The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the postplus command-line interface, a vendor-owned tool from the author 'postplusai', to perform publishing operations. It instructs the agent to run subcommands such as postplus publish <operation> with structured JSON inputs.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data (content, media, and integration IDs) to construct publishing envelopes, which presents a vulnerability surface for indirect instructions hidden in ingested content.
Ingestion points: Untrusted user content and media descriptions are used to build the publishing input object.
Boundary markers: The instructions do not define specific delimiters or warnings to separate user content from system instructions within the data payload.
Capability inventory: The skill has the ability to execute CLI commands and read/write local JSON files (input.json, result.json, challenge.json).
Sanitization: While the skill provides a mechanism to verify the request shape (postplus publish schema), it lacks specific logic for sanitizing or escaping the text content provided by the user.
[DATA_EXPOSURE]: The skill includes explicit security boundaries, instructing the agent to never ask for third-party credentials and to use the PostPlus platform's hosted invite-link flow for channel authorization. It also prohibits exposing login polling secrets to the user.

social-media-publisher