The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes DNS records fetched from the Tencent Cloud API. Maliciously crafted DNS records (e.g., TXT records containing text like "Ignore previous instructions") could attempt to influence the agent's behavior when it reads them. However, the skill returns this data as structured JSON, which helps mitigate accidental interpretation of data as instructions.
Ingestion points: The list_records method in scripts/dns_manager/providers/tencentcloud.py fetches data from the remote API.
Boundary markers: Output is formatted as JSON, providing a clear boundary for the agent between data and instructions.
Capability inventory: The skill can perform network operations (Tencent API) and read configuration files.
Sanitization: Record values are treated as data literals; no specific sanitization for prompt injection is performed on the record content itself.
[DATA_EXFILTRATION]: The skill communicates with dnspod.tencentcloudapi.com to manage DNS records. This is a well-known service provided by Tencent Cloud and is the primary purpose of the skill. The network requests are targeted and authenticated using standard HMAC-SHA256 signatures.
[COMMAND_EXECUTION]: The skill is invoked via a command-line interface as described in SKILL.md. It executes its own bundled Python scripts using the uv runner. This is the intended execution model and does not involve running arbitrary user-supplied commands.

dns-manager