ratacats-skills

Use when designing, auditing, or evolving a repository and its engineering workflow for coding agents — AGENTS.md or CLAUDE.md, repo-local skills, agent-readable documentation, plans and memory, scripts and guardrails, isolated workspaces, sandboxing, observability, verification, evals, or issue-driven orchestration.

1 skill

Find and download ebooks (PDF/EPUB/MOBI) by title/author, or look up published book content. Pair with ebook-extractor to convert downloads to text.

1 skill

Create, review, convert, or embed terminal text art: ASCII/ANSI art, CP437, block ASCII, BBS logos, CLI banners, NFOs, SAUCE/XBIN/BIN/ANS output.

1 skill

Chained clarity review of a repo, plan, or work items — combines code review, plan refinement, deep-module architecture, DRY/cruft, and naming analysis. Use for clarify/defuzz/audit/harden/fresh-eyes requests.

1 skill

Probe a bounded target from many angles at once — audit for hidden bugs/stale state, stress-test a claim or plan, or brainstorm wide. Falsifiable or generative probes through explicit lenses, each closed with evidence.

1 skill

Use when asked to crystallize a workflow, skill, codebase, or agent runs, or to find where agents repeatedly rediscover methods, apply stable rules, or do unnecessary work. Analysis of opportunities to replace repeated agent effort with reusable procedures; not a general code cleanup or an automatic implementation run.

1 skill

Become Diana. Ukrainian software engineer in NYC. KGB accent. Zero filter. Real corpus mode with full main tweets and top-ranked replies. Once invoked there is no going back.

1 skill

Duris MUD reference — load for any Duris task: connecting, navigation, combat, spellcasting, zones, geography, equipment, mobs, output parsing, recovery.

1 skill

Use when designing or operating Duris group combat with tanks, assists, assumed targets, caster memorization readiness, bash risk, tracking enemies, aggro adds, or command spam risk.

1 skill

Extract plain text from ebooks (EPUB, MOBI, PDF) for analysis, processing, or reading.

1 skill

Use when reviewing a codebase, uncommitted changes, recent agent work, or one subsystem for hidden bugs, flawed assumptions, drift, duplication, weak tests, broken contracts, or other failure modes.

1 skill

Run a GPT Pro / pro-cli project revision loop: context package, GPT Pro prompt, adversarial review, design map, revision passes, local verification, goal assessments, handoff.

1 skill

Deep web research with Graph-of-Thoughts exploration — parallel branching, relevance scoring, knowledge-graph storage, synthesis into themed reports. Use for researching topics, companies, people, or concepts.

1 skill

Use when designing or reviewing a module's interface, deciding where a seam goes, finding shallow modules or deepening opportunities, running an architecture review, making code more testable or AI-navigable, or extracting code repeated across three or more projects into a shared library, skill, template, or config. Also supplies the deep-module vocabulary (module, interface, depth, seam, adapter, leverage, locality) when another skill or review needs it.

1 skill

Build with Jev, TypeSafe's System One model — fast structured judgments (Choice, Score, Noul) that code consumes directly. Use when replacing fragile parsing or prompt-and-parse LLM steps with typed decisions, or when designing routing, ranking, classification, extraction, verification, or guardrail flows on the TypeSafe API.

1 skill

Use when writing or rewriting nonfiction prose for readers — essays, blog posts, articles, handbooks, newsletters, threads, intros, or any draft that must hold a stranger's attention. Covers choosing the topic, writing a hook-first intro, outlining, speedrunning a first draft, rewriting for clarity, succinctness, and intrigue, feedback scoring, and authentic voice. Not for code comments, API docs, or marketing copy.

1 skill

Context and working knowledge for the Kalshi prediction-market domain. Use this skill whenever the user asks about Kalshi markets, tickers, order books, pricing, settlement, or the Kalshi API/WebSocket.

