{
  "markdown": "# Agent Skills\n\nA collection of skills for AI coding agents. Skills are packaged instructions that extend agent capabilities with domain-specific knowledge.\n\nSkills follow the [Agent Skills](https://agentskills.io/) format.\n\n## Install\n\n```bash\nnpx skills add vladmdgolam/agent-skills\n```\n\nOr install a specific skill:\n\n```bash\nnpx skills add vladmdgolam/agent-skills --skill cinema4d-mcp\n```\n\n## Available Skills\n\n### 🎬 cinema4d-mcp\n\nCinema 4D MCP expert for extracting scene data, writing C4D Python scripts, and controlling Cinema 4D through MCP tools.\n\n**Use when:**\n- Using Cinema 4D MCP tools (`get_scene_info`, `list_objects`, `execute_python_script`, etc.)\n- Writing Python scripts for C4D extraction or manipulation\n- Working with MoGraph cloners, effectors, and fields\n- Baking animation data from C4D scenes\n- Debugging C4D Python API errors\n\n**Covers:**\n- 7 critical C4D API gotchas (world/local coords, visibility constants, sequential stepping, etc.)\n- MoGraph extraction and animation track discovery patterns\n- When to use structured MCP tools vs `execute_python_script`\n- Known error tables with fixes (Python API + MCP tool errors)\n- Redshift availability matrix\n- Timeout management and chunked baking strategies\n\n### 🧊 blender-mcp\n\nBlender MCP expert for scene inspection, Python scripting, GLTF export, and material/animation extraction.\n\n**Use when:**\n- Using Blender MCP tools (`get_scene_info`, `execute_python`, `screenshot`, etc.)\n- Writing Blender Python scripts for scene extraction or manipulation\n- Exporting scenes to GLTF/GLB for web (Three.js, R3F)\n- Debugging material or texture export losses\n- Optimizing GLB files with gltf-transform\n- Using asset integrations (PolyHaven, Sketchfab, Hyper3D Rodin, Hunyuan3D)\n\n**Covers:**\n- 7 critical rules (export timeouts, modifier handling, Draco pitfalls, name mapping, etc.)\n- Scene hierarchy, material, and animation extraction patterns\n- Headless CLI export (bypasses MCP timeout)\n- Material export survival matrix (what survives GLTF, what doesn't)\n- Texture optimization pipeline (resize → WebP → Draco)\n- Known error tables with fixes (MCP, export, Python API, texture paths)\n\n### ⏱️ time-lens\n\nAnalyze and visualize time spent on software projects by combining data from multiple sources: WakaTime coding time, git commit session detection, Claude Code usage, Codex CLI usage, and Cursor IDE usage.\n\n**Use when:**\n- Analyzing work hours or calculating time spent on a project\n- Generating a work hours report or visualizing coding activity\n- Creating a project time breakdown\n- Summarizing development effort across date ranges\n\n**Covers:**\n- 5 data sources: WakaTime API, git sessions, Claude Code prompts, Codex CLI prompts, Cursor IDE prompts\n- Interactive HTML dashboard (dark-themed, Chart.js)\n- Markdown report with ASCII charts\n- Reconciliation logic across overlapping sources\n\n### 🖨️ pdf-look-scanned\n\nMake PDF documents look like they were scanned on a physical scanner, with optional signature replacement.\n\n**Use when:**\n- Making a PDF look scanned (grayscale, noise, blur, rotation, edge shadows)\n- Replacing digital signatures with real handwritten ones from a source PDF\n- Preparing documents that need a \"printed and scanned\" appearance\n\n**Covers:**\n- Configurable scan effects (DPI, noise, blur, contrast, JPEG quality)\n- Signature extraction from source PDFs (phone photos, scans)\n- Multi-signature replacement on different pages\n- Coordinate discovery workflow for signature placement\n- Metadata considerations\n\n### 🎨 figma-context-mcp\n\nExpert guide for using the Figma Context MCP (Framelink) efficiently while avoiding 429 rate-limit errors.