The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8). It ingests untrusted data from arbitrary external URLs provided by users and processes this content through the agent.
Ingestion points: Data enters the agent context via the extraction API response in SKILL.md (Step 6).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying instructions embedded within the extracted web content.
Capability inventory: The skill has the ability to execute shell commands (curl) and write files to the local filesystem (echo >).
Sanitization: No sanitization or filtering of the extracted content is performed before it is presented to the agent or saved to disk.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several bash commands for configuration and operation.
Evidence: Uses mkdir and echo to create local configuration files in .listenhub/content-parser/ and curl combined with jq in a loop to poll the status of extraction tasks.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to external endpoints.
Evidence: It makes POST and GET requests to https://api.marswave.ai/openapi/v1/content/extract/ using curl. As this domain belongs to the skill's author ('marswaveai'), it is considered a vendor-owned resource.

content-parser