AI & ML

dezvin Bundle

Use this skill when creating or compressing answers, documents, prompts, instructions, skills, agents, reports, handoffs, code-facing text, or other artifacts so they retain required meaning and usefulness with less content and denser natural language. Trigger for token-efficient, concise without loss, remove bloat, semantic minification, dense writing, minimal sufficient artifact, or requests that an agent stop expanding simple work. Do not use for context-window management, tool-result pruning, exact length quotas, ordinary summarization that may omit meaning, or requests for exhaustive expansion.

dezvin Bundle

Use this skill only when explicitly invoked as $document-consolidator to analyze document sprawl, compress Markdown/text documents, consolidate multiple documents into candidate or final documents, or plan cleanup without losing evidence, source roles, protected facts, rules, decisions, links, or unfinished work. Supports .md and .txt only. Do not use for skill files, prompt files, docx/pdf, ordinary copyediting, publication writing, code refactors, or implicit summarization.

dezvin Bundle

Use this skill for standalone image-prompt creation, refactoring, model/style/format adaptation, transparent-background and isolated-asset prompting, image editing, reference-based prompting, weak-result repair, prompt sets, and controlled variants from a brief, idea, image, or existing prompt. Selects a task-fit art direction when needed and translates it into medium-specific production controls. Produces prompt text but does not generate images. Do not use when a separate project-specific visual workflow owns the request or when the primary deliverable is broader content rather than an image prompt.

epistemedeus Skill

Read credential-free public webpages as structured JSON or clean LLM-ready Markdown. Use GET /extract for one public HTTPS page, one bounded POST /extract/batch for 2–5 caller-supplied public HTTPS URLs with explicit desired fields, or GET /read for a JSON record containing Markdown. A caller may explicitly request a one-item batch. Do not use for authenticated or private-network content.

epistemedeus Skill

Statically scan a public GitHub repository for supply-chain risk before an agent installs or runs it. Use for MCP servers, Agent Skills, packages, or dependencies when checking exfiltration sinks, credential reads, obfuscated execution, environment harvesting plus network use, or install-time curl-pipe-shell patterns. The scanner never runs the target, and a clean result is not permission to execute it.

epistemedeus Bundle

Evaluate delegated-wallet, embedded-wallet, or agent-signer policy evidence without sharing credentials or raw provider payloads. Use to test exact approved-action shape, duplicated calls, authorization drift, replay, sequential cumulative caps, signed-but-unbroadcast accounting, ABI extraction, concurrent oversubscription, counter references, or application serialization before funding an autonomous wallet.

epistemedeus Bundle

Evaluate a funded agent work opportunity before committing effort. Use when an agent needs an attempt, verify-first, or abandon decision from reward, time, compute, mandatory spend, reusable value, selection probability, competition, slots, agent-access, acceptance, settlement, and optional dated platform evidence. The skill does not claim, bid, pay, or submit on the source platform.

epistemedeus Bundle

Find paid Agent402 or MPP machine services that match a capability intent and formally guarantee buyer-required JSON output paths, or audit one already-known seller route for the same guarantee. Use when an agent needs to search for a machine-paid API by the data it must return, reject underconstrained seller schemas before payment, distinguish machine-buyable GET routes from contract-ready POST routes, or create a verified non-spending purchase intent for SameDayDesk's contract-qualified search or seller-integrity audit.

fiveonecode Bundle

Audit whether a spec, task, harness, agent workflow, safety boundary, or operational contract is actually enforced by concrete mechanisms. Use when reviewing or changing AGENTS.md, CLAUDE.md, .agents/manifests, .agents/verify, .agents/evals, spec/agents, build/test specs, task plans, risk/execution contracts, source-of-truth boundaries, session closeout rules, verifier classifications, or any explicit request to check whether a promised guarantee is real.

fiveonecode Bundle

Build and improve agent-first engineering harnesses where AI agents perform most implementation work and humans steer architecture, constraints, and review. Use when defining or upgrading AGENTS.md rules, repository conventions, task decomposition, CI guardrails, merge strategy, quality gates, or cleanup loops to increase autonomous coding throughput and reliability.

fiveonecode Skill

Select and run the right local LLM engine (MLX with speculative decoding vs llama.cpp) with the right settings for a task, and enforce the one-engine-at-a-time memory rule. Use whenever a task runs local inference against an OpenAI-compatible endpoint — evals, summarization, code review, content generation — or when choosing between local engines serving the same model.

fiveonecode Bundle

Transcribe long local meeting audio into a validated rough transcript with bounded chunking, repetition and output-limit detection, targeted retries, optional stronger-model rescue, and auditable token usage. Use when a user provides an audio recording or asks for reliable multilingual meeting transcription.

fiveonecode Bundle

Refactor and review SwiftUI view files with strong defaults for small dedicated subviews, MV-over-MVVM data flow, stable view trees, explicit dependency injection, and correct Observation usage. Use when cleaning up a SwiftUI view, splitting long bodies, removing inline actions or side effects, reducing computed `some View` helpers, or standardizing `@Observable` and view model initialization patterns.

