Explore AI Agent Skills & Claude Prompts

Synergy skill combining fd (JAR locator), javap/jar (bytecode tools), and rg (call filter) to investigate Java library internals at bytecode level. Use when tracing what methods actually do vs what documentation claims: debugging framework behavior (Hibernate query execution, Spring transactions, JPA streaming, Blaze-Persistence delegation), discovering misleading APIs (getResultStream() secretly calling getResultList(), eager vs lazy loading), finding memory leaks in "lazy" code, locating implementations across Maven JARs, analyzing multi-class delegation chains, verifying performance claims, understanding unexpected framework behavior. Trigger phrases: "trace this method", "why is this slow", "does it really stream", "find implementation", "debug [framework] internals", "verify this claim", "locate this class", "check lazy loading", "is it truly lazy". Pipeline: fd finds JARs → jar tf lists classes → javap -c decompiles bytecode → rg filters method calls. Essential when documentation is unclear, misleading,

javap

Use javap, jar, and jdeps (all JDK built-in tools) to analyze Java bytecode, inspect library internals without source code, and trace actual method behavior. Load this skill when debugging framework behavior (Hibernate, Spring, JPA, Blaze-Persistence), verifying library claims ("does this really stream or does it load everything?"), tracing delegation chains across JARs, finding where methods are actually implemented, locating classes in Maven cache, discovering misleading APIs (methods that internally do the opposite of their name), analyzing dependencies, checking for memory leaks (ArrayList instantiation in "streaming" code), or understanding why a library behaves unexpectedly. Use even when the user doesn't say "javap" — phrases like "trace what this method does", "find this class in Maven", "why is this slow", "does it really stream", "verify the implementation", "debug Hibernate internals", "check the delegation chain" all trigger this skill. Critical when documentation is unclear, misleading, or absent

rg

⚡ IMPORTANT: Auto-trigger for text search. Use when: search for text, find files containing, locate TODOs, security scan secrets, trace class usage, grep pattern matching. Returns only matching lines. 30-98% token savings. 3-4x faster than grep/Select-String.

schedule Updated 21 days ago

rg

Use ripgrep (rg) to search text content inside files from the command line instead of Select-String (PowerShell), grep -r (Bash), or reading entire files. Load this skill for searching across codebases: finding annotations in catalog files, locating component owners, searching for API keys/secrets/tokens (security audit), tracking imports or dependencies, finding configuration values, discovering class usage, identifying error patterns, or searching logs. Use even when the user doesn't say "ripgrep" or "rg" — phrases like "search for", "find files containing", "which components use this system", "locate all TODOs", "find the annotation", "security scan for secrets", "what files import this module" all trigger this skill. rg respects .gitignore (skips build artifacts), outputs compact relative paths, and is 3-4x faster than Select-String on 10k+ file codebases. Essential for Backstage catalog operations (find unmapped components, search owners, locate systems by annotation). Saves 30-98% of tokens by extractin

xq

Use yq -p xml to read, extract, filter, or transform XML data from the command line instead of reading entire XML files. Load this skill for Maven projects (pom.xml), .NET projects (.csproj, .config, app.config), Java configuration (web.xml, persistence.xml, applicationContext.xml, Spring beans), build files (build.xml, .proj), or any XML configuration. Use when extracting Maven dependencies, checking project versions, listing artifacts, finding dependency scopes, analyzing .NET package references, inspecting servlet mappings, or auditing XML configs across a multi-module project. No separate tool installation needed — this is just yq with the -p xml flag. This skill applies even when the user says "what version is this pom", "list the dependencies", "check the target framework", "find Spring beans" — use yq -p xml instead of reading. Essential for large poms (50+ dependencies) or batch operations across multiple XML files. Saves 90-99% of tokens with surgical extraction.

xq

⚡ IMPORTANT: Auto-trigger for XML operations. Use when: extract Maven dependencies, check POM version, list artifacts, find Spring beans, audit XML configs. Surgical extraction 90-99% token savings. Essential for pom.xml, .csproj, app.config, web.xml files.

schedule Updated 21 days ago

gps-word

O(1) exact word lookup in the inverted index. Find all files and line numbers where a specific identifier is defined. Faster than search for exact symbol names. Use this skill for precise identifier lookups.

schedule Updated 2 months ago

yq

⚡ IMPORTANT: Auto-trigger for YAML/TOML/XML. Use when: extract YAML fields, check catalog metadata, list pipeline config, audit docker-compose, extract Kubernetes manifests. Supports YAML/TOML/XML. Batch operations save 95-98% tokens. 90-99% savings vs reading full files.

schedule Updated 21 days ago

yq

Use yq to read, extract, filter, or transform YAML, TOML, and XML data from the command line instead of reading entire files. Load this skill for YAML files (Backstage catalogs, docker-compose.yml, GitHub Actions workflows, Azure Pipelines, Kubernetes manifests, mkdocs.yml, .gitlab-ci.yml, agent frontmatter), TOML files (Cargo.toml, pyproject.toml), or XML files (pom.xml, .csproj, web.xml). Use when extracting entity metadata, auditing catalog files, checking pipeline configuration, listing service definitions, batch processing multiple YAML files, or converting between formats. This skill applies even when the user says "check the catalog entity", "what's the owner of this component", "list all systems", "audit these pipelines", "extract navigation from mkdocs" — use yq instead of reading files. Essential for batch operations: extracting the same field from 100+ catalog files saves 95-98% of tokens. Same jq-like filter syntax works across YAML/TOML/XML.

structural-search

Synergy skill combining fd (file finder) with rg (ripgrep) for two-dimensional codebase search: structural search by name/path/extension AND textual search by content/pattern. Load this skill for codebase exploration ("what's in this project?"), dependency discovery (find imports and their locations), architecture analysis (map component relationships), security scanning (find secrets in specific file types), configuration audits (locate settings files then search their content), catalog investigations (find unmapped components = files named az-project.yaml WITHOUT spec.system), multi-repo navigation, or any search requiring both "where are the files?" AND "what's inside them?". Use even when the user says "explore this codebase", "find all references to X", "security scan", "what uses this dependency", "map the architecture", "locate configuration for Y" — these all need bi-dimensional search. fd answers structural questions (file tree, extensions, paths), rg answers content questions (text patterns, code us

batch-config-audit

Synergy skill combining fd (file finder) with yq/jq/xq (config parsers) to audit configuration files in batch operations. Load this skill when extracting the same field from many config files at once: auditing Backstage catalog entities (check 749 systems for annotations, verify component owners, find unmapped components), scanning Maven projects (extract versions from all pom.xml), validating CI/CD pipelines (check cron schedules across repos), generating compliance reports, creating CSV exports of catalog metadata, batch-checking JSON/YAML/XML configs for security issues, or producing dashboards from distributed configuration. Use even when the user says "audit the catalog", "check all components", "list all system names", "export entity metadata", "validate pipeline configs", "generate a report of owners" — this is the batch audit pattern. This is the highest-impact synergy: transforms O(n × file_size) token cost into O(n × field_size), typically 95-99% reduction. Essential for Backstage catalog operations