Claude Product-Legal Plugin
Run Anthropic's product-legal plugin: 7 skills + launch-watcher daily agent. Calibration table from your past 10 launch reviews, PM self-serve triage, full launch reviews, 5-category marketing claims taxonomy, feature risk assessments. 8 lessons + certificate.
Anthropic shipped a product counsel plugin on May 12, 2026 — and it learns from your launch history
The product-legal plugin covers in-house product counsel workflows: launch reviews calibrated to your firm’s actual practice, marketing claims review across 5 taxonomic categories, feature risk assessments for the 10% of launches needing depth, and PM self-serve triage for the 80% of trivial questions. Built around a calibration table learned from your past 10 launch reviews — what your company actually blocks vs waves through, not generic best-practice frameworks.
The plugin reads your launch tracker (Jira / Linear / Asana) and your past launch reviews. From these, it captures: auto-block categories (always require legal hold), auto-wave categories (always cleared), case-by-case categories (mixed outcomes), risk thresholds, and your GC’s positions on marginal cases. Subsequent skills apply this calibration automatically — is-this-a-problem pattern-matches new PM questions; launch-review categorizes new launches against your framework; marketing-claims-review applies FTC substantiation framework calibrated to your firm’s enforcement posture.
This course walks the 7 skills: cold-start-interview, is-this-a-problem, launch-review, marketing-claims-review, feature-risk-assessment, matter-workspace, customize — plus the launch-watcher daily scheduled agent that surfaces upcoming launches needing review. You’ll learn the FTC §5 substantiation framework, the 5-category claim taxonomy, the strict-liability trap of absolute claims, common traps the skill catches (class shift, jurisdiction shift, regulator shift, implicit-consent overreach), and the daily cadence operational discipline.
You’ll come out with a CLAUDE.md profile at ~/.claude/plugins/config/claude-for-legal/product-legal/CLAUDE.md, a sample is-this-a-problem triage log demonstrating 10 PM questions answered same-minute, and a credential (PROD-XXXXXX) documenting the work. The plugin costs nothing beyond your Claude Pro subscription. The operating discipline this course teaches is the difference between AI that helps with launch reviews and AI that your firm actually runs as the everyday product counsel workflow.
What You'll Learn
- Install and cold-start the product-legal plugin using your past 10 launch reviews to build the calibration table
- Run `is-this-a-problem` for same-minute PM self-serve triage (Fine / Needs look / Hold)
- Run `launch-review` for category-by-category framework analysis with required-before-launch vs nice-to-have triage
- Apply `marketing-claims-review` 5-category taxonomy (puffery / factual / comparative / implied / absolute) with FTC substantiation framework
- Decide when to invoke `feature-risk-assessment` for the 10% of launches needing depth; document risk acceptance with GC sign-off
After This Course, You Can
What You'll Build
Course Syllabus
Prerequisites
- Active product counsel, in-house attorney reviewing launches, or product manager doing self-serve risk triage role
- Recommended: Some past launch review history (10+ ideal; the plugin handles fewer with 'limited evidence' flag)
- Claude Cowork or Claude Code installed; ability to install plugins from the Anthropic marketplace
Who Is This For?
- Product counsel (in-house + outside)
- Product managers doing self-serve risk triage
- Marketing leads responsible for claims compliance
- GC overseeing escalated launch reviews
- Compliance officers tracking launch tracker activity
- Outside counsel doing private-practice product work (matter-workspace)
Frequently Asked Questions
Does the plugin replace platforms like LaunchDarkly, ProductBoard, or LinearB?
No — those are product-management platforms. The plugin sits adjacent: it consumes launch tracker tickets (Jira/Linear/Asana) and produces legal analysis. The platforms manage product work; the plugin manages product counsel work.
What if my firm has fewer than 10 past launch reviews?
The plugin reads what's available and flags calibration as 'limited evidence — re-run with more data when available.' Under 5 past reviews still yields a useful starting point but should re-run cold-start as your library grows. The skill is honest about confidence based on sample size.
How is this different from generic risk frameworks like ISO 31000 or COSO?
Generic frameworks are aspirational and one-size-fits-all. The product-legal calibration is empirical and firm-specific — built from what your firm actually did, not what a textbook says you should do. The empirical approach captures regulatory history, consent decrees, GC posture, and board direction that generic frameworks miss.
PM self-serve via is-this-a-problem — what about ethical concerns?
PMs use is-this-a-problem for triage, not legal opinion. The skill produces ✅ Fine for clearly cleared questions (no review needed) and ⚠ Needs look / 🛑 Hold for everything else (routed to legal). PMs cannot bypass review for non-Fine outcomes. This isn't PM self-lawyering — it's triage.
How does this work with other practice-area plugins?
Product-legal pairs with: Commercial-Legal (MSA/SOW on launches), Privacy-Legal (consent + data flow for product features), IP-Legal (TM clearance for product names + marketing claims), and Regulatory-Legal (jurisdiction-specific compliance). Many firms run the full practice-area plugin stack.
