The Agent Skills Directory

[PROMPT_INJECTION]: The skill does not contain instructions that attempt to override agent safety protocols or bypass system constraints.
[DATA_EXFILTRATION]: The skill's operations are confined to the project's documentation directory (docs/content/docs/reference/roadmap/). There is no evidence of sensitive data being accessed or transmitted to external domains.
[COMMAND_EXECUTION]: The skill specifies a Commit + push process for updating documentation. These actions are performed within the context of the project's version control and are aligned with the skill's primary purpose of maintaining roadmap items.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by reading existing roadmap .mdx files and external research links. While this creates a potential surface for indirect prompt injection if those files contain malicious instructions, the skill uses these as reference material for design decisions and lacks high-risk execution capabilities that would make such an injection critical.
Ingestion points: Reads roadmap files in docs/content/docs/reference/roadmap/ and linked research dossiers.
Boundary markers: None explicitly defined in the prompt interpolation.
Capability inventory: File system read/write, git commit, and git push.
Sanitization: None specified for the content read from external roadmap files.

brainstorm