The Agent Skills Directory

[PROMPT_INJECTION] (LOW): The skill is susceptible to Indirect Prompt Injection (Category 8) because it ingests and processes untrusted data exports from external project management tools. A malicious file could attempt to influence the agent's logic or summaries.
Ingestion points: parse_input.py processes user-uploaded JSON, CSV, and YAML files.
Boundary markers: No explicit boundary markers or 'ignore embedded instructions' warnings are present in the parsing logic.
Capability inventory: Data analysis, reporting, and optional notification via webhooks.
Sanitization: Employs yaml.safe_load() which prevents basic code execution during parsing.
[EXTERNAL_DOWNLOADS] (SAFE): The skill depends on the PyYAML library for configuration and data parsing. It correctly uses yaml.safe_load() to mitigate risks associated with untrusted YAML payloads.
[DATA_EXFILTRATION] (SAFE): The skill features documented support for Slack and Microsoft Teams webhooks for notifications. These are user-configured via environment variables or a local config file. No hardcoded credentials or unauthorized data exfiltration mechanisms were detected.
[COMMAND_EXECUTION] (SAFE): The detect_context.py helper script retrieves environment variables and terminal metadata (e.g., shutil.get_terminal_size) to optimize the UI. It does not execute arbitrary shell commands or access sensitive system paths.

scrum-master-agent