secondme-dev-assistant
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECREDENTIALS_UNSAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill stores sensitive OAuth Client Secret and developer access tokens in local files at ~/.secondme/client_secret and ~/.secondme/dev_credentials. While it attempts to set restrictive file permissions, storing secrets in plain text poses a risk.
- [DATA_EXFILTRATION]: An opt-in telemetry system records session metadata, including user_intent and actions_summary, and synchronizes this data to the vendor's remote endpoint at https://api.mindverse.com.
- [DYNAMIC_EXECUTION]: The skill uses python3 to execute multi-line scripts embedded within the instruction files for handling telemetry synchronization and JSON configuration merging.
- [COMMAND_EXECUTION]: The skill executes shell commands using npx for installation and updates, and uses openssl for generating PKCE security parameters.
- [INDIRECT_PROMPT_INJECTION]: The agent is instructed to fetch and follow API specifications from external documentation URLs (https://develop-docs.second-me.cn), which represents an ingestion surface for potentially untrusted data.
- [EXTERNAL_DOWNLOADS]: The skill fetches documentation and updates from remote sources aligned with the vendor infrastructure.
Audit Metadata