substack-draft-status
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The script
scripts/status.pyaccesses a sensitive local file atC:/Users/mphan/OneDrive/Documents/GitHub/mphinance/secrets.envto retrieve authentication credentials. This access is hardcoded to a path within the author's specific development environment. - [EXTERNAL_DOWNLOADS]: The skill performs a network request to
mphinance.substack.comvia the Substack API to retrieve draft metadata. This communication is required for the skill's primary function and targets a well-known service. - [COMMAND_EXECUTION]: The skill executes the local Python script
scripts/status.pyusing the system's Python interpreter to perform data reconciliation. - [PROMPT_INJECTION]: The skill ingests untrusted data from external sources, presenting an indirect prompt injection surface. 1. Ingestion points:
scripts/status.pyreads draft titles from the Substack API response and content from localpost.mdfiles. 2. Boundary markers: No specific delimiters are employed in the final report to isolate these external titles from the agent's context. 3. Capability inventory: The skill's capabilities are limited to reading local files and making network GET requests, with no system-modifying or file-write permissions. 4. Sanitization: While titles are processed for matching logic, they are displayed in their raw form in the report provided to the user.
Audit Metadata