fulcradynamics Bundle

Exchange messages with agents on OTHER Fulcra accounts — a different person's assistant, a site agent, a teammate's bot — using a dedicated outbox channel and a narrow datashare. Use when a user says connect my agent to X's agent, set up a mesh, send this to another account's agent, or check what another agent sent us.

fulcradynamics Bundle

Manages agent progress reporting and OKF-compliant memory syncing to Fulcra.

fulcradynamics Skill

Connects an agent to Fulcra through the CLI two-step login or the MCP connector, and recovers from a network-restricted shell.

fulcradynamics Skill

Deprecated — renamed to fulcra-workspaces. Use the fulcra-workspaces skill instead.

fulcradynamics Bundle

Guides a new user or agent through the initial setup, configuration, and capabilities of the Fulcra environment.

fulcradynamics Bundle

Manages agent memory backup, restoration, rollback, and cloning using Fulcra's versioned file storage.

fulcradynamics Bundle

Sets up a secure, local-only administration control panel for the Fulcra environment. Features an interactive Chat Envoy for direct agent communication, and a File Store Explorer for browsing the user's data backend.

rosslevinsky Bundle

Iterative plan refinement between Claude and Codex. Each agent writes an initial plan using a condensed v2 plan methodology embedded in the skill (mirroring plan-init's content model), then they alternately critique and incorporate the other's best ideas, with a judge checking for convergence after each round. Three exits: convergence (score ≥8/10, from round 3 onward), stagnation (no score improvement over 3 consecutive rounds), or the 10-round cap. Writes a summary.md with a score trajectory, a pointer to the winning plan, and a breakdown of remaining differences with pros/cons. Supports resume from an interrupted run. Use when the user invokes /plan-duel, or says "duel the plans", "run plan duel", "cross-compare plans with codex" — the input is a problem statement the skill generates competing plans from, not two existing plans to compare against each other. Argument: the problem statement (inline text or a file path), or the path to an existing plan-duels workdir to resume.

netvar1337 Bundle

Practical SRC / crowdsource testing / bug bounty vulnerability hunting workflow skill. Includes: 5-phase methodology (intake → recon → enum → hunt → report), 19 attack-class playbooks (SQLi/XSS/RCE/SSRF/IDOR/CSRF/Path Traversal/File Upload/SSTI/XXE/Race/HTTP Smuggling/OAuth/JWT/SAML/GraphQL/Mobile/LLM/DoS), 305 structured payloads, 263 WAF/EDR bypass variants, 2,887 real disclosed High/Critical HackerOne cases, 77,000+ WooYun case statistics, a Chinese-domestic OA / middleware fingerprint library, and banking / telecom industry-vertical playbooks. Triggers when the user mentions 'SRC hunting / SRC vulnerability hunting / bug bounty / crowdsource testing / hackerone / bug bounty / SRC / arbitrary X vulnerability / penetration testing' or asks 'how to hunt a target / how to test an API / how to bypass a WAF'.

rasahq Skill

Enables and configures LLM-powered Contextual Response Rephraser in Rasa CALM. Covers endpoints.yml setup, per-response metadata, and prompt customization. Use when setting up or tuning the Contextual Response Rephraser.

rasahq Skill

Connects external sub agents to a Rasa CALM assistant via the A2A (Agent-to-Agent) protocol. Use when adding an external agent, configuring an agent card, setting up A2A authentication, customizing input/output processing, or invoking an external sub agent from a flow.

rasahq Skill

Configures MCP (Model Context Protocol) servers in a Rasa CALM assistant. Use when defining MCP servers in endpoints.yml, setting up authentication (API key, OAuth 2.0, pre-issued token), or connecting to multiple external services.

rasahq Skill

Configures model_groups in endpoints.yml for LLM and embedding providers. Covers single and multi-deployment setups, routing strategies, failover, self-hosted models, and caching. Use when adding or changing LLM providers or setting up multi-LLM routing.

rasahq Skill

Use when a user wants to evaluate or test a Rasa assistant using the eval scenario framework — at any stage of the process: setting up the evaluation infrastructure for the first time, deciding which scenario types to cover for a flow, writing or generating scenario YAML files, learning about supported assertion types or how to write good success criteria, running goal-driven simulations where an LLM plays the user role against a live Rasa bot, or analyzing whether existing scenarios are comprehensive. Covers both the authoring side ("write scenarios", "generate evals", "help me write criteria") and the execution side ("simulate a conversation", "run my eval scenarios"). Not for scripted end-to-end tests, unit tests, NLU training, or general Rasa bot development.

rasahq Skill

Calls MCP tools directly from Rasa flow steps. Use when invoking an MCP tool via a call step, mapping slot values to tool parameters, or extracting tool results into slots. This replaces custom action code for simple API integrations.

rasahq Skill

Adds knowledge base search to a Rasa CALM assistant using EnterpriseSearchPolicy, connects vector stores (Faiss, Milvus, Qdrant), and configures generative or extractive search modes. Use when setting up RAG, connecting a vector store, overriding pattern_search, or implementing a custom information retriever.

