The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill explicitly directs the creation of instructions in SKILL.md and references/executar-specs-template.md to run the agent with the --dangerously-skip-permissions flag. This flag is a critical security bypass that allows the agent to execute shell commands and modify files without individual user approval, effectively neutralizing the platform's primary safety controls.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, untrusted PRDs and requirements documents into executable subagent tasks without any sanitization or boundary enforcement.
Ingestion points: PRDs, requirement documents, and user stories provided as input to the agent (referenced in SKILL.md).
Boundary markers: The templates provided (spec-template.md) do not include markers or instructions to differentiate between structural instructions and the potentially malicious data being processed.
Capability inventory: The generated outputs lead to subagents performing file modifications and shell executions (described in spec-template.md under 'Comandos de Validação').
Sanitization: The skill lacks any mechanism to sanitize or validate the content of the ingested documents before using them to generate executable specifications.
[COMMAND_EXECUTION]: The orchestration workflow involving EXECUTAR-SPECS.md is specifically built to automate the execution of development tasks. When combined with the instruction to skip permission prompts, it facilitates an environment where unverified and potentially harmful commands can be executed autonomously across the project.

sdd-generator