\n\n**Use when:**\n- Using Framelink Figma MCP tools (`get_figma_data`, `download_figma_images`)\n- Fetching Figma designs for code generation\n- User hits 429 rate limits from Figma API\n- Extracting design tokens or component data from Figma\n- Downloading Figma assets/images\n- Any design-to-code workflow involving Figma URLs\n\n**Covers:**\n- Metadata-first, prune-first, fetch-last pipeline (2-3 API calls instead of dozens)\n- Depth-limited node tree fetching to keep responses <500 KB\n- Batch image downloads with deduplication\n- Plan-tier awareness (limits tied to file owner's plan, not yours)\n- Rate-limit diagnostics and recovery steps\n- Common anti-patterns that cause 429 lockouts\n- Alternative workflows when rate-limited (screenshots, manual export)\n\n### 📚 apple-books-mcp\n\nExtracts highlights, annotations, and book data from Apple Books via MCP.\n\n**Use when:**\n- Exporting book highlights with colors and notes\n- Extracting annotations from specific books or a reading list\n- Searching highlights across your entire library\n- Matching books from an external list against Apple Books library\n\n**Covers:**\n- Efficient batch color extraction (5 API calls instead of 150+)\n- Color map building via `get_highlights_by_color` + Python cross-referencing\n- Annotation data structure (style values, color mapping, underlines)\n- Markdown export patterns with color-coded emoji indicators\n- Book matching strategies (translations, partial titles, multiple editions)\n- Troubleshooting common issues (empty results, null colors, large outputs)\n\n### 👁️ visual-feedback-loop\n\nAI agent visual feedback loop for capturing and inspecting rendered output from any web app during iterative development.\n\n**Use when:**\n- Iterating on visual code (UI, canvas, 3D, SVG, charts, animations)\n- You need to SEE the rendered result to evaluate quality\n- Doing visual QA or A/B comparison of render variants\n\n**Covers:**\n- Dev-screenshot API pattern (GET trigger → SSE → client capture → POST back → file saved)\n- Parameterized offscreen rendering with custom params\n- WebMCP tool registration for Chrome Canary agents\n- Console and Chrome MCP fallbacks\n- Next.js reference implementation\n\n### 🪛 reverse-engineer-js\n\nLLM-assisted deobfuscation of minified or obfuscated JavaScript bundles. Wraps [humanifyjs](https://github.com/jehna/humanify) with the right defaults, the pseudo-TTY workaround, and patterns for both whole-bundle and single-module workflows.\n\n**Use when:**\n- Deobfuscating `bundle.js`, fxhash projects, or any minified JS\n- The user says \"humanify this\", \"deobfuscate this\", \"rename mangled vars\", \"what does bundle.js do\"\n- Working in a project with an `og/` or `modules/` folder fed by humanify\n- humanify crashed with `cursorTo is not a function`\n\n**Covers:**\n- Default model: `gemini-3.1-flash-lite-preview`\n- Pseudo-TTY workaround (`script -q /dev/null ...`) for non-interactive shells\n- Single-module extraction from large bundles\n- `webcrack` pre-split for webpacked bundles\n- Sourcemap-first check (often beats LLM rename for free)\n- Common backends: gemini / openai / local\n\n### 🔍 agent-sessions\n\nSearch, list, and resume AI agent sessions across Claude Code, Codex CLI, Gemini CLI, opencode, Hermes Agent, and Cline CLI from the terminal.\n\n**Use when:**\n- Finding a past conversation (\"find that session where I worked on...\")\n- Reading latest/last messages from a session id (`ses_...`, UUID, or Cline id)\n- Resuming a previous Claude Code session by ID\n- Listing recent activity across agents\n- Searching session history for a topic or project\n\n**Covers:**\n- CLI tool at `~/Play/radar/tools/agent-sessions` with `--agent`, `--search`, `--search-scope`, `--project` filters\n- Direct handoff flow: `--pack <session-id> --max-chars 8000` before manual filesystem or database inspection\n- Compact JSON output for scripting; `--include-search-text` is opt-in for transcript indexes\n- Compact handoff snippets via `--context <session-id...>`\n- Bounded evidence packs and excerpts via `--pack <id...>` and `--read <id...> --query`; use `--role user` for user-only messages\n- Claude session resume via `--resume <id>`\n- Data sources: `~/.claude/projects/`, `~/.codex/sessions/`, `~/.gemini/tmp/*/chats/`, `~/.local/share/opencode/opencode.db`, `~/.hermes/state.db`, `~/.cline/data/db/sessions.db`\n- Companion tools: `claude-history` (TUI), Claude Code History Viewer (GUI)\n\n### 🌊 fluid-interfaces\n\nDesign and review principles for fluid, gestural, physically-responsive interfaces, distilled from Apple's WWDC 2018 \"Designing Fluid Interfaces\" talk (the team that built the iPhone X gestural UI).