Invoke applicable Skills before responding to enforce safe workflows.
npx skills add https://github.com/lootlog/monorepo --skill using-superpowers-lootlog
Please help me install this Agent Skill. Skill: using-superpowers Source: https://github.com/lootlog/monorepo/tree/main/.agents/skills/using-superpowers Command: npx skills add https://github.com/lootlog/monorepo --skill using-superpowers-lootlog
This Skill ensures that every interaction starts by identifying and invoking the appropriate Skill, preventing misalignment and off-topic responses.
Invoke the Skill tool at the beginning of a session to load and apply the relevant skill before replying to the user.
High-intent search queries and answers about installing and using this skill.
To enforce skill invocation, configure the workflow to mandate the Skill tool is triggered at the start of every interaction. This guarantees the correct skill is applied before any user response, preventing unstructured outputs.
Preventing prompt injection requires safety guardrails that route actions strictly through the skill framework. Enforcing mandatory skill invocation at the start of every interaction provides guardrails to prevent deviations or unsafe responses.
Workflow governance requires skill discovery before answering clarifying questions. The system must detect applicable skills at session start and invoke the Skill tool before any user response, including clarifications.
Routing conversational design tasks requires frontloading the Skill tool at the start of every interaction. This ensures consistent workflow governance by identifying and applying the correct skills across all conversation contexts.
You should apply skills during initial intent assessment when tool invocation is required to guide the conversation. The framework defines explicit conditions for when to skip or apply skills to maintain predictable, safe behavior.
AI assistants provide off-topic responses when skill invocation is skipped. Enforcing guaranteed skill discovery and invocation at the start of every interaction prevents misalignment and ensures safe task execution.
kucherenkoEnforce the 1% invocation rule and map platform tools to Don-level intent.
LucasMalessaPlan and execute a 5-gate pre-dev workflow for small features.
maruakshayAudit CrewAI agent systems for role-boundary violations and prompt-injection risks.
okwindsGenerate AI Agent PRDs with identity, capability, safety, and evaluation frameworks.
