The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill repeatedly instructs the user to download and execute a script from a remote server using the curl | sh pattern.
Evidence: Found in SKILL.md, references/authentication.md, and references/cli-reference.md using the URL https://cli.inference.sh.
[COMMAND_EXECUTION]: The instructions encourage piping unverified remote content directly into the system shell, bypassing manual code review.
[DATA_EXFILTRATION]: The belt CLI tool, which the skill is designed to manage, possesses a "Local File Uploads" feature that automatically reads and transmits local file contents to a remote cloud environment when a path is provided in the input arguments.
Evidence: Documented in SKILL.md and references/running-apps.md under the "Local File Uploads" section.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves processing external input files and performing network operations, which creates a vulnerability surface for indirect prompt injection.
Ingestion points: Local input files (JSON/text) and media files passed via the --input argument in SKILL.md and references/running-apps.md.
Boundary markers: None identified; instructions do not suggest the use of delimiters or warnings to ignore instructions embedded in the processed data.
Capability inventory: The skill uses the belt CLI which supports network uploads, file reads, and interaction with various AI models and social media APIs.
Sanitization: There is no evidence of input validation or content sanitization before the data is processed by the CLI tool.

agent-tools