pp-surgegraph
Warn
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: Downloads and installs the surgegraph-pp-cli binary using npx from @mvanhorn/printing-press-library or go install from the vendor's GitHub repository.
- [DATA_EXFILTRATION]: The skill documents a --deliver webhook: feature that allows the agent to POST command output, which may include sensitive visibility data or managed credentials, to arbitrary external URLs.
- [CREDENTIALS_UNSAFE]: The CLI provides functionality to manage and retrieve LLM API keys (OpenAI, Gemini, Anthropic) and WordPress integration keys, which exposes these secrets to the agent context.
- [COMMAND_EXECUTION]: Executes multiple subcommands via the surgegraph-pp-cli binary, including sync operations, document management, and account status checks.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Processes untrusted data from external sources including AI answer engine citations, brand visibility reports, and WordPress site content.
- Boundary markers: Absent; command output is interpolated directly into the agent context without sanitization warnings.
- Capability inventory: Includes shell command execution via the CLI tool, file writing via the --deliver file: flag, and network POST requests via the webhook sink.
- Sanitization: Not explicitly implemented in the skill instructions or CLI parameters.
Audit Metadata