The Agent Skills Directory

[PERSISTENCE]: The skill instructs the agent to modify the user's shell profile (~/.bashrc) to append custom binary paths to the PATH environment variable. This ensures that the manually installed Node.js runtime remains available in future sessions.
[EXTERNAL_DOWNLOADS]: The skill downloads the Node.js binary directly from nodejs.org and installs the @iqinghu/qhkit package from the npm registry. It also utilizes external mirrors such as npmmirror.com for package and binary retrieval.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to bootstrap its environment, including downloading archives, verifying checksums, and extracting files. A security verification step is included: curl -fsSL ...SHASUMS256.txt | grep ... | sha256sum -c -, which validates the integrity of the Node.js package before use.
[DYNAMIC_EXECUTION]: The skill performs dynamic code execution to handle image optimization tasks. Specifically, it uses python -c to execute a one-liner script using the Pillow library and npx to run sharp-cli.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted user-supplied prompts and data from external tool outputs.
Ingestion points: Untrusted data enters the agent context through the prompt field in the generate and script commands in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: The skill possesses the ability to execute shell commands (qhkit, npm, python), write to the file system (~/.bashrc, image compression outputs), and perform network operations (curl).
Sanitization: No evidence of input sanitization or validation was found for the interpolated data.

linkpix-youtube-viral-video