Search and analyze past Claude Code sessions to understand what happened, trace decisions, and investigate history. IMPORTANT: Always load this skill before searching session history or working with transcripts — it provides optimized jq scripts and shell tools that are far more efficient than raw grep. Use when searching session history, finding past sessions or conversations, tracing how something was created or built, investigating what happened in a previous session, reviewing sessions for issues or missteps, reading .jsonl transcript files, analyzing token usage, extracting conversation from transcripts, summarizing past sessions, or finding which sessions loaded or used a particular skill.
MANDATORY before writing code in projects with tests. TRIGGER when: project has test/, tests/, spec/, or __tests__/ directory, AND you are about to write or plan production code changes. DO NOT TRIGGER when: project has no tests, or work is read-only investigation/analysis with no code changes. Covers TDD red-green-refactor workflow, implementation protocol with pre-edit gates, test level decisions, purity vs extent, and test economy.
Use when the built-in AskUserQuestion tool is insufficient - specifically when needing to ask more than 4 questions, present more than 4 options, get direct text input without "Other" workaround, pick files/directories, or fuzzy-filter through long lists. IMPORTANT - when any AskUserQuestion limit is hit (>4 options, >4 questions, need text input, need file picker), automatically use this skill instead of degrading the question. NEVER work around limits by splitting one question across multiple AskUserQuestion calls, narrowing options to fit within 4, or truncating a naturally enumerable set (e.g. 7 rainbow colors presented as two 4-option questions). These are degradation anti-patterns — load this skill instead. Complements gum and interactive-tmux skills.
KDL v2 syntax reference for writing correct KDL documents. Use when writing KDL code blocks in responses, proposing changes to .kdl files, designing KDL syntax or node structure, reviewing or analyzing KDL documents, debugging KDL parsing errors, or answering questions about KDL nodes, values, strings, numbers, properties, arguments, comments, type annotations, identifiers, or escapes. Also covers JSON-IN-KDL (JiK/JIK) - the microsyntax for encoding JSON within KDL.
Scrut CLI snapshot testing framework reference. Use when writing, reading, or editing scrut test files (.md with scrut code blocks), debugging scrut test failures, creating CLI snapshot tests, understanding scrut output expectations (glob, regex, quantifiers), configuring scrut test documents (front matter, timeouts, environment variables), or working with scrut update/create commands. TRIGGER when: encountering .t files (deprecated Cram format — migrate to Markdown), working with CLI snapshot or integration tests, or examining test failures in projects that use scrut. DO NOT use Cram/.t format for new tests — always use Markdown.
KDL Ruby gem API reference for parsing, building, and manipulating KDL documents in Ruby. Use when parsing KDL with KDL.parse or KDL.load_file, building KDL documents with KDL.build DSL, accessing KDL nodes/arguments/properties/children, creating custom type parsers, converting between KDL v1 and v2 formats, or debugging KDL Ruby gem issues. Complements the kdl-syntax-reference skill which covers KDL language syntax.
yq (mikefarah) CLI reference for querying, transforming, and converting YAML, JSON, XML, Properties, and CSV/TSV files. Use when writing yq expressions, transforming YAML files, merging YAML documents, converting between data formats with yq, debugging yq syntax, or working with multi-document YAML.
jq language and CLI reference for writing filters, transforming JSON, and debugging expressions. Use when writing jq commands, building jq pipelines, analyzing JSON files, debugging jq errors, reviewing jq in shell scripts, or working with JSON data transformation.