The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes structured error responses from external Adobe API endpoints, which could theoretically contain malicious instructions.
Ingestion points: Adobe API response bodies and error messages processed during troubleshooting steps in SKILL.md.
Boundary markers: Not explicitly defined; the skill presents findings for user review and manual intervention.
Capability inventory: The skill utilizes Bash (via curl) for network requests and status verification.
Sanitization: Relies on manual user validation of command output and suggested fixes.
[EXTERNAL_DOWNLOADS]: Fetches real-time service incident reports from Adobe's official status monitoring API.
The skill provides a command to retrieve status data from status.adobe.com. This uses an established service and is standard for operational visibility.
[COMMAND_EXECUTION]: Provides validated command templates for common integration tasks.
Includes curl templates for OAuth 2.0 token generation using Adobe's Identity Management Services (IMS).
Utilizes nslookup for network connectivity diagnostics targeting official Adobe domains like firefly-api.adobe.io and pdf-services.adobe.io.
Securely handles credentials by referencing environment variables (e.g., ${ADOBE_CLIENT_ID}) rather than hardcoding sensitive values.
[REMOTE_CODE_EXECUTION]: The skill includes a pipe to Python (python3 -m json.tool) which was flagged by automated scanners. Analysis confirms this is not remote code execution; it is the use of a standard built-in Python module to pretty-print JSON output for human readability.

adobe-common-errors