The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git rev-parse and git diff to identify files that have changed since a specific commit hash. These are local, read-only operations that are necessary for the tool's documented purpose.
[PROMPT_INJECTION]: The skill processes data from .decodie/index.json and uses the verified_sha field as an argument in a git shell command. This represents an attack surface for indirect prompt injection if the index file is manipulated by an untrusted source. 1. Ingestion points: .decodie/index.json (SKILL.md). 2. Boundary markers: None present. 3. Capability inventory: Shell command execution via git (SKILL.md). 4. Sanitization: No explicit validation or escaping of the SHA input is mentioned. Mitigation: Ensure that values read from the index file conform to expected hexadecimal or git-ref formats before interpolation into shell commands.

decodie-flag-stale