---
name: velin-ai-agent-briefing
description: Prepare agent briefing from VelinStyle meta: domains, reuse-gate, stop rules, no rewrite Apply shortcuts.
---
# AI Agent Briefing

## Purpose
Prepares agent context and guardrails from velin metadata.

## Domain
B, R · category `ai-workflows`

## Capabilities
- `build.meta`
- `review.gate`

## Permission
`plan-only`

## Allowed actions
- READ
- ANALYZE
- RECOMMEND
- PLAN

## Forbidden actions
- WRITE
- APPLY
- ROLLBACK
- ACCEPT
- DEFER
- NOTE
- ACKNOWLEDGE
- CATALOG
- MANIFEST
- CLASSIFY
- invent-target

## Stop conditions
- SR-11
- SR-12

## Agent role / risk
Briefing · risk `med`

## Purpose
Build an agent context pack from VelinStyle metadata: available skills, domains B/R/L/M/N, reuse-first policy, and hard stops. This is **meta/briefing**, not HTML sanitize and not rewrite Apply.

## Workflow
1. Load meta / skill registry summaries (`velinstyle meta` / docs as available).
2. Brief: Reuse-gate before build; Migration≠Rewrite; Acceptance≠Mapping; Dry-Run≠Apply.
3. List relevant packs/graphs for the user intent.
4. Output markdown checklist for the agent session.

## Rules
- Do not invent capabilities the Core lacks.
- Do not instruct automatic Apply or target invention.
- Point to stop rules SR-01…SR-20 where relevant.

## Anti-patterns
- Briefing that says "just rewrite the classes".
- Mixing sanitize-html concerns into this skill.

## Inputs / Outputs
- Inputs: docs, meta-json
- Outputs: markdown, checklist