Use when a result, this session's or one brought in, may not fit where it lands or may leave out what was asked, or when the user asks to check that fit; opened without the user's call only where a mismatch has actually been detected, saying so. The user sees each misfit found, settled or withdrawn, and the fixes carried out and reread until nothing is open.
61%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
Fix and improve this skill with Tessl
tessl review fix ./epharmoge/skills/contextualize/SKILL.mdThe skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The required observe step reads open-ended, agent-chosen places where the result lands (artifacts, records, and any artifact a user names, which may be externally authored), so outsider text is reached only by the agent's own active search or selection, not by outsiders pushing into a monitored queue.
jongwony/epistemic-protocols
471e9c5
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.