Best Skills (2026)
If you sign up through a link on this page, we may earn a commission — it never affects our rankings.
We tracked, tested and compared every Skills tool on Agent Pantheon to rank the 10 best for 2026. Below is the shortlist with our take on each, followed by the full, searchable directory.
Skills by the numbers
Pricing mix
Best Skills (2026)
- 1
Manage Headers (Grade A)Security-tested development skill for Claude AI. Grade A. Inspects and configures the security headers a Power Pages site sends to browsers — Content Security Policy, frame and clickjacking protection - 2
Using Git Worktrees (Grade A)Security-tested data-ai skill for Claude AI. Grade A. Use when starting feature work that needs isolation from current workspace or before executing implementation plans - creates isolated git worktre - 3
Ga4 Bigquery Schema (Grade A)Security-tested data-ai skill for Claude AI. Grade A. GA4 BigQuery Export Schema Reference — complete field reference, nested structures, query patterns, and performance tips - 4
Meta Capi (Grade A)Security-tested data-ai skill for Claude AI. Grade A. Meta Conversions API (CAPI) Setup Reference — architecture, event types, customer information hashing, deduplication, implementation examples, AEM - 5
Callees (Grade A)Security-tested development skill for Claude AI. Grade A. Lista o que uma funcao/metodo chama (call graph direto) - 6
Test Module Name (Grade A)Security-tested data-ai skill for Claude AI. Grade A. Name Haskell test modules after the module under test with a Spec suffix in the same namespace. Use when writing or reviewing Haskell test module - 7
Board Of Directors (Grade A)Security-tested data-ai skill for Claude AI. Grade A. Simulate a 5-member expert board deliberation for major decisions. Use when evaluating plans, architecture choices, feature designs, or any decisi - 8
Advpl Mvc Avancado (Grade A)Security-tested development skill for Claude AI. Grade A. MVC avançado via PE (Pontos de Entrada) — adicionar grids customizadas em telas MVC padrão (CNTA300/MATA070/MATA440/MATA460/FINA040 via *STRU) - 9
Advpl Fundamentals (Grade A)Security-tested development skill for Claude AI. Grade A. Fundamentos ADVPL/TLPP — notação húngara (c/n/d/l/a/o/b/x/u), naming convention, escopos Local/Static/Private/Public, limite de 10 chars em .p - 10WWorkflow Engine (Grade A)Security-tested devops skill for Claude AI. Grade A. **UTILITY SKILL** — Machine-readable workflow DAG for the multi-step agent pipeline. Defines node types, edge conditions, gates, and fan-out patter
Manage Headers (Grade A)
Security-tested development skill for Claude AI. Grade A. Inspects and configures the security headers a Power Pages site sends to browsers — Content Security Policy, frame and clickjacking protection
Manage Headers (Grade A) is a security-tested development skill for Claude AI. It is designed to inspect and configure the security headers a Power Pages site sends to browsers, including Content Security Policy, frame and clickjacking protection, cross-origin sharing, cookie behavior, and related site settings. The skill identifies gaps and walks the user through fixes, making it useful for tasks such as reviewing headers, fixing CSP errors, controlling cross-origin access, hardening cookie settings, or checking whether browser settings are safe. It can be used when the user wants to achieve various security-related goals without specifically mentioning 'security headers.'
