The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data including source code, requirements, and test coverage reports (LCOV, JSON, XML).
Ingestion Points: coverage_analyzer.py parses external report content; test_generator.py processes raw source code and requirement strings.
Boundary Markers: The instructions in SKILL.md and HOW_TO_USE.md do not explicitly require the agent to use strict delimiters (like XML tags) when interpolating untrusted external code or report data into the prompt context, which could allow maliciously crafted code comments or coverage metadata to influence agent behavior.
Capability Inventory: The skill involves the agent executing several local Python scripts via the CLI (test_generator.py, coverage_analyzer.py, tdd_workflow.py, etc.) to generate output and analyze project state.
Sanitization: While the Python scripts use structured parsing (e.g., json.loads, xml.etree.ElementTree.fromstring), there is no explicit sanitization logic to detect or neutralize prompt injection attempts hidden within the analyzed source code or data files.
[EXTERNAL_DOWNLOADS]: The documentation (README.md, references/ci-integration.md) mentions dependencies on well-known testing frameworks and services such as pytest, jest, vitest, and Codecov. These references are intended for user environment setup and CI integration, and are targeting well-known, trusted industry services.
[COMMAND_EXECUTION]: The skill instructs the agent to run various local Python scripts to perform its duties. These scripts (scripts/*.py) are part of the skill package and perform legitimate analysis tasks such as calculating cyclomatic complexity and parsing LCOV files.

tdd-guide