reasoning-with-parallel-attack-pathsUse when orchestrators need diverse security analysis by running the SAME attacker prompt across MULTIPLE LLMs (Claude, GPT-4, Gemini, DeepSeek) in parallel using LiteLLM
llm-evaluationImplement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.
behavior-first-architecture-analysisUse when analyzing codebase architecture to prevent keyword-based assumptions - forces execution tracing over name searching, requires counter-evidence before claiming gaps, provides parallel agent prompt templates for triangulating true system behavior
architecting-state-managementUse when designing state management for React features, architect agent needs recommendations, or evaluating state solutions - provides four-tier decision framework (TanStack Query for server state, Context for global client state, Zustand for complex features, useReducer for state machines) with sy
enforcing-information-architectureUse when refactoring frontend sections with 20+ files, deciding where new components belong, or organizing React codebases - provides file structure standards based on complexity tiers (flat, tab-based, hook-based, feature-module patterns) matching metrics/vulnerabilities gold standards
integrating-tanstack-componentsUse when integrating TanStack Router, Query, Table, and Virtual together - provides architecture patterns, data flow, and type-safe integration
auditing-agentsUse when auditing agents for compliance - validates against the gold standard pattern (EXTREMELY-IMPORTANT block, Step 1/2/3 structure, 7 universal skills). 9 audit phases.
creating-agentsUse when creating new agents - TDD workflow (RED-GREEN-REFACTOR) enforcing gold standard structure with EXTREMELY-IMPORTANT block, Step 1/2/3, and 7 universal skills.
fixing-agentsUse when fixing agent compliance issues - Claude applies all fixes using Edit tool with explicit patterns for each issue type (block scalar, name mismatch, field order, tables). No CLI dependency.
listing-agentsUse when listing all agents - displays agents grouped by 8 categories (architecture, development, testing, quality, analysis, research, orchestrator, mcp-tools) with descriptions and counts.
renaming-agentsUse when renaming agents safely - validates source/target, updates filename and frontmatter, finds and updates all references across commands/skills/agents, verifies integrity.
searching-agentsUse when finding agents by keyword - searches names, descriptions, types, skills with relevance scoring. Fast discovery in 30-60 seconds.
updating-agentsUse when modifying existing agents - guides through TDD update workflow (RED-GREEN) with compliance validation
verifying-agent-skill-invocationUse when verifying agent correctly uses mandatory skills - spawns agent with pressure scenarios, evaluates skill invocation via output metadata, reports PASS/FAIL per skill.
writing-claude-hooksThis skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse,
claude-marketplace-managementCreate and manage Claude Code plugin marketplaces. Use when creating marketplace.json, adding plugins to marketplaces, distributing plugins to teams via GitHub, configuring extraKnownMarketplaces in settings.json, or migrating plugins between marketplaces. Covers marketplace structure, plugin entrie
auditing-tool-wrappersUse when validating MCP tool wrapper compliance - runs 12-phase audit on wrappers and services
designing-progressive-loading-wrappersUse when architecting MCP wrappers or API SDK wrappers that need token/context optimization - covers progressive loading patterns, response filtering strategies, token reduction techniques for LLM context windows
fixing-tool-wrappersUse when remediating MCP tool wrapper compliance issues - auto-fixes deterministic problems found by audit
listing-toolsUse when viewing all available MCP tool services - lists services from .claude/tools/ with descriptions from package.json
orchestrating-api-tool-developmentUse when creating REST API wrappers with TDD - orchestrates test writing, implementation, and review for token-optimized HTTP client wrappers
orchestrating-mcp-developmentUse when creating MCP wrappers - orchestrates discovery, architecture, batched TDD, review, audit for 100% tool coverage
reviewing-mcp-wrappersUse when reviewing MCP wrapper implementations - provides HOW-to methodology for validating against architecture plans, checking code quality, running verification commands, and writing structured review documents with verdicts
setting-up-mcp-serversUse when setting up a new MCP server for wrapping - detects if MCP is configured, guides installation from npm/PyPI/GitHub/local, adds config to mcp-client.ts, verifies MCP responds. Supports npm (npx), Python (uvx), Go, and OAuth MCPs.
