The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to facilitate the debugging process by running test frameworks, executing CLI commands, and performing HTTP requests via curl. It also dynamically generates and executes small scripts (e.g., in Node.js or Python) to isolate and reproduce reported bugs. These actions are consistent with the skill's primary purpose.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its analysis of external, potentially attacker-controlled data.
Ingestion points: Symptom descriptions provided in arguments, stack trace contents from application errors, and project configuration or dependency files (.fullstack-dev/config.json, package.json, requirements.txt).
Boundary markers: The prompt instructions lack explicit delimiters or specific 'ignore' instructions for content found within ingested evidence files.
Capability inventory: The skill has powerful system access via Bash (command execution), Edit and Write (file modification), and the Agent tool (sub-agent dispatch).
Sanitization: There is no evidence of automated sanitization or filtering of error logs and stack traces before they are presented to the agent for analysis.

debug