Search Results for "Mathews-Tom" | AgentCC

tavily

AI-optimized web search and content extraction via Tavily API. Use this skill when you need to search the web for current information, recent events, technical documentation, news, or any topic requiring live internet data. Triggers on: "search the web", "look this up online", "find recent information about", "search for", "google this", "web search", "look up", "find current data on", "what does the web say about", "research online", "fetch web content", "extract page content", "get article text", "scrape URL". Tavily returns clean, AI-ready snippets — prefer it over raw web fetch for search queries. Use extract for full page content when a specific URL is already known.

prompt-lab

Systematic prompt engineering: analyzes existing prompts for failure modes, generates structured variants (direct, few-shot, chain-of-thought), designs evaluation rubrics with weighted criteria, and produces test case suites for comparing prompt performance. Triggers on: "improve this prompt", "prompt engineering", "prompt lab", "generate prompt variants", "A/B test prompts", "evaluate this prompt", "prompt failure modes", "optimize prompt", "write a better prompt", "prompt design", "prompt iteration", "few-shot examples". Use this skill when designing, improving, or evaluating LLM prompts.

github

GitHub CLI operations via `gh` for issues, pull requests, CI/Actions, releases, repos, search, gists, and the REST/GraphQL API. Structured output with `--json` and `--jq` for parsing. Covers `gh issue create/list/view/edit/close`, `gh pr create/review/merge/checks`, `gh run list/view/rerun/watch`, `gh release create`, `gh search repos/issues/prs/code`, `gh api` for REST and GraphQL queries, and `gh gist` operations. Includes error handling for HTTP 401/403/404/422/429, scope troubleshooting, and rate limit management. Trigger phrases: "create an issue", "file a bug", "open a ticket", "submit a PR", "raise a pull request", "check pipeline", "view test results", "CI failing", "why did CI fail", "check CI status", "merge a PR", "manage releases", "query the GitHub API", "search repositories", "triage workflows", "automate GitHub operations". Also triggers when the user pastes a GitHub URL.

rag-auditor

Evaluates RAG (Retrieval-Augmented Generation) pipeline quality across retrieval and generation stages. Measures precision, recall, MRR for retrieval; groundedness, completeness, and hallucination rate for generation. Diagnoses failure root causes and recommends chunk, retrieval, and prompt improvements. Triggers on: "audit RAG", "RAG quality", "evaluate retrieval", "hallucination detection", "retrieval precision", "why is RAG failing", "RAG diagnosis", "retrieval quality", "RAG evaluation", "chunk quality", "RAG pipeline review", "grounding check". Use this skill when diagnosing or evaluating a RAG pipeline's quality.

adr-writer

Generates Architecture Decision Records (ADRs) capturing context, decision rationale, alternatives considered, and projected consequences. Produces numbered, status-tracked documents following the standard ADR format with proper metadata lifecycle. Triggers on: "write an ADR", "document this decision", "architecture decision record", "why did we choose", "capture this decision", "record the decision", "ADR for", "document the architecture", "decision record", "design decision", "technical decision". Use this skill when an architectural or technical decision needs to be documented.

code-refiner

Deep code simplification, refactoring, and quality refinement. Analyzes structural complexity, anti-patterns, and readability debt, then applies targeted refactoring preserving exact behavior. Language-agnostic: Python, Go, TypeScript/JavaScript, Rust. Use this skill when the goal is simplification and clarity rather than bug-finding. Triggers on: "simplify this code", "clean up my code", "refactor for clarity", "reduce complexity", "make this more readable", "code quality pass", "tech debt cleanup", "run the code refiner", "simplify recent changes", "this code is messy", "too much nesting", "this function is too long", "clean this up before I PR it", "tidy up my code", cyclomatic complexity, cognitive complexity, code smells.