testing-mcp-wrappersUse when testing MCP wrappers - provides MCP-specific testing methodology for unit tests, schema validation, and integration tests with response filtering and token optimization patterns
updating-tool-wrappersUse when modifying existing MCP tool wrappers - test-guarded changes with regression prevention
mcp-tools-1passwordUse when accessing 1password services - provides 4 tools for get-item, list-items, read-secret, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-chariotUse when accessing chariot services - provides 2 tools for query, schema. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-chrome-devtoolsUse when accessing chrome-devtools services - provides 26 tools for click, close-page, drag, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-context7Use when accessing context7 services - provides 2 tools for get-library-docs, resolve-library-id. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-currentsUse when accessing currents services - provides 8 tools for get-projects, get-run-details, get-runs, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-featurebaseUse when accessing featurebase services - provides 25 tools for create-article, create-changelog, create-comment, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-ghydraUse when accessing ghydra services - provides 36 tools for Ghidra reverse engineering (instances, functions, memory, data, structs, analysis, UI). References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-jadx-mcp-serverUse when accessing jadx-mcp-server services - provides 19 tools for Android APK reverse engineering including class analysis, method/field inspection, code retrieval, and rename operations. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-linearUse when accessing Linear API - 57 exported GraphQL wrappers covering issues, projects, initiatives, cycles, teams, users, comments, labels, attachments, workflow states, reactions, archive operations, subscribers, favorites, documents, and issue relations via OAuth 2.0. Note 7 additional wrappers e
mcp-tools-nebulaUse when accessing nebula services - provides 5 tools for access-key-to-account-id, apollo, public-resources, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-perplexityUse when accessing perplexity services - provides 4 tools for perplexity_ask, perplexity_reason, perplexity_research, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-praetorian-cliUse when accessing praetorian-cli services - provides 15 tools for aegis-list, assets-get, assets-list, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-pyghidraUse when performing binary analysis with Ghidra - reverse engineering, vulnerability detection, decompilation. REQUIRES persistent PyGhidra MCP server. HTTP client only (no direct wrapper imports).
mcp-tools-registryUse when calling any external API or service (Linear, Praetorian CLI, GitHub, Playwright) - provides registry of MCP tool wrappers in .claude/tools/ with TRUE progressive loading achieving 0 tokens at session start (was 71.8k tokens). REQUIRED before implementing API calls. Tools load ONLY when impo
mcp-tools-salesforceUse when accessing salesforce services - provides 12 tools for create-scratch-org, delete-org, deploy-metadata, and more. References mcp-tools-registry for Bash + tsx execution patterns. Enables granular agent access control.
mcp-tools-serenaUse when accessing Serena services - provides 23 tools for semantic code analysis and editing via Language Server Protocol. Symbol operations, memory persistence, file search, and workflow guidance.
mcp-tools-shodan-apiUse when performing security reconnaissance with Shodan - provides token-optimized wrappers for host search, IP lookups, and DNS enumeration
claude-plugin-security-auditorAutomatically audits Claude Code plugins for security vulnerabilities, best practices, CLAUDE.md compliance, and quality standards when user mentions audit plugin, security review, or best practices check. Specific to claude-code-plugins repository standards.
claude-plugin-settingsThis skill should be used when the user asks about "plugin settings", "store plugin configuration", "user-configurable plugin", ".local.md files", "plugin state files", "read YAML frontmatter", "per-project plugin settings", or wants to make plugin behavior configurable. Documents the .claude/plugin
claude-plugin-structureThis skill should be used when the user asks to "create a plugin", "scaffold a plugin", "understand plugin structure", "organize plugin components", "set up plugin.json", "use ${CLAUDE_PLUGIN_ROOT}", "add commands/agents/skills/hooks", "configure auto-discovery", or needs guidance on plugin director
processing-large-artifactsUse when any artifact (document, codebase, PDF) exceeds practical context limits - provides parallel decomposition strategies for analysis, splitting, or verification using calibrated token thresholds. Invoke when user asks to analyze large files, split documentation, verify splits, or when files ex
auditing-skillsUse when validating skill compliance - reads skills, applies 30-phase rules from phase-details.md (index of split references), reports violations
closing-rationalization-loopholesUse when agents skip skill instructions despite explicit requirements
converting-books-to-skillsUse when converting technical books to skills with chapter detection, keyword extraction, and progressive disclosure.
converting-epub-to-markdownUse when converting EPUB ebooks to markdown format compatible with the books-to-skills workflow.
