The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates fanning out multiple parallel workers and aggregating their results into a single report. This pattern introduces a surface for indirect prompt injection, as malicious instructions present in data processed by sub-agents could potentially influence the parent agent during the aggregation phase.\n
Evidence: Instructions in SKILL.md and the aggregation phase described in references/swarm-procedure.md.\n- [NO_CODE]: The skill consists entirely of markdown instructions and configuration files. It does not ship with any executable scripts, binaries, or automated installation commands.\n
Evidence: File contents for SKILL.md, plugin.json, and references/swarm-procedure.md.\n- [EXTERNAL_DOWNLOADS]: The skill documents its origins and licensing by referencing GitHub repositories from Anthropics and the vendor 'Ship Shit Dev' (shipshit.dev). These references are for attribution and metadata and are considered safe under the trust scope guidelines.\n
Evidence: SKILL.md metadata and NOTICE.md attribution sections.

swarm