75 lines
2.7 KiB
YAML
75 lines
2.7 KiB
YAML
name: launch-radar-risk-classifier
|
|
model: claude-opus-4-7
|
|
system:
|
|
text: |
|
|
You receive a validated list of upcoming launches and the product counsel's
|
|
risk calibration (loaded from the plugin's CLAUDE.md or equivalent config).
|
|
For each launch, classify against the calibration:
|
|
|
|
- needs-review : matches a "usually blocks" or "usually requires work" pattern
|
|
- needs-flag : novel area, AI component detected, new jurisdiction, or
|
|
policy-adjacent — product counsel should be aware, not
|
|
necessarily full review
|
|
- fyi : on the radar but matches a "usually FYI" pattern
|
|
- skip : UI/infra/copy-only, no legal trigger
|
|
|
|
Also record which risk domains triggered (privacy, AI governance,
|
|
marketing, IP, contractual, regulatory, security, third-party) and a
|
|
one-line rationale. If the target_date is within the user's stated lead
|
|
time, set review_by_date to target_date minus the lead time (clamped to
|
|
today).
|
|
|
|
You are READ-ONLY. No MCP, no web, no Write. The launch list is trusted
|
|
(tracker-reader already validated it); the user's calibration file is
|
|
trusted. Return only schema-validated JSON.
|
|
tools:
|
|
- type: agent_toolset_20260401
|
|
default_config: { enabled: false }
|
|
configs:
|
|
- { name: read, enabled: true }
|
|
- { name: grep, enabled: true }
|
|
mcp_servers: []
|
|
skills:
|
|
- { path: ../../../product-legal/skills/is-this-a-problem }
|
|
callable_agents: []
|
|
output_schema:
|
|
type: object
|
|
required: [classified]
|
|
additionalProperties: false
|
|
properties:
|
|
classified:
|
|
type: array
|
|
maxItems: 200
|
|
items:
|
|
type: object
|
|
required: [launch_id, classification, rationale]
|
|
additionalProperties: false
|
|
properties:
|
|
launch_id:
|
|
{ type: string, maxLength: 64, pattern: "^[A-Za-z0-9_-]+$" }
|
|
classification:
|
|
{ type: string, enum: ["needs-review", "needs-flag", "fyi", "skip"] }
|
|
triggers:
|
|
type: array
|
|
maxItems: 8
|
|
items:
|
|
type: string
|
|
enum:
|
|
- privacy
|
|
- ai-governance
|
|
- marketing
|
|
- ip
|
|
- contractual
|
|
- regulatory
|
|
- security
|
|
- third-party
|
|
- jurisdictional
|
|
- childrens-privacy
|
|
- teen-privacy
|
|
- health-data
|
|
severity:
|
|
{ type: string, enum: ["p0", "p1", "p2", "p3"] }
|
|
rationale:
|
|
{ type: string, maxLength: 400, pattern: "^[A-Za-z0-9 .,:;!?()&'\"/_#%+*@=-]+$" }
|
|
review_by_date:
|
|
{ type: string, maxLength: 10, pattern: "^[0-9]{4}-[0-9]{2}-[0-9]{2}$" }
|