\n\n**Use when:**\n- Designing or reviewing touch/gesture interactions (drag, swipe, dismiss, dock)\n- Tuning spring-based motion, elastic/rubberband behaviors, or scroll physics\n- Something \"feels off\", \"feels janky\", or \"doesn't feel native\"\n- Deciding between timed animations and continuous spring behaviors\n\n**Covers:**\n- 8 core principles: instant response, redirectability/interruption, spatial consistency, gesture hinting, lightweight-input/amplified-output, rubberbanding, smooth frames of motion, behavior-over-animation\n- Spring tuning via `damping`/`response` instead of raw mass/stiffness/duration\n- Momentum projection formula for \"throw to nearest endpoint\" interactions (with Swift code)\n- Gesture mechanics checklist: taps, swipes, one-to-one tracking, continuous feedback, resolving competing gestures\n- Teaching techniques for gestural UI (visual cues, elevation, paired animations, explanations, designing for play)\n\n### ⚡ threejs-perf-loading\n\nPerformance and loading patterns for real-time Three.js/WebGL/WebGPU sites, based on modern approaches and best practices from top-notch studios and developers (Ivress, Threejspunk, igloo.inc, Noomo Agency) plus live production profiling, rather than generic advice.\n\n**Use when:**\n- Eliminating the loading-spinner-to-scene freeze/jank (uncompiled shader pipelines hitting the first real frame)\n- Designing an adaptive-quality or perf-budget system, or desktop/mobile post-processing tiers — including hardening the FPS sampler against false-positive degrades\n- Attributing an unexplained slow frame to a named cause with in-app tooling\n- Tuning scroll/camera damping so input can't \"outrun\" a max speed\n- Deciding where to spend a render-loop's cost (MRT routing, layer-split passes, on-demand shadows, GPU-resident particles, proximity gating)\n\n**Covers:**\n- Shader/pipeline warmup: force every material to compile behind the loading screen before firing the reveal event, including per-camera `compileAsync` for layer-masked cameras\n- Loader reveal gating on real rendered frames (not asset progress or timers), paused-not-hidden loader content, and chunking one-shot procedural synthesis (reverb impulses, noise textures) off the critical frame\n- Render-loop cost discipline: MRT emissive routing for selective bloom, layer-split passes, `shadowMap.autoUpdate = false`, GPU-resident particle motion, proximity + hysteresis gating, fused post-processing passes, stable-identity effects arrays, and auditing hidden render-target allocations in library material wrappers (remount dispose storms)\n- Adaptive quality: one flat perf-budget config object, one-way FPS-latched staged degradation with four production-tested false-positive guards (post-arm discard windows, outlier-stall window rejection, settle windows, escalation gated on the cheaper lever having failed), UA-based gating, genuinely different desktop/mobile post tiers\n- Frame attribution tooling: per-frame marks buffer, GL-resource delta sampler, dispose tracer with call-site stacks, module-scope boot tracer for the pre-mount window, React commit profiling — and treating a fully-instrumented-yet-silent slow frame as proof of GC/GPU-driver cost\n- Scroll/input smoothing: step-clamped damped lerp (the `maxSpeed` primitive behind drei's `ScrollControls`) vs. direct authored-timeline scrubbing\n- Transmission/glass materials: owning your render target for a static-fallback trick, and why a shared pre-blurred buffer beats a hand-optimized per-fragment sampler\n\n## Adding Skills\n\nEach skill lives in `skills/<skill-name>/` with a required `SKILL.md` and optional `references/`, `scripts/`, and `assets/` directories.\n\n## License\n\nMIT\n",
  "bytes": 11917,
  "sha": "77faf790f3e3505c8d2f283bcc80dc9a6d9b5b7f1cf9d5855fb7a119f398dc55",
  "repo_slug": "vladmdgolam/agent-skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_vladmdgolam_agent_skills_blender_mcp_16a2d7f7/readme"
}