The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill's installation guide in references/installation.md instructs users to execute sudo pacman -S tamarin-prover for Arch Linux environments, which grants the toolchain administrative access to the system.
[COMMAND_EXECUTION]: The skill requires the execution of various shell commands to function, including the primary tamarin-prover executable, maude (logic engine), dot (Graphviz visualization), and a local Python script scripts/inspect-model.py.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for fetching and installing software from external sources, including a Homebrew tap (tamarin-prover/tap/tamarin-prover), the Nix package manager, and cloning the project's source code from a GitHub repository (https://github.com/tamarin-prover/tamarin-prover.git).
[INDIRECT_PROMPT_INJECTION]: The scripts/inspect-model.py script serves as a surface for indirect injection as it processes untrusted .spthy protocol files.
Ingestion points: The script reads the text content of every .spthy file located within the directory structure provided to the --root argument.
Boundary markers: There are no explicit delimiters or instructions within the processing logic to ignore potentially malicious embedded instructions in the protocol models.
Capability inventory: The skill possesses capabilities for system command execution, local file access, and hosting an interactive web-based proof environment (tamarin-prover interactive).
Sanitization: The script uses regular expressions to extract protocol metadata but does not sanitize the resulting strings before they are processed or presented to the agent context.

tamarin-prover