The Agent Skills Directory

[SAFE]: The skill implements and teaches strong security practices for handling third-party integrations, including constant-time HMAC verification and strict schema validation for inbound data.
[EXTERNAL_DOWNLOADS]: The skill provides links to reference implementations in GitHub repositories (e.g., github.com/hexastack/hexabot-channel-facebook). These are official resources provided by the vendor for developer guidance and do not involve unverified or hidden downloads.
[DATA_EXFILTRATION]: While the skill involves handling API credentials and sending data to external messaging platforms, it explicitly instructs against logging secrets and promotes the use of a dedicated CredentialService for secure resolution of tokens.
[PROMPT_INJECTION]: The skill defines a mechanism for ingesting untrusted data from external messaging platforms (webhooks).
Ingestion points: Inbound webhook payloads processed in index.channel.ts via Zod parsing.
Boundary markers: Uses strict Zod schemas to define expected data shapes and ignore malformed inputs.
Capability inventory: Includes network requests via fetch to messaging APIs in the outbound encoder.
Sanitization: Implements HMAC signature verification and explicit schema validation to ensure data integrity before it enters the system context.

hexabot-channel-creator