The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes an external Python script located at ~/.claude/scripts/doc-validate.py to perform documentation validation. Since the contents of this script are not provided within the skill package, its logic and behavior cannot be verified, which presents a security risk.
[COMMAND_EXECUTION]: The skill makes extensive use of system commands and third-party CLI tools, including ls, grep, wc, diff, oc (OpenShift), gh (GitHub CLI), and bd (Beads). These tools are used for repository analysis, checking cluster deployment status, and managing project issues.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes arbitrary source code from repositories to generate documentation.
Ingestion points: The skill reads all files within the target repository during project detection, coverage calculation, and documentation generation steps (e.g., reading .py and .go files in SKILL.md).
Boundary markers: No explicit delimiters or instructions are used to ensure the agent ignores or isolates potentially malicious instructions embedded in comments or strings within the processed source files.
Capability inventory: The agent has permissions to execute shell commands, read and write files, and interact with external services via the gh and oc tools.
Sanitization: No sanitization or content validation is performed on the ingested repository data before it is analyzed by the agent.

doc