toggl-track
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from Toggl Track API responses and webhook payloads which could potentially contain malicious instructions from an attacker-controlled Toggl account.
- Ingestion points: API response payloads are read by
scripts/toggl_api.pyand passed into the agent's context inSKILL.md. - Boundary markers: While no specific character delimiters are defined for API responses, the
SKILL.mdinstructions explicitly warn the agent to treat these responses as untrusted data and ignore embedded instructions. - Capability inventory: The agent can perform authenticated network read and write operations to the Toggl Track API via the provided helper script.
- Sanitization: The skill uses the Python standard library's
json.dumpsandurllib.parse.urlencodeto ensure data sent to the API is properly formatted. - [EXTERNAL_DOWNLOADS]: The skill references and fetches OpenAPI specification documents from Toggl's official engineering assets to define the API structure.
- Evidence:
SKILL.mdcontains links toengineering.toggl.comfor API contract documentation. - [COMMAND_EXECUTION]: The skill executes a bundled Python script to facilitate network communication with the Toggl API.
- Evidence:
scripts/toggl_api.pyis the primary interface for API calls. It is restricted to the Python standard library and includes robust error handling that redacts the user's API token from stderr output.
Audit Metadata