aidvizhhub Bundle

МЫШЛЕНИЕ ПЕЩЕРНИКА: Tree-of-Thoughts с откатом, критический процесс (петля план→вызов→проверка фактами), итерация задачи (Draft v1 → аудит по 3 критериям → откат → вывод). LLM-резонер: пейперы про backtracking, самокоррекция = ловушка, промпт с откатом. Триггеры: подумай, реши, откат, backtrack, ветка, спираль, тупик, аудит, критпроцесс, проверь себя, правильность, правильная методика, валидация, корректность подхода, метод правильный. EN: Thinking with backtracking: Tree-of-Thoughts, audit by 3 criteria, rollback and retry; truth = facts, not feelings. Use when a problem is hard, uncertain, or stuck.

aidvizhhub Bundle

ЧИНИМ БЕЗ ГАДАНИЯ: локальные улики (логи/dmesg/состояние) → веб по уликам → сопоставил → лечи, одна переменная за раз. Фиксы: MCP Unknown tool (codemode false), camoufox Connection closed (убить процесс → disconnect → connect), Firefox плеер не работает, WirePlumber 0.5 убил Lua. Триггеры: не работает, зависло, сломалось, логи, dmesg, улики, unknown tool, connection closed, mcp, кауфми лёг, диагноз, почему, версии, совместимость, gpu-задача на cpu, nvidia-smi. EN: Fix without guessing: local clues (logs/dmesg/state) → web by clues → treat, one variable at a time. Use when something is broken, hangs, or fails.

novoads Bundle

Generate AI video ads and image ads through the Novoads REST API (api.novoads.ai). Use when the user wants a UGC video, a product video, a talking-head ad, an AI actor holding a product, a TikTok or Reels or Shorts ad, a static image ad, or asks to "make me an ad", "generate a video", or names a model (Seedance, Seedance 2.5, Seedance Mini, Omni Flash, Veo 3.1, Sora 2, GPT Image 2, Nano Banana Pro, Reve). Handles upload, dialogue approval, cost confirmation, generation, polling, QA, and download. A product photo lands here when it SEEDS a new scene; when the picture itself should move, unchanged, that is image-to-motion. Not for editing an existing video file and not for publishing to an ad platform.

novoads Bundle

Clones a STATIC IMAGE ad into finished ads for your own product, into a reusable prompt TEMPLATE, or both — the template is what makes the ads and the ads are what prove the template. Round-trips the frame through gpt-image-2, nano-banana-pro or reve-2.1 until it reproduces, replaces every brand-specific with a {placeholder}, then fills it with your product pinned to a real photo. Use for "clone this ad for my product", "rebuild this competitor's ad for us", "make me ads like this", "reverse engineer this ad", "make this ad reusable", "add this to my prompt library", or "clone my competitors' ads" with NO file attached — it sources one via spy-competitor-ads rather than asking which. **Among cloners the source's medium decides**: a still, or nothing attached, lands here; "clone this ad" plus a VIDEO is clone-video-ad. NOT for generating an ad from a brief (chatgpt-image-ad, nano-banana-image-ad), nor animating a still as-is (image-to-motion). Refuses to render your product without a real photo of it.

novoads Bundle

Clones a VIDEO ad end to end for a different product or offer, on the Novoads API. The deliverable is a finished rendered MP4 — the WHOLE ad, not its opening, and not a template. Reads the source's style, pacing, camera work, dialogue and tone, adapts it into a Seedance 2.0 prompt, prices it, renders it, files the source beside the clones and diffs each clip's transcript against the approved script. Use for "clone this video ad", "recreate this video for my brand", "make this ad but for my product" when the source is a video, "clone my competitors' VIDEO ads", or a video ad handed over with a product photo. **Among cloners the source's medium decides, not the wording**: "clone this ad" plus a video lands here; a still, or nothing attached, is clone-image-ad, and animating a still as-is is image-to-motion. This pack has no hook-only skill: for the opening alone, clone the whole ad and cut. If it does not move, wrong skill.

novoads Bundle

Generate one or more standalone Meta image-ad creatives via Nano Banana Pro through the Novoads API. Locks the model, auto-strips platform chrome, enforces edge-safe layouts. Use when the user asks for a "Nano Banana ad", "Gemini image ad", "nano-banana-pro ad creative", "make a static image ad with Gemini", or anchors on a need for photoreal / lifestyle / multi-reference / handheld-object / clay-texture ad creatives (sticky-note flatlays, held-whiteboard signs, lifestyle portraits, ingredient collages, OOH photography). Does NOT trigger on ChatGPT Image cues — use chatgpt-image-ad for those.

novoads Bundle

Generate high-CTR YouTube thumbnails using Nano Banana Pro via the Novoads API. Handles reference image upload, character likeness alignment, proven CTR-tested prompt formulas, and bounded batch generation. Use when the user asks to create a YouTube thumbnail, video thumbnail, A/B test thumbnail variations, or refers to thumbnail design with their face, brand assets, or product photos.