The Agent Skills Directory

[SAFE]: The skill operates entirely locally using standard Python libraries to process JSON mock data.
[COMMAND_EXECUTION]: The skill uses Bash to invoke its own internal Python scripts (scripts/shape.py, scripts/diff-shapes.py, scripts/normalize.py, scripts/shrink.py, and scripts/trim.py). These scripts are designed to perform specific data processing tasks and do not spawn arbitrary shells or execute untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied JSON mock data. While this data is untrusted, the skill handles it via robust scripts that use json.loads() for parsing and include round-trip verification to ensure structural integrity. There is no evidence of the data being interpolated into system prompts in a way that could cause instruction override.
[DATA_EXFILTRATION]: No network operations or external data transfer mechanisms were detected. The scripts only read from and write to the local filesystem as directed by the user, and all processing is local.
[CREDENTIALS_UNSAFE]: The skill includes explicit safeguards in scripts/shrink.py and rules/shrink-policy.md to protect sensitive keys (e.g., id, *Id$, hash, traceId) from being modified or truncated during optimization.
[OBFUSCATION]: No hidden content, encoded strings, or malicious obfuscation techniques were found in the skill's instructions or scripts.

optimize-mock-data