The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data into the agent's context, creating a surface for indirect prompt injection.
Ingestion Points: Customer feedback is ingested from various third-party platforms including Slack, Intercom, Salesforce, Zendesk, and email, as documented in SKILL.md and assets/productboard-insight-triage-workflow.md.
Boundary Markers: There are no explicit instructions or boundary markers provided to help the agent distinguish between customer feedback and its own instructions when processing these 'Insights'.
Capability Inventory: The skill provides a wide range of capabilities via the Productboard REST API, including creating/updating features, managing releases, and pushing data to engineering trackers like Jira and Linear (references/productboard-api-patterns.md).
Sanitization: While the skill mentions sanitizing feature descriptions for 'External Roadmaps' (SKILL.md), it lacks specific mechanisms or guidance for the agent to sanitize or ignore potential malicious instructions embedded within the raw customer verbatims it triages.
[SAFE]: No obfuscation, persistence mechanisms, or malicious command execution were detected. All network operations are directed at the official Productboard API domain (api.productboard.com), which is a well-known service for this tool. The skill correctly identifies the use of environment variables for sensitive tokens as a best practice.

productboard-expert