The Agent Skills Directory

[SAFE]: No security concerns were identified during the analysis of this skill. It operates entirely on local project files and does not utilize network access or perform any operations that could lead to data exfiltration.
[INDIRECT_PROMPT_INJECTION]: While the skill scans and reads content from various project directories (e.g., src/, design/), it follows a collaborative protocol that requires explicit user approval before generating the final report. This human-in-the-loop requirement mitigates risks associated with processing potentially untrusted data from the codebase.
[COMMAND_EXECUTION]: The skill utilizes Bash for routine tasks such as counting files and estimating lines of code. This usage is restricted to local project auditing and is proportional to the skill's stated purpose.
[DATA_EXPOSURE_EXFILTRATION]: The skill's metadata (allowed-tools) does not include network-enabled tools like curl or wget, and the instructions do not attempt to access sensitive system files outside of the project scope.

project-stage-detect