The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill processes untrusted data from external logs which could contain malicious instructions or malformed code designed to exploit the agent or local compiler.
Ingestion points: GitHub Action logs fetched via gh api and stored in fuzzer_logs.txt (File: SKILL.md).
Boundary markers: Absent. There are no delimiters or instructions to treat the log content as non-executable text.
Capability inventory: The skill uses the Bash tool to run nargo and the Write tool to create src/main.nr and Prover.toml (File: SKILL.md).
Sanitization: Absent. The agent is directed to copy sections directly from logs into the local environment.
[Command Execution] (LOW): The skill invokes nargo execute on code extracted from fuzzer logs. This behavior is the primary intended purpose of the skill for developers, though it involves running unverified code.
[External Downloads] (LOW): Performs network operations via gh api to fetch logs from the noir-lang/noir repository. Since the organization is not on the trusted list, this is flagged as a low-severity network operation.

extract-fuzzer-repro