The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill requires administrative privileges to perform system-level updates and installation of dependencies into protected directories.
Evidence: Execution of sudo apt-get update and sudo apt-get install -y build-essential wget to prepare the build environment.
Evidence: Use of sudo make install to place compiled binaries into the /usr directory.
[REMOTE_CODE_EXECUTION]: The skill downloads and executes code from non-official sources and insecure channels.
Evidence: Downloads source code (ta-lib-0.4.0-src.tar.gz) via an unencrypted http:// connection from SourceForge, followed by local compilation and execution of build scripts (./configure, make).
Evidence: Recommends downloading and installing pre-compiled binary wheel files from a non-official, individual-maintained GitHub repository (github.com/cgohlke/talib-build).
[COMMAND_EXECUTION]: The skill incorporates a user-supplied argument ($PYTHON_VERSION) directly into a shell execution string without evident sanitization.
Evidence: The command $PYTHON_VERSION -m venv venv is susceptible to command injection if the user provides a manipulated string containing shell metacharacters.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by ingesting untrusted data that is later used in privileged tool operations.
Ingestion points: The skill uses AskUserQuestion to collect API keys (OpenAlgo, Crypto), secret keys, and local database file paths.
Boundary markers: No explicit delimiters or instructions are used to prevent the agent from interpreting instructions embedded within the user-provided data.
Capability inventory: The skill possesses Bash and Write capabilities, allowing it to execute commands and modify files based on ingested data.
Sanitization: No evidence of validation or sanitization is present for the user-provided keys or file paths before they are written to the .env configuration file.

setup