The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches conference-related data and executes a CLI tool from the author's official infrastructure and package registries.
Evidence: SKILL.md references endpoints at https://ai.engineer and uses npx @aidotengineer/aie to run the conference CLI.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes external data (speaker bios, session descriptions) that could theoretically contain malicious instructions.
Ingestion points: SKILL.md (API endpoints), scripts/fetch-schedule.sh, and scripts/fetch-speakers.py ingest data from https://ai.engineer.
Boundary markers: Not present in the provided implementation examples.
Capability inventory: Performs network requests and data processing using curl, fetch, requests, and urllib.
Sanitization: No explicit sanitization or filtering of the fetched text content was observed in the examples.
[SAFE]: All identified external resources, including domains and software packages, are owned by the skill's author (aidotengineer) or represent well-known technology services. The skill's behavior aligns with its stated purpose of providing conference metadata, and no malicious patterns were detected.

aie-europe-2026