The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides installation instructions that involve downloading and immediately executing remote scripts from an unverified external domain. This bypasses security reviews and can lead to arbitrary code execution.
Evidence in SKILL.md: curl -fsSL https://fabro.sh/install.sh | bash and curl -fsSL https://fabro.sh/install.md | claude.
Further evidence: codex "$(curl -fsSL https://fabro.sh/install.md)" uses command substitution to execute fetched content.
[INDIRECT_PROMPT_INJECTION]: The skill defines agent pipelines that read from external specification files and interpolate that content directly into prompts, which is a known attack surface for indirect prompt injection.
Ingestion points: The workflow examples in SKILL.md read natural language specifications from the specs/ directory (e.g., specs/{feature_name}.md).
Boundary markers: The skill does not demonstrate or require the use of delimiters or 'ignore embedded instructions' warnings for this interpolated content.
Capability inventory: The skill includes capabilities for CLI command execution, SSH access to sandboxes (fabro ssh), and port forwarding (fabro preview).
Sanitization: There is no evidence of sanitization, filtering, or validation of the content read from these files before it is used in the prompt field of the DOT graph.
[COMMAND_EXECUTION]: The skill relies on and instructs the user to execute various CLI tools that manage system resources and remote access.
Commands like fabro ssh and fabro preview provide high-privilege access to isolated environments and network ports.
[EXTERNAL_DOWNLOADS]: The skill performs multiple network requests to fetch installation scripts and documentation from fabro.sh, a domain not identified as a trusted or well-known service.

fabro-workflow-factory