crafting-effective-readmes
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data when reviewing existing project files (such as
package.json, source code, and existing README files) to generate or update documentation. \n - Ingestion points: Context gathering step in
SKILL.mdandREADME.mdwhere the agent is instructed to "Check against actual project state".\n - Boundary markers: The skill does not define explicit delimiters or warnings to ignore instructions that might be embedded in the project files being documented.\n
- Capability inventory: The skill is designed for documentation tasks; it does not contain instructions for subprocess execution, network exfiltration, or privileged file system operations.\n
- Sanitization: No specific sanitization or filtering of the ingested code content is implemented.\n- [SAFE]: All external documentation references and links target well-known services and established open-source repositories, such as GitHub and standard-readme.\n- [SAFE]: The use of placeholders for sensitive environment variables in templates (e.g.,
API_KEYintemplates/internal.md) follows standard documentation practices and does not include hardcoded credentials.
Audit Metadata