The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a 'wake' command on the user's host system to ensure the Chrome extension remains active, subject to user consent.
[COMMAND_EXECUTION]: It requires the use of the 'curl' utility to perform HTTP PUT operations to upload media bytes to dynamically generated URLs provided by integrated tools.
[DATA_EXFILTRATION]: The skill reads from a local configuration file located at '~/.uplika//voice.md' to retrieve writing style preferences and tone settings.
[DATA_EXFILTRATION]: Network operations are performed to communicate with vendor-managed tools and to upload media content to external endpoints.
[EXTERNAL_DOWNLOADS]: The 'media_from_url' tool allows the skill to retrieve and attach photos or videos from arbitrary public HTTPS addresses provided by the user or external content.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, including keyword reports and Naver blog search results, which could contain adversarial instructions designed to influence the post generation.
[INDIRECT_PROMPT_INJECTION]: Evidence Chain: 1. Ingestion points: Data is retrieved from external sources via 'list_naver_keyword_reports' and by browsing live Naver search results. 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the ingested content. 3. Capability inventory: The skill has the capability to execute shell commands ('wake', 'curl') and publish content to external accounts. 4. Sanitization: No specific methods for validating or sanitizing the retrieved external content are described in the instructions.

naver-post