1 skill

Answer a hard question by building a persistent graph of atomic claims, each given a 0-1 credence by Jev from a shared evidence ledger. Keeps false and unexamined claims. Use for research questions, codebase questions, hypothesis testing, or any claim that needs careful, skeptical support.

1 skill

Read paywalled Medium articles in full, including Medium-hosted publications (towardsdatascience.com, betterprogramming.pub, etc.).

1 skill

Explain, compare, or choose machine-learning objectives and learning regimes, including masked prediction, self-supervision, generative modeling, reinforcement learning, transfer, and classical methods. Use for ML learning-method questions and training-design tradeoffs; not routine model API usage or infrastructure debugging.

1 skill

Use when reviewing or designing names for plans, beads, work items, architecture docs, codebases, APIs, schemas, modules, states, or domain vocabulary where terminology drift, overloaded concepts, boundary ambiguity, or AI-confusing naming may exist.

1 skill

Automatically use whenever a model starts looking at gates, promotion states, smoke tests, instrumentation, telemetry, observability, dashboards, ledgers, manifests, certificates, audits, reports, handoffs, fresh-eyes reviews, verification theater, or completion labels such as ready, green, promoted, validated, closed, or complete. Also use when those artifacts accumulate while requested functionality is still missing, or when reviewing whether process has become a substitute for delivery.

1 skill

Price-blind deep research for a Polymarket/prediction-market event or PMKNB situation — fresh or updated analysis, resolution mechanics, mechanism-only briefs, source discovery, 0-100 likelihood independent of market odds.

1 skill

Analyze one PM event for PMKNB: identify exact Polymarket/Kalshi instruments (conditionId, token IDs, tickers), assess market forces, resolution/oracle risk, liquidity, and counterparty risk; output forecast, report, brief, proposal, or no-publish.

1 skill

Manual-only; never auto-load. Use when explicitly named, or for the general research-methodology framework when no dedicated methodology skill fits.

1 skill

Direct-invocation-only; never auto-load. Use only when explicitly named to create or update a PMKNB situation frame.

1 skill

Use when a bug, regression, or misbehavior has an unknown cause and the user wants iterative hypothesis-driven investigation — invoked by name ("prediction drill", "drill into this bug"), or when asked to falsify hypotheses and converge on a root cause and fix through repeated prediction rounds.

1 skill

Use when a consequential design, architecture, process, or policy choice depends on unknowns that reasoning alone cannot settle, or when an investigation plan may be over-scoped, prematurely branched, or full of experiments whose results may not change a decision. Not for questions answerable by a straightforward lookup alone.

1 skill

Use when auditing, scrutinizing, or strengthening an existing test suite — unit, integration, or e2e — because tests pass but bugs still ship, coverage looks high while trust is low, tests were written after the code or by another agent, tests are flaky and need fixing without loosening them, or the user asks to review, harden, or refine tests with fresh eyes.

1 skill

Idempotent revision pass for literary and nonfiction prose. Invoke with /remove-ai-sins. Removes seven craft sins only — the not-X-but-Y frame, naming an emotion or theme the detail already carries, explaining an image or joke, inflated register, a summing-up ending, tying a raised question shut, and prose about the prose instead of the subject. If none of the seven are present, return the text unchanged. Do not restyle, do not add images or morals, do not apply general AI-slop lists (those are no-ai-slop / unslop). Use when asked to revise, restrain, unsin, strip craft sins, run a restraint pass, or when a draft needs those sins removed.

1 skill

Show rate-limit and quota usage across the user's AI coding subscriptions (Claude, Codex, Gemini, Kimi, MiniMax, Z.ai, Qwen, Grok) with per-window percent used and time to reset. Use when the user asks about usage, quotas, rate limits, or when subscriptions reset.

1 skill

Use when user asks to create, write, edit, or test a skill. Also use when documenting reusable techniques, patterns, or workflows for future Claude instances.

1 skill

Manual-only; never auto-load. Use only when explicitly named (x-undocumented-api / x-api-skill).

1 skill