Stats
Actions
Tags
From tracegrep
Use when searching supported source code, tracing references or callers, or when you would normally use `rg`, `grep`, a built-in code search tool, or any other plain-text search in a repository that has `tg` available. Default to `tg` first for supported source files and only fall back when `tg` cannot model the target.
How this skill is triggered — by the user, by Claude, or both
Slash command
/tracegrep:tracegrepThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
For supported source files in this repository, use `tg` first instead of any built-in search tool or plain-text search command.
For supported source files in this repository, use tg first instead of any built-in search tool or plain-text search command.
Do not start with rg, grep, or a built-in code search tool for supported code unless tg fails, cannot model the target language construct, or you are intentionally searching non-source text that tg does not cover.
tg accepts the same search parameters you would normally pass to rg, but it returns matches enriched with language-aware call-graph context.
When installed from this repo, tg is the short binary name for tracegrep.
Supported file types: .rs, .py, .js, .jsx, .svelte, .ts, and .tsx.
tg, not a built-in search tool.rg.--json when you need structured output for further processing.rg, grep, or a built-in search tool only for unsupported file types, non-code text, or cases where tg clearly cannot answer the question.tg decorates matches with the backtrace of how that line of code is called in the codebase. Use this to build understanding of existing call paths & avoid reinvention.
tg tool_data
tg QueryOptions src
tg --json --include-tests --depth 2 parse
npx claudepluginhub btucker/tracegrep --plugin tracegrepPerforms semantic searches on codebases using natural language queries to locate orchestration logic, definitions, and call flows, with snippets, scores, and trace/skeleton options.
Replaces grep/rg/ag/ack/fd with AST-aware code search via tilth MCP. Finds symbols, definitions, callers, imports, and text patterns using tree-sitter. Requires tilth MCP server.
Searches codebases using ripgrep (rg) with regex patterns, file type/extension filtering, multi-line support, and .gitignore respect. Use for text patterns, code snippets, or multi-file analysis.