maa-pipeline-graph
Warn
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands, specifically ripgrep (rg), to recursively scan the workspace for files related to pipeline visualization. It also explicitly directs the agent to execute discovered Python scripts using the local interpreter (e.g.,
python tools/pipeline_to_mermaid.py). - [DYNAMIC_EXECUTION]: The skill employs a pattern of discovering and running local executable content at runtime. It instructs the agent to find scripts matching specific name patterns and run them to generate output, which could lead to the execution of malicious code if the project repository is compromised or contains untrusted scripts.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple ingestion points to build state machine representations, which could be exploited to influence the agent's behavior through malicious content in project documentation or configuration.
- Ingestion points: Project files including
basic_info.md,interface.json, various Pipeline JSON files, and Python source code inagent/**/*.py. - Boundary markers: No explicit delimiters or instruction-bypass warnings are used when reading data from these external files into the agent's context.
- Capability inventory: The skill has access to shell execution (
rg), local file system read/write, and Python script execution. - Sanitization: While it uses basic regex sanitization (
re.sub(r'[^A-Za-z0-9_]', '_', name)) to ensure technical validity of Mermaid IDs, it does not implement sanitization or validation to prevent prompt injection or data manipulation within the parsed content.
Audit Metadata