The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill demonstrates a high-risk surface for indirect prompt injection (Category 8) due to its core function of analyzing external text.
Ingestion points: The skill ingest untrusted data from user-specified files (e.g., existing_content.txt, blog_post.md) into the agent's context via brand_voice_analyzer.py and seo_optimizer.py.
Boundary markers: There are no delimited boundaries or system instructions to ignore embedded prompts within the files being analyzed.
Capability inventory: The skill possesses the capability to execute shell commands (python, grep, cp) and modify the filesystem.
Sanitization: No evidence of sanitization or filtering for malicious instructions within the input files is provided.
Command Execution (MEDIUM): The skill relies on the execution of subprocesses and shell commands to perform its duties. While the scripts are local, directing an agent to execute commands like grep -f references/brand_guidelines.md content.txt on potentially attacker-controlled content provides a vector for command manipulation or information leakage if the input files are crafted to exploit shell syntax.
Unverifiable Code (MEDIUM): The skill references multiple external scripts (scripts/brand_voice_analyzer.py, scripts/seo_optimizer.py) and reference files which are not present in the analyzed set. The security of the skill depends entirely on the contents of these unverified local scripts.

content-creator