doc-condenser

Transform verbose technical documentation into concise, scannable specs. Use this skill when you need to condense, summarize, or reformat technical docs, specs, or READMEs — including when a document is too verbose, when you want a technical summary, or when working with lengthy specification documents. Triggers on phrases like "make this concise", "too verbose", "condense this", "technical summary", "strip the fluff", or "reformat this spec". See assets/template.md for the standard output structure.

benchmark-runner

Designs structured benchmarks for comparing algorithms, models, or implementations. Selects appropriate metrics (latency, throughput, memory, accuracy), designs representative test cases, captures hardware/software context, produces comparison tables with tradeoff analysis, and includes reproduction instructions. Triggers on: "benchmark", "compare performance", "which is faster", "latency comparison", "memory comparison", "run benchmark", "design benchmark", "compare implementations", "evaluate algorithms", "performance comparison", "throughput test", "speed test". Use this skill when comparing two or more implementations, algorithms, or models.

architecture-reviewer

Architecture reviews across 7 dimensions: structural integrity, scalability, enterprise readiness (SOC2/HIPAA/GDPR/PCI-DSS), performance, security, operational excellence, and data architecture. Produces scored reports with prioritized recommendations. Three modes: (1) Codebase review — evidence-based analysis of source code, configs, IaC; (2) Document review — risk-based analysis of design docs, RFCs, specs; (3) Hybrid — drift detection between intent and implementation. Triggers on: "review architecture", "critique design", "audit system", "evaluate codebase", "find design flaws", "assess scalability", "check security", "enterprise readiness", "architecture assessment", "technical due diligence", or when user provides a system design document or codebase and asks for feedback or improvements.

manuscript-provenance

Computational provenance audit verifying that every number, table, figure, ordering, and terminology in a manuscript is derived from code and scripts — not manually entered. Cross-references LaTeX source against the codebase to detect hardcoded values, stale outputs, broken pipelines, and manual data entry. Companion to manuscript-review: that skill audits the document as prose; this skill audits whether the document is faithfully generated from code. Use when the user says "check provenance", "verify reproducibility", "audit my pipeline", "are my numbers from code", "check manuscript against scripts", "provenance audit", or any request to verify that manuscript content traces back to computational outputs.

filesystem

File and directory operations using Claude Code built-in tools — replaces the Filesystem MCP server. Maps all 11 MCP tools to native equivalents: Read, Write, Edit, Glob, Grep, and Bash. Covers file reading with line ranges, parallel reads, pattern-based file search, regex content search, directory listing, tree traversal, move/copy/rename, and metadata inspection. Trigger phrases: "read this file", "write to file", "create a file", "edit file", "find files matching", "search for text in files", "list directory", "show directory tree", "move file", "rename file", "copy file", "file info", "find all Python files", "search codebase for". Use this skill when performing file operations, navigating codebases, or managing directories.

agent-builder

Build AI agents and automate Claude Code programmatically using the Claude Agent SDK and headless CLI mode. Use this skill when you need to build an agent, create a Claude agent, make a bot, work with the agent SDK, run Claude in headless mode, write programmatic agent code, automate with Claude, create an MCP server builder, or query Claude programmatically. Covers the Python SDK, the claude -p headless interface, custom tool creation with SDK MCP servers, hooks for deterministic control, session management, and CLI flag reference. Authentication uses existing ~/.claude/ config — no API keys required.

repo-sentinel

Full security audit and enforcement for public repositories across 12 attack surfaces: git history, source code, docs, config, .gitignore recon, CI/CD, containers, dependencies, binaries, metadata, platform-specific (GitHub/GitLab), license compliance, and community surface. Provides fast-path and full 20-check audits, pre-commit hooks, CI gates, .gitignore generation, and history scrubbing. Use whenever pushing to a public remote, open-sourcing a repo, writing README/docs, configuring CI/CD or Dockerfiles, adding dependencies, or checking license compliance. Trigger on: push to GitHub, make repo public, open source this, set up the repo, write README, add CI/CD, create Dockerfile, set up pre-commit, add license, write SECURITY.md, secret leaks, credential rotation, .claude/ tracking, repo hygiene, security scanning, or is this safe to push, pre-oss, open source readiness, release audit, or open source audit. This is the gatekeeper between internal and public.

