The Agent Skills Directory

[SAFE]: The skill exclusively uses the Ruby standard library (json, net/http, uri, fileutils), which eliminates risks associated with third-party dependency vulnerabilities or malicious package supply chain attacks.
[SAFE]: Network operations are restricted to the official Heroku API (api.heroku.com), a well-known and trusted service. No data is sent to unauthorized third-party domains.
[SAFE]: Credentials are handled securely in scripts/save_token.rb by storing the API token in a local file with restrictive permissions (0600), ensuring only the local user can access the token.
[SAFE]: Potential data exposure is mitigated in scripts/config_vars.rb by masking configuration variable values by default, requiring an explicit --reveal flag for the user to view secrets.
[SAFE]: Infrastructure management actions, such as restarting dynos or scaling resources, are designed to require explicit user confirmation through the agent interface before execution.
[SAFE]: Indirect Prompt Injection Surface Analysis: 1. Ingestion points: Data returned from Heroku API calls in all management scripts (e.g., apps.rb, logs.rb). 2. Boundary markers: Not explicitly implemented in the console output of the scripts. 3. Capability inventory: Infrastructure modification tools including restart.rb (DELETE request) and scale.rb (PATCH request). 4. Sanitization: No specific sanitization of API-returned strings before display. This represents a low-risk surface inherent to the tool's management functionality, mitigated by the agent's own safety protocols.

heroku