The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it is designed to ingest and analyze external data from conference CFP (Call for Papers) pages and previous talk descriptions.
Ingestion points: The agent is instructed in SKILL.md (Step 1) and references/event-research-checklist.md to read external CFP pages, code of conduct text, and existing talk titles to extract constraints and signals.
Boundary markers: The skill does not prescribe specific delimiters (like XML tags or markdown blocks) to isolate untrusted external content from the agent's instructions, though it does ask the agent to state constraints back to the user.
Capability inventory: The skill is restricted to text generation and maintaining a local state file (devrel-context.md). It does not possess capabilities for arbitrary command execution, network exfiltration of sensitive data, or privileged file system access.
Sanitization: There are no explicit instructions for sanitizing or escaping the content fetched from external URLs before processing.
[SAFE]: The skill references established and legitimate industry resources such as speakerline.io for research purposes. These are standard tools for the developer relations use-case.
[SAFE]: The skill includes handoff references to other specialized skills within the same author's namespace (e.g., samber/developer-relations-skills@tech-talk-outline), which is a standard practice for modular agent workflows.

conference-cfp-submission