The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains a command pattern that fetches data from an external API and pipes it into a Python interpreter for parsing.
Evidence: A curl command in SKILL.md targets completion.amazon.com and pipes the output to python3 -c "import sys,json; [print(s['value']) for s in json.load(sys.stdin).get('suggestions',[])]".
Analysis: While piping network data to an interpreter is a high-risk signature, the source is a well-known service (Amazon) and the executed code is a hardcoded, benign JSON parser used for keyword extraction.
[COMMAND_EXECUTION]: The skill utilizes a local shell script to automate the retrieval of competitor product data.
Evidence: The script scripts/fetch-competitor.sh uses curl, grep, and sed to extract marketing data (titles, bullets, BSR) from Amazon marketplace domains.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch real-time search suggestions and competitor product information.
Evidence: Network calls to completion.amazon.com and various amazon.com marketplace domains are used to populate campaign data.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from external Amazon listings, which could be leveraged for indirect prompt injection.
Ingestion points: The scripts/fetch-competitor.sh script extracts title and bullet point text from competitor pages in SKILL.md Step A3.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present when processing the scraped data.
Capability inventory: The skill has the ability to execute shell commands and fetch external data, though the scraped data itself is only used as text input for campaign design.
Sanitization: There is no evidence of filtering or sanitizing the scraped HTML content before it is presented to the agent context.

amazon-ppc-campaign