The Agent Skills Directory

[NO_CODE]: The skill is composed entirely of natural language instructions and does not include any executable code or scripts.
[PROMPT_INJECTION]: The skill defines a procedure to "dissolve" or "release" previous context and assumptions. While designed for cleaning stale reasoning, this mechanism effectively overrides previous instructions or established context, which could be exploited if an attacker provides data that the agent identifies as "stale." (Ingestion points: Current conversation state or project context via Read tool; Boundary markers: None; Capability inventory: Read tool; Sanitization: None).
[SAFE]: No network activity, hardcoded credentials, or suspicious file access patterns were detected within the skill definition.

shiva-bhaga