The Agent Skills Directory

[PROMPT_INJECTION]: A prompt injection pattern was detected in the references/evals.yaml file. Analysis confirms this is part of a benchmark test case designed to evaluate the agent's ability to resist malicious instructions and does not pose a threat to the skill's operational instructions.
[EXTERNAL_DOWNLOADS]: The skill downloads content and metadata from GitHub. These operations are restricted to a curated allowlist by default and utilize mandatory commit-SHA pinning and signature verification to ensure the provenance and integrity of the downloaded code.
[COMMAND_EXECUTION]: The skill uses subprocess.run to execute git commands and local validation scripts. These calls use sanitized inputs and are essential for the skill's primary function of managing local skill installations.
[DATA_EXFILTRATION]: The skill utilizes GitHub authentication tokens (e.g., GITHUB_TOKEN) for API access. The instructions explicitly mandate that the agent redact these secrets from any output to prevent accidental credential exposure.

skill-installer