The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions do not contain any patterns attempting to override agent behavior, bypass safety filters, or extract system prompts. The language is purely instructional and focused on the task of planning.
[DATA_EXFILTRATION]: There are no hardcoded credentials, sensitive file path accesses, or unauthorized network operations. The file operations are limited to saving documentation in a standard directory structure (docs/plans/).
[REMOTE_CODE_EXECUTION]: The skill does not involve downloading or executing external scripts. While it generates bash commands for verification, these are intended for execution within a controlled development environment as part of the implementation process.
[COMMAND_EXECUTION]: The skill includes placeholders for shell commands to verify tasks. These commands are generated based on the implementation context and are not executed automatically or without user oversight.
[INDIRECT_PROMPT_INJECTION]: The skill processes external inputs such as specifications or design documents to generate plans. While this represents a surface for indirect prompt injection, the risk is mitigated by the structured template and the requirement for explicit verification steps. The skill does not possess high-privilege capabilities that would be automatically triggered by malicious input in a spec file.
[OBFUSCATION]: No obfuscation techniques such as Base64 encoding, zero-width characters, or homoglyphs were detected in the skill content.

writing-plans