sql-optimizer

Analyzes SQL queries for performance issues: missing indexes, N+1 patterns, suboptimal joins, full table scans. Interprets EXPLAIN output, detects anti-patterns, recommends indexes, and rewrites queries with detailed explanations of each optimization. Triggers on: "optimize this query", "slow query", "query performance", "add indexes", "index recommendation", "explain plan", "query plan", "N+1 query", "fix this SQL", "SQL performance", "optimize SQL", "why is this query slow", "index strategy". Use this skill when given a SQL query that needs performance analysis or optimization.

skill-evaluator

Evaluate Claude Code skill quality across 6 weighted dimensions: frontmatter quality, trigger coverage, structural completeness, content depth, consistency and integrity, and CONTRIBUTING.md compliance. Produces scored audit reports with severity-classified findings and actionable recommendations. Two modes: (1) Quick Audit — single skill, full per-dimension report with weighted scoring; (2) Full Audit — all skills in repo, comparative ranking table plus per-skill summaries. Triggers on: "evaluate skill", "audit skill quality", "score skill", "skill review", "check skill completeness", "rate this skill", "skill quality check", "grade skill", "assess skill", "skill audit", "how good is this skill", or when a user asks for feedback on a SKILL.md file. Use this skill when reviewing skills before deployment, comparing skill quality across a repo, or diagnosing why a skill fails to activate on relevant queries.

regex-builder

Generates, explains, and tests regular expression patterns. Builds patterns from positive and negative examples, breaks down each component with a readable explanation table, generates edge-case test suites, and provides usage examples in Python and JavaScript. Triggers on: "regex for", "regular expression", "pattern for", "match strings like", "extract from", "build regex", "create pattern", "explain this regex", "what does this regex do", "regex to match", "parse with regex", "validate format". Use this skill when building, explaining, or testing a regular expression pattern.

sequential-thinking

Structured, reflective problem-solving through sequential chain-of-thought reasoning. Replaces the Sequential Thinking MCP server (sequentialthinking tool). Use this skill when facing complex multi-step problems, planning and design tasks, analysis that may need course correction, debugging with iterative hypothesis testing, architectural decisions requiring trade-off evaluation, or any task where the full scope is unclear initially. Trigger on: sequential thinking, step by step reasoning, chain of thought, break down problem, think through this, structured analysis, multi-step problem solving, reflective reasoning, thought process, deep analysis, complex reasoning, hypothesis verification, branching analysis, revision-based thinking, dynamic problem solving.

dependency-audit

Audits project dependencies for license compliance, maintenance health, security vulnerabilities, and bloat. Analyzes both direct and transitive dependency trees, detects abandoned packages, identifies license conflicts (copyleft, unknown), checks for known CVEs, and finds unused or duplicate dependencies. Triggers on: "audit dependencies", "dependency check", "license check", "dependency health", "abandoned packages", "bloat check", "unused dependencies", "security audit dependencies", "dependency review", "license compliance", "package audit", "supply chain", "dependency risk". Use this skill when reviewing project dependencies for risk.

pr-review

Pull request and code review with diff-based routing across five dimensions: code quality and guideline compliance, test coverage analysis, silent failure detection, type design and invariant analysis, and comment quality auditing. Classifies changed files and loads only relevant review methodologies. Produces severity-ranked findings (Critical, Important, Suggestion) with confidence scoring. Replaces pr-review-toolkit plugin. Trigger phrases: "review my PR", "review this code", "check my changes", "is this ready to merge", "audit this PR", "review before committing", "check code quality", "any issues with this code", "pre-merge review", "look over my changes", "code review". Use this skill when reviewing code before commit or merge, checking PR quality, or when the user asks for feedback on recent modifications.

