tsa-find

name: tsa-find version: 2.0.0 description: | Fast file + content search with code-aware sizing. Replaces Read/Grep/find for routine "where is this file" / "grep for X" / "show me lines 10-20 of Y" questions. Returns file paths + line numbers + a sized chunk, not the whole file.

Use when:

• "Find files matching " / "show me all *.yml under config/"
• "Grep for 'TODO' / 'FIXME' / 'TODO\(perf\)' / regex anywhere"
• "How big is " / "is this file too large to read fully"
• "Show me lines 50-80 of " / "read just the relevant slice"
• "Find all files matching name + containing string"

Replaces: native find + grep + cat invocations (~3-10k tokens for big repos) with single MCP calls (200-500 tokens). allowed-tools: - mcp__tree-sitter-analyzer__search - mcp__tree-sitter-analyzer__project - mcp__tree-sitter-analyzer__structure - mcp__tree-sitter-analyzer__health - Bash - Read

tsa-find — File / text search, sized for agents

The "I just want to find X" skill. Wraps fd + rg + sized partial reads with consistent output formatting.

Tool routing

Procedure

Single search

search action=content query="TODO" roots=["tree_sitter_analyzer/"] include_globs=["*.py"]

Returns: matches: [{file, line, content}] with sized previews. Always includes file:line so the agent can cite without reading the file.

Sized partial read (the killer feature)

Before reading a large file blind, use:

health action=scale file_path="tree_sitter_analyzer/ast_cache.py"
# returns {line_count: 938, file_metrics: {file_size_bytes: 38918, total_lines: 938, code_lines: 661, ...},
#          llm_guidance: {analysis_strategy: "This is a large file...", recommended_tools: ["structure action=read", ...]}}
# (there is no top-level is_large / recommendation — judge size from line_count / file_metrics.file_size_bytes,
#  and read llm_guidance.recommended_tools for the next-step suggestion)

Then extract only what you need:

structure action=read file_path="..." start_line=800 end_line=870

Avoids reading 80KB when you need 2KB.

Combined find+grep

search action=grep file_pattern="test_*.py" content_pattern="def test_synapse"
# returns: list of test functions matching both criteria

CLI equivalents

uv run tree-sitter-analyzer --project-root . --outline   # list project files
uv run tree-sitter-analyzer --check-tools                # verify fd/rg available
uv run tree-sitter-analyzer <file> --partial-read        # sized read with --start-line / --end-line

Anti-patterns

• DON'T Read a large file before checking scale — burns tokens
• DON'T Bash grep -rn for things search action=content handles — slower + noisier
• DON'T call search action=batch with a single query — it enforces a ≥2-query minimum (raises "must be at least 2 queries"); use search action=content for one pattern
• DON'T re-search the same query twice in one session — cache the result mentally