The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Cargo.lock files through an included Python auditing script. This creates a vulnerability surface where a malicious lockfile could attempt to inject instructions into the agent's context via unsanitized package names or version strings.
Ingestion points: scripts/check-uniffi-lock.py reads user-provided Cargo lockfiles.
Boundary markers: There are no specific delimiters or ignore-instructions defined for handling the output of the dependency audit.
Capability inventory: The skill utilizes shell commands through Gradle, Cargo, and Python, which provide significant system access if an injection were to occur.
Sanitization: The audit script extracts and displays TOML values directly without sanitization or escaping.
[REMOTE_CODE_EXECUTION]: The skill involves the download and installation of binary tools from external sources.
Fetches and installs the UniFFI binding generator from the Ubique Innovation GitHub repository using cargo install --git.
Resolves official plugin and runtime dependencies from Maven Central during the Gradle build process.
[COMMAND_EXECUTION]: The skill requires the use of CLI tools to build the project and verify the interop boundary.
Executes ./gradlew for binding generation, project compilation, and running smoke tests.
Executes cargo for Rust library builds and toolchain management.
Runs a local Python script to verify UniFFI version alignment across workspaces.

ubique-uniffi