The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions utilize 'silently identify' and 'silently infer' patterns for rhetorical traditions and calibration variables. These directives are used for internal chain-of-thought reasoning to improve the accuracy of the diagnostics before presenting the final report. While this technique hides the agent's internal reasoning steps from the user, it is used here for task performance rather than to conceal malicious intent or bypass safety guardrails.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and analyze untrusted text provided by users (manifestos). While this constitutes an attack surface for indirect prompt injection, the skill lacks the necessary capabilities to cause harm. There are no tools available for file writing, network exfiltration, or command execution, and no boundary markers are specified to separate user text from instructions. The impact of a successful injection would be limited to manipulating the coaching report's content within the current conversation.

manifesto-coach