creating-skillsUse when creating new skills with TDD - guides through RED-GREEN-REFACTOR phases, progressive disclosure, location selection, and research integration
deleting-skillsUse when removing skills from the codebase - guides through reference discovery, impact analysis, user confirmation, directory removal, and cleanup verification to ensure no orphaned references remain
fixing-skillsUse when audit failed - applies fixes based on audit output by routing to appropriate handlers (Deterministic/Hybrid/Claude-Automated/Human-Required)
formatting-skill-outputUse when outputting structured findings (audit results, agent recommendations) - provides deterministic table formatting via CLI to solve non-deterministic Claude output
listing-skillsUse when viewing all available skills - lists skills from core and library with status information
migrating-skillsUse when promoting/demoting skills between core and library - handles two-tier system transitions including gateway routing updates, command structural pattern changes (skill: vs Read()), invocation syntax updates, and dependency tracking
pressure-testing-skill-contentUse when testing skills with pressure scenarios - applies RED-GREEN-REFACTOR to verify skills resist agent bypass attempts.
processing-large-skillsUse when needing to understand complex skills with multiple reference files - provides parallel analysis of skill components with semantic synthesis, dependency tracing, and cross-file relationship mapping. Invoke when user asks to understand, compare, or trace skills with 5+ files.
renaming-skillsUse when renaming existing skills - safe 11-step protocol with reference updates across gateways, commands, and other skills, plus integrity verification
searching-skillsUse when finding skills by keyword - searches core and library with fuzzy matching, scored results (exact > substring)
syncing-gatewaysUse when library skills are added, renamed, or deleted - synchronizes gateway routing tables with library skills, detects orphaned skills and broken paths using instruction-based workflows
updating-skillsUse when modifying existing skills - TDD update workflow (RED-GREEN) with research integration (orchestrating-research), Context7 staleness detection, progressive disclosure extraction, capacity planning (<500 lines), and compliance validation.
generate-blogpostUse when creating blog posts from GitHub repos - analyzes source code, captures author voice from reference materials, researches existing content, and synthesizes technically accurate posts with authentic voice
vuln-blogpostGenerate interactive branded HTML blog outlines from vulnerability show-and-tell transcripts with anonymization, Writer Mode, FAQ/AEO, code scaffolds, CVE cross-references, social drafts, and publishing checklists
adhering-to-makefile-best-practicesUse when creating or maintaining Makefiles - enforces modern patterns (pattern rules, automatic variables), prevents common mistakes (wildcard misuse, missing .PHONY), ensures maintainability (variables, progressive disclosure), and guides structure (namespacing, file organization)
aws-cognitoProvides AWS Cognito authentication and authorization patterns for Go backend. Use when integrating Cognito, implementing auth flows, or managing user pools.
burp-integrationUse when integrating Chariot with Burp Suite DAST via GraphQL API - implementing scan workflows, processing results, managing site provisioning, or debugging Burp API integration issues.
designing-go-interfacesUse when designing Go interfaces, deciding interface{} vs generics vs concrete types, creating library APIs, or reviewing interface design - applies "accept interfaces, return structs" and caller-defines-interface patterns from Dave Cheney and Uber Go. Handles "god interface", "premature abstraction
go-best-practicesUse when writing Go code - covers CLI structure (thin main.go + pkg/runner), function organization (important first, helpers last), and early returns to avoid nesting
go-errgroup-concurrencyUse when implementing parallel processing in Go with error handling - covers errgroup API, 7 concurrency patterns (mutex, atomics, channels, cancellation, stage-specific workers), production examples (TruffleHog 24K stars, Nuclei 26K stars), benchmarks, anti-patterns, and Chariot concurrency standar
implementing-go-pipelinesUse when implementing multi-stage processing pipelines in Go following the official Go blog canonical pattern - streaming data processing, producer-consumer patterns, graceful cancellation with done channel. Handles "goroutine leak", "pipeline deadlock", "memory exhaustion from unbuffered channels".
