easy-yapi-assistant
Pass
Audited by Gen Agent Trust Hub on Aug 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses and reads configuration files from the user's home directory at
~/.easyapi/. These files contain global API export rules and metadata that apply across all projects on the system. - [COMMAND_EXECUTION]: The skill relies on bundled shell scripts (e.g.,
read_rule_file.sh,get_existing_rules_for_key.sh,list_rule_files.sh) to perform file system operations and keyword searches within the project workspace. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data including Java/Kotlin source code and user-defined rule files (
.rules,.properties). This content is used to generate suggestions for API documentation rules, which could be manipulated via malicious comments or code patterns in the analyzed project. - Ingestion points: Project source files (via
rg),.easyapi/*.rules, and legacy.easy.api.config*files. - Boundary markers: The skill instructions rely on the agent showing diffs to the user rather than technical delimiters in the prompts.
- Capability inventory: The agent can propose file writes and perform search operations across the project structure.
- Sanitization: The skill requires explicit human approval for every file modification and does not execute the generated Groovy scripts itself.
Audit Metadata