The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions and reference files contain scripts that read local repository files to gather debugging data. While this involves data access, it is constrained to the workspace and is necessary for the skill's function.
Evidence: references/locate.md reads source files (.dart, .py, .ts) to analyze error handling. references/slow.md reads requirements.txt and package.json to identify installed profiling tools.
[COMMAND_EXECUTION]: The skill uses standard local commands to analyze code patterns and repository history.
Evidence: references/regression.md executes git log and git rev-list to identify changes between releases. references/locate.md uses grep to identify how stack traces are handled in the codebase.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted content from the local project environment.
Ingestion points: The skill reads source code files and configuration manifests (e.g., SKILL.md, references/locate.md, references/reproduce.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat file content as untrusted or to ignore embedded instructions.
Capability inventory: Across its scripts, the skill has the capability to read files and execute git-related commands.
Sanitization: The skill does not implement sanitization or validation for the code or log data it retrieves before the agent processes it.

debug