underperforms, or evaluating retrieval quality. Output: architecture decisions with cited numbers, not vibes.

ad-frameworks

Compact playbook of ad copy frameworks (AIDA, PAS, problem-agitation-solution, before-after-bridge, 4Us, FAB) and when to apply each. Use when structuring the body of an ad — script, body copy, landing-page sections, or any persuasive piece. Includes a decision guide that matches framework to audience awareness level (Schwartz).

content-translate

Translate and localize documents, code comments, UI strings, and structured content while preserving formatting, adapting to domain-specific terminology, and maintaining consistency across large files via batching.

ai-evals-design

Operational playbook for designing production LLM evaluation in 2026. Covers eval framework selection (Promptfoo, DeepEval, Maxim, LangSmith, Braintrust, Phoenix, Galileo, Patronus), golden dataset construction (size, stratification, versioning), synthetic data generation, LLM-as-judge design with G-Eval methodology (Liu et al., EMNLP 2023), preference-leakage mitigation (Wong et al. 2025), statistical significance testing (McNemar's, paired t, Bayesian pairwise, bootstrap), RAG metrics (RAGAS, TruLens RAG Triad), agent trajectory evaluation, drift detection, eval-driven development (Red Hat, March 2026), CI/CD prompt-regression gates, and benchmark literacy (GAIA, BFCL, SWE-Bench Verified, MMLU-Pro, GPQA Diamond, HELM, τ-Bench). Use when designing an eval suite, validating a prompt change, gating a deploy, or diagnosing production quality drift. Output: eval verdicts with statistical evidence, not vibe judgments.

ai-prompt-injection-defense

Operational playbook for defending LLM applications against prompt injection (direct + indirect), and red-teaming them in 2026. Covers OWASP Top 10 for LLM Applications v2025 (LLM01–LLM10), OWASP Top 10 for Agentic Applications 2026 (ASI01–ASI10), foundational research (Greshake et al. on indirect injection, Anthropic Many-Shot Jailbreaking, Constitutional AI), documented real-world exploits (Slack AI Aug 2024, EchoLeak CVE-2025-32711, GitHub Copilot RCE CVE-2025-53773, Bing/Sydney), defense layers (XML tag separation per Anthropic docs, structured outputs, content filters via Llama Guard 3 / OpenAI Moderation / Anthropic classifiers, NeMo Guardrails), and red-team tooling (garak, PyRIT, Promptfoo red-team mode). Use when auditing an LLM app for injection, designing defense-in-depth, mapping OWASP risks, or building a red-team suite. Output: red-team findings + defense recommendations with reproducible payloads.

ad-linkedin

What to produce for a LinkedIn ad — Sponsored Content (single image, video, carousel, document), Message Ads, and Text Ads. Encodes the exact slots (intro text, headline, description), character limits, asset ratios, the CTA enum per format, and the professional-context rules that decide whether B2B audiences engage or scroll past. Use whenever the task is producing LinkedIn ad copy or asset briefs. Strictly platform-specific — does not cover audience filters, matched audiences, or LinkedIn Insight Tag setup.

schedule Updated 23 days ago

social-mastodon

Ground-truth 2026 playbook for writing posts, replies, and threads on Mastodon and the wider Fediverse. Covers the strictly chronological timelines (no algorithm), hashtags as the sole discovery mechanism, the content-warning (CW) conventions, alt-text as a cultural norm, boost vs favorite semantics, instance selection, the local vs federated timeline, and why X/Bluesky tactics don't transfer. Activate when drafting for Mastodon or Fediverse-compatible servers (Pleroma, Akkoma, GoToSocial, Firefish, etc.).

video-spec-sheet

Canonical reference for platform-by-platform video output specs: aspect ratio, length cap, codec, bitrate, frame rate, color, caption style and safe-zone bands. Use when picking output settings, validating an export, or spec-checking before publishing. Covers TikTok, Reels, Shorts, IG/FB feed, Stories, X, LinkedIn — all four format families: short-form vertical, square/feed, long-form 16:9, story 9:16.