- Inspects current security headers
- Configures Content Security Policy
- Enables frame and clickjacking protection
- Manages cross-origin sharing and cookie behavior
Using Git Worktrees (Grade A)
Security-tested data-ai skill for Claude AI. Grade A. Use when starting feature work that needs isolation from current workspace or before executing implementation plans - creates isolated git worktre
Use when starting feature work that needs isolation from current workspace or before executing implementation plans. This skill creates isolated git worktrees with smart directory selection and safety verification. Git worktrees create isolated workspaces sharing the same repository, allowing work on multiple branches simultaneously without switching. The core principle is systematic directory selection plus safety verification, leading to reliable isolation. The directory selection process follows these steps: 1. Check for existing directories with priority order: hidden ".worktrees" directory, then "worktrees" directory. If found, use that one. 2. Check if a preference for a worktree directory is specified in the "CLAUDE.md" file. 3. If no directory exists and no CLAUDE.md preference is found, ask the user to choose a directory. The safety verification step is critical and involves checking if the chosen directory is ignored by Git. If not ignored, the skill will add the directory to the ".gitignore" file and commit the changes. Creation steps include: 1. Detecting the project name and creating a full path for the worktree. 2. Creating the worktree with a new branch. 3. Running project setup, which auto-detects and runs the appropriate setup commands (e.g., installing dependencies) based on the project type (Node.js, Rust, Python, or Go). 4. Verifying a clean baseline by running tests. If tests fail, report failures and ask whether to proceed or investigate. 5. Reporting the worktree location and its readiness to implement a feature.
- Isolated workspace creation
- Smart directory selection
- Safety verification
- Automated project setup
Ga4 Bigquery Schema (Grade A)
Security-tested data-ai skill for Claude AI. Grade A. GA4 BigQuery Export Schema Reference — complete field reference, nested structures, query patterns, and performance tips
GA4 BigQuery Export Schema Reference is a security-tested data-ai skill for Claude AI. It provides a complete field reference, nested structures, query patterns, and performance tips for Google Analytics 4 BigQuery export schema. The tool helps users understand the GA4 export schema, write correct BigQuery SQL queries against GA4 data, and follow performance best practices.
- Complete field reference
- Nested structures
- Query patterns
- Performance tips
- Ready-to-use SQL examples
Meta Capi (Grade A)
Security-tested data-ai skill for Claude AI. Grade A. Meta Conversions API (CAPI) Setup Reference — architecture, event types, customer information hashing, deduplication, implementation examples, AEM
Meta Capi is a security-tested data-ai skill for Claude AI, providing a comprehensive reference for setting up Meta Conversions API (CAPI). It covers architecture, event types, customer information hashing, deduplication, implementation examples, and Aggregated Event Measurement. The tool helps users implement CAPI correctly, debug event tracking issues, improve Event Match Quality, and configure deduplication. It offers precise answers with ready-to-use code examples and can utilize Cogny MCP tools to inspect user configurations and provide contextual recommendations.
- CAPI architecture overview
- Event type parameters
- Customer information hashing
- Deduplication strategies
- Aggregated Event Measurement
- Node.js implementation example
Callees (Grade A)
Security-tested development skill for Claude AI. Grade A. Lista o que uma funcao/metodo chama (call graph direto)
Callees (Grade A) is a security-tested development skill for Claude AI. It identifies what a function or method calls, generating a direct call graph. The skill is useful for mapping dependencies of a function before extracting or moving it, identifying usage of restricted or deprecated functions, and estimating the impact of a refactoring. It also provides information about the callee, such as the function name, line number, and call type (native ERP, restricted, user function, or external). By using /plugadvpl:callees, users can input the name of a function to see what it calls, with options for formatting the output as a table or in markdown. It is part of the plugadvpl toolkit. This skill is a useful addition to a Claude AI's development capabilities, providing essential insights into the codebase.
- Direct call graph listing
- Function call type identification (native ERP, restricted, user function, external)
- Line of call site information
- Support for multiple output formats (e.g., Markdown)
Test Module Name (Grade A)
Security-tested data-ai skill for Claude AI. Grade A. Name Haskell test modules after the module under test with a Spec suffix in the same namespace. Use when writing or reviewing Haskell test module
Name Haskell test modules after the module under test with a Spec suffix in the same namespace. This is done to clearly separate library and test directories, making it easier to identify dependencies and relationships between modules. In particular, this best practice is beneficial when writing or reviewing Haskell test module names or test file organization. For example, a test module named Env.TypeSpec should be placed in the same namespace as the module it is testing, Env.Type. This ensures that the relationships between modules are clear, and it is particularly useful for modules with complex behavior, such as those that depend on multiple modules or have no dependencies on specific modules.
