fstar-verification

Pass

Audited by Gen Agent Trust Hub on Oct 2, 2026

Risk Level: SAFEPERSISTENCEPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation in README.md includes links to download binary releases of the F* compiler from the official project repository on GitHub.
  • [PRIVILEGE_ESCALATION]: The installation instructions in README.md describe using sudo to install the opam package manager and the z3 SMT solver via system package management tools.
  • [PERSISTENCE]: README.md contains instructions for the user to append environment variable exports (FSTAR_HOME and PATH) to their ~/.bashrc file to ensure the tool remains available across shell sessions.
  • [COMMAND_EXECUTION]: The skill details the execution of various development and verification tools including fstar, z3, make, and ocamlopt as part of its core functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and verify external source code files (.fst), which constitutes a standard attack surface for indirect prompt injection if the files contain instructions targeting the AI agent.
  • Ingestion points: Processes F* source files (.fst), build configurations (.fstar), and Makefiles.
  • Boundary markers: No specific delimiters or boundary warnings are mentioned to separate external code from agent instructions.
  • Capability inventory: Invokes local compilers, solvers, and build tools via shell commands.
  • Sanitization: The skill does not describe specific sanitization or validation of input code beyond the language's own type-checking and verification rules.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 2, 2026, 08:04 PM
Security Audit — agent-trust-hub — fstar-verification