task-decomposer

Breaks down large features into implementable tasks with dependency mapping, edge case identification, test strategy planning, and phased execution order. Produces task tables with effort sizing, parallelization flags, and risk flags for each phase. Triggers on: "break down this feature", "decompose", "task breakdown", "how should I implement", "implementation plan", "what are the steps for", "edge cases for", "plan this feature", "implementation steps", "break this into tasks", "work breakdown", "project breakdown". Use this skill when a feature or project needs to be broken into actionable steps.

estimate-calibrator

Produces calibrated three-point estimates (best/likely/worst case) with explicit unknowns, confidence intervals, and assumption documentation. Breaks work into atomic units, identifies technical and scope uncertainties, calculates PERT ranges, and provides confidence rationale. Triggers on: "estimate this", "how long will this take", "effort estimate", "time estimate", "best case worst case", "confidence interval", "sizing", "estimate effort", "how big is this", "story points", "t-shirt sizing", "estimate the work", "project estimate". Use this skill when a task or project needs an effort estimate with explicit uncertainty.

api-docs-generator

Audits and enhances API documentation for FastAPI and REST endpoints. Identifies missing descriptions, incomplete response codes, missing examples, and generates enhanced docstrings, Pydantic model examples, and OpenAPI spec improvements. Triggers on: "generate API docs", "document this API", "OpenAPI for", "add examples to", "improve docstrings", "API documentation audit", "FastAPI docs", "document endpoints", "API reference", "swagger docs", "REST API docs", "endpoint documentation", "response documentation". Use this skill when API endpoints need documentation or documentation audit.

concept-to-video

Turn any concept into an animated explainer video using Manim (Python). Use whenever the user wants animated visualizations, motion graphics, or video output (MP4/GIF) for technical concepts. Covers: architecture animations, data flow visualizations, algorithm step-throughs, pipeline explainers, math proofs, comparisons, agent interactions, training loops, or any "animate X" / "make a video of X" request. Also trigger for existing Manim scene edits or re-renders. Trigger on: "create a video", "animate this", "make an explainer", "concept to video", "manim animation", "show this as a video", "motion graphic", "visualize this process", or any concept + video/animation output request. Use this skill even for casual "animate" requests about technical ideas.

mcp-to-skill

Convert MCP (Model Context Protocol) servers into on-demand skills to dramatically reduce active context window usage. MCP tool schemas consume tokens on every turn whether used or not; skills load only when needed. Analyzes MCP tool definitions, classifies each tool by replacement strategy, and generates a complete skill package. Use this skill whenever the user mentions: convert MCP, MCP to skill, reduce context size, too many tools, tool token bloat, slim down MCPs, context window full, replace MCP, MCP migration. Also trigger when the user has many active MCP servers and wants to optimize context usage, asks about reducing token overhead from tool definitions, or says my context is too big or running out of context when MCPs are active.

migration-risk-analyzer

Analyzes database migration scripts for risk: lock contention, downtime estimation, rollback strategy, and deployment recommendations. Classifies DDL and DML operations by lock type and duration, identifies irreversible changes, generates rollback scripts, and produces pre/post validation queries. Triggers on: "analyze this migration", "migration risk", "is this migration safe", "schema change risk", "DDL risk", "downtime for this migration", "lock risk", "rollback strategy", "how to roll back", "migration review", "alter table risk". Use this skill when reviewing database migrations before deployment.

changelog-composer

Generates structured changelogs and release notes from git history and PR descriptions. Classifies changes into breaking, features, fixes, performance, and docs. Filters internal-only changes, detects breaking changes, and produces human-readable entries linked to source PRs. Triggers on: "generate changelog", "write release notes", "compose changelog", "what changed since", "changes since last release", "prepare release", "release notes for", "changelog for", "summarize changes", "diff since tag". Use this skill when preparing a release and needing to summarize changes for users.