implementing-go-semaphore-poolsUse when implementing bounded concurrency in Go with semaphores and worker pools - rate limiting external APIs, high-throughput parallel processing (40K+ items/hour), variable-cost task scheduling. Complements go-errgroup-concurrency. Handles "too many goroutines", "rate limit exceeded", "out of mem
implementing-graphql-clientsUse when implementing GraphQL clients in Go with batch queries, rate limiting, and performance optimization - GitHub/GitLab/Azure DevOps APIs, cursor-based pagination, variable binding security. Handles "rate limit exceeded", "GraphQL injection", "slow API queries". Achieves 3x performance improveme
reviewing-backend-implementationsUse when executing code reviews for Go backend implementations - provides 5-step review process validating plan adherence, code quality, and verification commands with severity classification
structuring-go-projectsUse when evaluating or designing Go project structure - provides systematic framework for assessing compliance with official Go guidance and community patterns, decision trees for directory organization, and criteria for appropriate structure based on project complexity
building-web-crawlersUse when implementing web crawlers in Go for security scanning - provides Colly and Katana patterns, three-mode architecture (standard/headless/passive), event-driven callbacks, rate limiting, and Chariot platform integration
capabilities-janus-chainsUse when building Janus chains - workflow design, tool orchestration, state management
capabilities-scanner-integrationUse when integrating security scanners - Nmap, Masscan, output parsing, result normalization
capabilities-vql-developmentUse when developing VQL artifacts - Velociraptor query language, Aegis capabilities, artifact testing
creating-noseyparker-rulesenforcing-go-capability-architectureimplementing-detection-pluginsimplementing-go-cli-applicationsimplementing-go-plugin-registriesimplementing-yaml-template-enginesintegrating-standalone-capabilitiesmapping-python-dependencies-to-goorchestrating-nerva-developmentwriting-nerva-sctp-moduleswriting-nerva-tcp-udp-moduleswriting-nerva-testsoptimizing-go-scanner-performanceorchestrating-brutus-developmentorchestrating-capability-developmentporting-python-capabilities-to-goreviewing-capability-implementationsselecting-plugin-implementation-patterntranslating-python-idioms-to-govalidating-live-with-shodanverifying-port-equivalencewriting-nuclei-signaturesyaml-mastererror-handling-patternschariot-brand-guidelinesconstructing-graph-queriesdebugging-react-infinite-loopsenforcing-react-19-conventionsfrontend-animation-designercoordinating-competing-systemsgraph-layout-coordinationhierarchical-graph-navigationimplementing-semantic-zoom-graphsoptimizing-large-data-visualizationsub-clustering-algorithmstable-graph-hybrid-patternsvisualizing-with-d3working-with-sigma-jsmigrating-chariot-ui-componentsmigrating-saved-queriesoptimizing-react-performanceoptimizing-vite-buildsprefetching-react-routespreloading-react-resourcespreventing-react-hook-infinite-loopsreviewing-frontend-implementationssecuring-react-implementationsusing-context-apiusing-eslintusing-modern-react-patternsusing-shadcn-uiusing-tanstack-queryusing-tanstack-routerusing-tanstack-tabledeveloping-integrationsintegrating-with-awsintegrating-with-azureintegrating-with-azuredevopsintegrating-with-bitbucketintegrating-with-bugcrowdintegrating-with-extrahopintegrating-with-featurebaseintegrating-with-gcpintegrating-with-githubintegrating-with-gitlabintegrating-with-hackeroneintegrating-with-jiraintegrating-with-oracle-cloudintegrating-with-panoramaorchestrating-integration-developmentvalidating-integrationsoptimizing-neo4j-queriesquerying-neo4j-with-cypherorchestrating-feature-developmentbash-defensive-patternsavoiding-barrel-filesdocumenting-with-tsdocimplementing-branded-types-typescriptimplementing-result-either-patternimplementing-retry-with-backoffoptimizing-llm-api-responsesoptimizing-typescript-performancepattern-matching-typescriptsanitizing-inputs-securelystructuring-hexagonal-typescriptstructuring-workspace-packagesusing-typescript-const-assertionsusing-typescript-satisfies-operatorvalidating-with-zod-schemasdocxoptimizing-github-seopdfpptxxlsxmapping-to-cis-benchmarksmapping-to-cwemapping-to-doramapping-to-fda-cybersecuritymapping-to-fedrampmapping-to-iso-27001mapping-to-mitre-attackmapping-to-mitre-d3fendmapping-to-nist-csfmapping-to-sans-top-25cloud-advanced-patternscloud-cost-optimizationcloud-lambda-vs-ec2-decisionscreating-featurebase-contentsyncing-featurebase-contenttroubleshooting-featurebase-syncfirebase-cli-deploymentintegrating-with-linearorchestration-prompt-patternsadhering-to-uiux-lawscustomizing-nighthawk-profilesformatting-plextrac-findingsgenerating-security-test-instructionsnrf-revieweranalyzing-cve-customer-impactgenerating-threat-intelligence-reports