The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external sources such as application logs, version control history, and error messages to diagnose bugs. This creates a surface where an attacker who can write to these logs could embed adversarial instructions aimed at the agent.
Ingestion points: Reads files and output via tail -f logs/error.log, git log, and grep commands in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore embedded commands within the log data it processes.
Capability inventory: The skill has access to powerful tools including Bash, Write, and Edit, which could be abused if an injection is successful.
Sanitization: There is no logic provided to sanitize or filter the content of ingested logs before they are processed by the agent.
[COMMAND_EXECUTION]: The skill provides instructions to execute several shell commands and a local Python script to gather context and generate reports.
Evidence: env | grep -i debug in SKILL.md
while targeted at debugging, this command can expose environment variables that might contain sensitive configuration or credentials if they include the string 'debug'.
Evidence: python3 scripts/debug_report.py --name <error-message> --output debug-report.md in SKILL.md and README.md
executes a local Python script included with the skill to generate markdown templates.

debugger