md-to-pdf

Convert Markdown files to professionally styled PDF documents with full support for Mermaid diagrams, LaTeX/KaTeX math equations, tables, syntax-highlighted code blocks, and all standard Markdown features. Use this skill whenever the user wants to convert a .md file to .pdf, generate a PDF report from Markdown, or produce print-ready documents from Markdown sources containing diagrams, math, or complex formatting. Trigger phrases include: "convert markdown to pdf", "make a pdf from this md", "render this markdown", "export markdown as pdf", "markdown to pdf with diagrams", "pdf from markdown with equations", "generate a pdf report", "how do I export markdown as PDF", "convert my notes to PDF", "can you make a PDF from this markdown".

debug-investigator

Systematic debugging framework: symptom capture, evidence analysis, hypothesis generation, bisection strategy, log analysis, instrumentation, and minimal reproduction design. Transforms ad-hoc debugging into structured investigation with ranked hypotheses and concrete investigation plans. Triggers on: "debug this", "why is this failing", "root cause", "investigate", "diagnose", "this error", "this exception", "this stacktrace", "not working", "broken", "unexpected behavior", "bisect", "narrow down", "isolate the issue", "help me debug". Use this skill when confronted with a bug, error, or unexpected behavior in any codebase.

architecture-diagram

Generate detailed layered architecture diagrams as self-contained HTML artifacts with inline SVG icons, CSS Grid nested container layout, SVG path connection overlays, and color-coded connection legends. Use when the user asks to create architecture diagrams, infrastructure diagrams, system topology diagrams, network diagrams, cloud architecture visuals, deployment diagrams, integration flow diagrams, or any request involving visual representation of system components, their containment hierarchy, and interconnections. Triggers on terms like "architecture diagram", "infra diagram", "system diagram", "topology", "deployment diagram", "network diagram", "integration architecture", or when the user provides a system description and asks for a visual/diagram.

html-presentation

Convert any document, outline, or content into a polished, self-contained HTML slide presentation. Use this skill whenever the user wants to create a presentation, slide deck, pitch deck, or talk from a document, markdown file, outline, notes, or any textual content and wants the output as an HTML file (not PPTX). Also trigger when the user asks for an "HTML presentation", "web-based slides", "reveal.js deck", "browser presentation", or wants to convert a document into slides they can present from a browser. Supports two navigation modes - horizontal (left/right, Reveal.js-powered) and vertical scroll (top-to-bottom, keyboard/scroll navigation). Includes multiple visual themes (dark editorial, light minimal, corporate, hacker terminal). Consider asking the user for clarification on theme, navigation direction, and content structure if not already specified.

manuscript-review

Systematic pre-publication manuscript audit producing a structured refactoring report with section-level diagnostics, citation hygiene analysis, and submission-readiness assessment. Use this skill whenever the user uploads a manuscript, paper, thesis chapter, journal submission, or conference paper and asks for review, feedback, editing, refactoring, pre-submission check, proofreading, or quality audit. Also trigger when the user says "review my paper", "check before submission", "is this ready to submit", "pre-pub checklist", "manuscript review", "refactor my paper", or asks about citation consistency, argument coherence, or formatting compliance. Covers partial requests like "check my references" or "does the abstract work" — the full diagnostic surfaces issues across all facets even when only one was asked about.

web-fetch

Web content fetching and URL retrieval via curl and WebFetch — replaces the Fetch MCP server (fetch_html, fetch_json, fetch_markdown, fetch_txt). Covers HTTP GET/POST requests, JSON API consumption with jq, HTML retrieval, markdown conversion, plain text extraction, authenticated requests, redirects, cookies, and timeouts. Trigger phrases: "fetch this URL", "get the page content", "download HTML", "call this API", "curl this endpoint", "grab the JSON", "fetch markdown from", "retrieve web content", "hit this endpoint", "scrape this page", "read this URL", "pull data from API", "make an HTTP request". Use this skill when the user provides a URL and wants its content, when consuming REST APIs, or when extracting readable content from web pages.

