note-mabza
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a feedback loop where user-provided content is captured and stored in the
.agents/feedback/directory to be processed later by/improve-mabza. - Ingestion points: User feedback and discussion content are captured in
SKILL.md(Step 2: Écrire l'entrée) and stored in Markdown files. - Boundary markers: The skill uses Markdown sections (
## Observation,## Contexte) and YAML frontmatter but lacks specific "ignore instructions" delimiters for the content provided by the user. - Capability inventory: The skill performs file system writes (
mkdirfor the directory andwritefor the Markdown files). - Sanitization: There is no evidence of sanitization or escaping for the user content before it is written to the feedback files, which could lead to command or prompt injection if the processing tool (
/improve-mabza) executes or interprets the content of these logs.
Audit Metadata