day4-wrap-and-analyze
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
npx skills add ai-native-camp/camp-1 --agent claude-code --yesduring the initial setup. This command installs the latest curriculum from the vendor's official repository.\n- [EXTERNAL_DOWNLOADS]: The skill fetches external content from theai-native-camprepository vianpx. This is a standard procedure for updating the course content.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes content from several reference markdown files and user responses to manage the training flow.\n - Ingestion points: Files located in the
references/directory (e.g.,block1-build-session-wrap.md).\n - Boundary markers: Uses structural headers like
## EXPLAINand## EXECUTEto distinguish between interaction phases.\n - Capability inventory: Spawns shell processes for installation and performs file system operations.\n
- Sanitization: Relies on a structural state machine to control data flow.\n- [PROMPT_INJECTION]: The main
SKILL.mddefines a 'STOP PROTOCOL' that uses imperative language to dictate interaction pacing. These directives are structural interaction guidelines designed to manage information flow rather than malicious attempts to bypass safety filters.
Audit Metadata