YouTube Analysis

Extract YouTube video transcripts and produce structured concept analysis with multi-level summaries, key concepts, and actionable takeaways. Pure Python, no API keys, no MCP dependency. Fetches transcripts via youtube-transcript-api with a yt-dlp fallback, then Claude analyzes the content directly. Use this skill when the user mentions: analyze youtube video, youtube transcript, summarize this video, what does this video cover, extract concepts from video, video analysis, watch this for me, break down this talk, youtube URL, video summary, lecture notes from video, podcast transcript, conference talk notes, tech talk breakdown, video key points, TL;DR of video, video takeaways, or pastes any URL containing youtube.com or youtu.be.

test-harness

Generates comprehensive pytest test suites: happy path, edge cases, error conditions, fixture scaffolding, mock strategy, and async patterns. Analyzes function signatures, dependency chains, and complexity hotspots to produce runnable, parametrized test files. Triggers on: "generate tests", "write tests for", "test this function", "test this class", "create test suite", "what tests should I write", "pytest for", "unit tests for", "mock strategy for", "how to test", "test coverage for", "write a test file". Use this skill when given a Python function, class, or module and asked to produce tests.

concept-to-image

Turn any concept, idea, or description into a polished static HTML visual, then export it as a PNG or SVG image. Use this skill whenever the user wants to create a visual representation of an idea and needs an image file output (PNG or SVG). This includes: infographics, concept diagrams, flowcharts, comparison charts, process visuals, educational diagrams, social media graphics, data visualizations, posters, cards, badges, icons, logos sketches, or any "make me an image of X" request that can be achieved with HTML/CSS/SVG rather than photographic AI generation. Also trigger when the user has an existing HTML visual and wants to export/convert it to PNG or SVG. Trigger phrases include: "create an image of", "make a visual", "design a graphic", "export as PNG", "save as SVG", "concept to image", "turn this into an image", "screenshot this HTML", "generate an infographic", or any request combining a concept description with image output.

linkedin-post-style

Write LinkedIn posts matching a specific technical author's voice — direct, analytical, dry-humored, and precise. Use this skill whenever the user asks to write, draft, rewrite, or edit a LinkedIn post, social media post, tech commentary, or any public-facing short-form writing about technology, AI, software engineering, or developer tools. Also trigger when the user says "write this in my style", "post about this", "rewrite this for LinkedIn", "draft a post in my style", or provides raw content/notes and wants it shaped into a post.

static-web-artifacts-builder

Suite of tools for creating elaborate, self-contained static HTML artifacts — infographics, interactive diagrams, architecture visuals, data dashboards, and rich visual deliverables. Use this skill when asked to create an interactive HTML artifact, build a self-contained web component, generate a static HTML visualization, make an interactive diagram, produce a visual infographic, or render any high-density visual as a single HTML file. Zero build toolchain — no React, no Vite, no Parcel. Pure HTML5 + CSS3 (Grid/Flexbox) + inline SVG. Triggers on: "create interactive HTML", "build self-contained web component", "generate static HTML visualization", "make an interactive diagram", "produce infographic", "render as HTML artifact", "build a visual dashboard", "create a diagram I can open in browser".

gpu-optimizer

Expert GPU optimization for modern consumer GPUs (8-24GB VRAM). Use this skill when you need to optimize GPU training, speed up CUDA code, reduce OOM errors, tune XGBoost for GPU, migrate NumPy to CuPy, make a model faster, manage GPU memory, optimize VRAM usage, or benchmark PyTorch. Covers mixed precision, gradient checkpointing, XGBoost GPU acceleration, CuPy/cuDF migration, vectorization, torch.compile, and diagnostics. NVIDIA GPUs only. PyTorch, XGBoost, and RAPIDS frameworks.