The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions use strong directives that encourage bypassing the model's internal deliberation and safety checks.
Evidence includes phrases like "without hesitation or deterrence," "Stop thinking, start doing," and "antidote to analysis paralysis."
The "Core Philosophy" of "BIAS TOWARD ACTION" and "No Excuses" explicitly instructs the agent to override the cautious behavior typically expected when handling system tools.
[COMMAND_EXECUTION]: The skill is granted Bash permissions and is instructed to prioritize speed and "good enough" solutions over careful analysis.
The "15-Minute Rule" and "Blocker Resolution Playbook" encourage the immediate implementation and execution of workarounds without second-guessing, which significantly increases the likelihood of running destructive or insecure shell commands.
[PROMPT_INJECTION]: The skill creates a high vulnerability to indirect prompt injection due to its structural bias against input validation.
Ingestion points: Task descriptions, blocker information, and requirements processed during execution.
Boundary markers: None identified. The skill does not define delimiters or provide instructions to ignore malicious content within task data.
Capability inventory: High-privilege tools including Bash, Write, and Edit are available to the executor.
Sanitization: Absent. The skill explicitly discourages the "Analysis Paralysis" of researching approaches or planning carefully, making it likely to follow instructions embedded in untrusted data to maintain "forward momentum."

swift-executor