The Agent Skills Directory

[COMMAND_EXECUTION]: The skill explicitly instructs the agent to run shell commands found in project documentation to verify their accuracy. In the "Audit" step, it requires the agent to "Dry-run install commands against package manifests and CI." In the "Verify" step, it states to "Prefer running documented commands when the environment allows."- [REMOTE_CODE_EXECUTION]: By instructing the agent to execute commands derived from untrusted input files (such as READMEs or docs), the skill introduces a direct path for remote code execution. A malicious repository could contain harmful commands in its documentation that would be executed by the agent during the verification phase.- [PROMPT_INJECTION]: The skill is highly vulnerable to indirect prompt injection because its core function involves processing untrusted external documentation that may contain adversarial instructions.
Ingestion points: The agent reads README, docs/, API docs, CHANGELOG, and CONTRIBUTING files as specified in the "Scope" section of SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore instructions found within the processed documentation.
Capability inventory: The skill grants the agent the capability to execute shell commands and read various files across the repository.
Sanitization: There are no instructions to validate or sanitize the commands found in the documentation before they are executed.

unslop-docs