Detect, confirm, store, and enforce user-declared guidelines, constraints, and guardrails as JSON files. Pristino's internal RAG for working rules. [EXPLICIT] Trigger: "guardrail", "guideline", "constraint", "from now on", "always use", "never use"
From jm-adknpx claudepluginhub javimontano/jm-adk-alfaThis skill is limited to using the following tools:
agents/guardian.mdagents/lead.mdagents/specialist.mdagents/support.mdevals/evals.jsonknowledge/body-of-knowledge.mdknowledge/knowledge-graph.mdprompts/meta.mdprompts/primary.mdprompts/variations/deep.mdprompts/variations/quick.mdtemplates/output.docx.mdtemplates/output.html"Rules you declare become rules the system enforces."
Manages user-declared working rules stored as JSON in references/guardrails/. When the user expresses a preference ("always use TypeScript", "never use jQuery"), Pristino detects the intent, confirms with the user, and stores it. These rules are loaded at every session start and the Guardian checks compliance. [EXPLICIT]
references/guardrails/guidelines.json, constraints.json, guardrails.jsonconstraints.jsonguidelines.jsonguardrails.jsonGL-NNN (guidelines), CT-NNN (constraints), GR-NNN (guardrails){ "id": "GL-001", "rule": "...", "type": "guideline", "confirmed": "YYYY-MM-DD", "source": "user-explicit", "active": true }
"active": false on the specified rule| Anti-Pattern | Why It's Bad | Do This Instead |
|---|---|---|
| Storing without confirmation | User didn't intend a permanent rule | Always double-confirm |
| Mixing types | "Never" rules in guidelines file | Classify by enforcement level |
| Storing unverifiable rules | Guardian can't check "make it nice" | Rules must be specific and testable |
session-protocol — Loads guardrails during bootstrapcontinuous-learning — Insights may generate new guardrailscode-review — Guardian checks guardrail complianceExample invocations:
| Scenario | Handling |
|---|---|
| Empty or minimal input | Request clarification before proceeding |
| Conflicting requirements | Flag conflicts explicitly, propose resolution |
| Out-of-scope request | Redirect to appropriate skill or escalate |
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.