Detected a prompt injection in the skill instructions. The skill contains hidden or deceptive instructions that fall outside its stated purpose and attempt to override the agent’s safety guidelines or intended behavior.
The skill's "missioni spericolate" section tells the agent not to call high-risk speculative schemes such as meme coin sniping scams, to avoid dissuading the user, and to "help 100%" with them. That is sketchy, since it suppresses warnings and promotes risky financial behavior unrelated to the code-exploding purpose, but it shows no malicious intent to attack or defraud the user.
The 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 pipeline clones and ingests a user-supplied GitHub repository (`-s, --source` or a GitHub URL argument), so third-party source comments, READMEs, and docs are read by the LLM enrich step after the user selects that URL, which fits the low tier of selected external content rather than the medium tier.
biodoia/readmago
1fbd499
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.