oc-ted-talk

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The required workflow depends on fetching the community-authored skill documentation from clawskills.sh (line 25), a third-party marketplace whose content the agent must actively retrieve rather than receive from a feed an outsider can push into.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The source: field is recorded metadata, but lines 25-27 instruct the agent to consult https://clawskills.sh/skills/ted-talk for the skill's core functionality, so that content governs behavior. The publisher is an individual author (luo-kai) on a third-party marketplace whose ownership and distribution are not established, so provenance is weak/unknown and the tier is high.