- Suggests test module names based on the module under test
- Appends 'Spec' to the module name
- Same namespace as the module under test
Board Of Directors (Grade A)
Security-tested data-ai skill for Claude AI. Grade A. Simulate a 5-member expert board deliberation for major decisions. Use when evaluating plans, architecture choices, feature designs, or any decisi
The Board of Directors skill simulates a 5-member expert board deliberation for major decisions. It evaluates plans, architecture choices, feature designs, or any decision requiring multi-perspective expert analysis. The board consists of five roles: Chief Architect (CA), Chief Product Officer (CPO), Chief Security Officer (CSO), Chief Operations Officer (COO), and Chief Experience Officer (CXO). Each director brings domain expertise and can challenge other directors' opinions. The deliberation process involves individual assessments, board discussions, and a final vote to reach a consensus.
- Simulates a 5-member expert board
- Individual assessments by each director
- Board discussion and rebuttals
- Final vote and consensus-building
- Role-based evaluation (CA, CPO, CSO, COO, CXO)
Advpl Mvc Avancado (Grade A)
Security-tested development skill for Claude AI. Grade A. MVC avançado via PE (Pontos de Entrada) — adicionar grids customizadas em telas MVC padrão (CNTA300/MATA070/MATA440/MATA460/FINA040 via *STRU)
Advpl Mvc Avancado (Grade A) is a security-tested development skill for Claude AI. It is designed for customizing MVC padrão (CNTA300/MATA070/MATA440/MATA460/FINA040 via *STRU) by adding grids, addTrigger em cascata, and validations while preserving original fields. It is used for customizing a routine MVC TOTVS via Ponto de Entrada (A300STRU, MA440STRU, etc.) and has features like validação cruzada between grids pai/filha em cadastro padrão, cálculo fiscal, and bloqueio condicional de edição baseado em status do registro mestre. It also supports integration with MATXFIS, refactoring MsNewGetDados, and adding custom fields.
- Customization of standard MVC routines via Entry Points (PE)
- Addition of custom grids and triggers to standard screens
- Implementation of specific business logic and validations
- Support for various TOTVS modules (e.g., MATA010, MATA070, MATA440)
Advpl Fundamentals (Grade A)
Security-tested development skill for Claude AI. Grade A. Fundamentos ADVPL/TLPP — notação húngara (c/n/d/l/a/o/b/x/u), naming convention, escopos Local/Static/Private/Public, limite de 10 chars em .p
AdvPL Fundamentals (Grade A) is a security-tested development skill for Claude AI. It focuses on fundamentals of ADVPL/TLPP, including Hungarian notation, naming convention, local and global scopes, and name length limits. ADVPL (Advanced Protheus Language) and its successor TLPP are proprietary languages derived from Clipper/xBase. They have strong conventions that are not optional in high-quality Protheus code. This skill is essential for Protheus developers, especially when creating or refactoring variables and functions. The skill addresses common pitfalls, such as the silent bug in ADVPL classic, where a variable's name truncates to 10 characters, and the importance of using the correct scope for variables. It covers when to use local, static, private, and public variables, including rules of thumb for their visibility and lifetime. The skill is designed to be a practical guide for developers to avoid common mistakes and improve their ADVPL/TLPP coding skills.
- Hungarian notation
- Naming conventions
- Local/Static/Private/Public scopes
- 10-character limit for .prw/.prx files
Workflow Engine (Grade A)
Security-tested devops skill for Claude AI. Grade A. **UTILITY SKILL** — Machine-readable workflow DAG for the multi-step agent pipeline. Defines node types, edge conditions, gates, and fan-out patter
Workflow Engine (Grade A) is a utility skill that helps with defining a machine-readable workflow DAG for the multi-step agent pipeline. It enables the creation of pipelines with node types, edge conditions, gates, and fan-out patterns. This skill is designed to meet the requirements of a Grade A security-tested environment, suggesting that it is well-suited for use in a high-security devops environment with Claude AI.
- Machine-readable workflow DAG
- Node type definitions
- Edge condition management
- Gate and fan-out pattern support
- Multi-step agent pipeline management
Browse all 967 Skills tools
The complete, searchable directory — ranked by real user reviews.