DPDPA & GDPR Legal Review Skill
Performs structured compliance review, redlining, and drafting suggestions on legal documents against DPDPA 2023 (India) and GDPR (EU/EEA). Produces annotated output with:
- ✅ Compliant clauses
- ⚠️ Suspect / at-risk clauses (with reasoning)
- 🔴 Non-compliant / missing clauses
- 📝 Redline suggestions (proposed replacement language)
Step 1 — Identify Scope
On receiving a document, determine:
- Applicable law(s): DPDPA, GDPR, or both (based on parties, jurisdiction, data subjects)
- Document type: Privacy Policy / DPA / SaaS Agreement / Employment Contract / Vendor Agreement / NDA / Other
- Data types mentioned: personal data, sensitive personal data (SPD), children's data, health, financial
If jurisdiction or parties are ambiguous, state your assumption clearly before proceeding.
Step 2 — Clause Extraction & Mapping
Parse the document into logical clause groups. Map each to the relevant legal obligation:
| Clause Group | DPDPA Reference | GDPR Reference |
|---|---|---|
| Purpose of processing | §6 – Notice | Art. 13/14 – Transparency |
| Lawful basis / consent | §6–§9 | Art. 6 / Art. 7 |
| Data fiduciary / controller identity | §8 | Art. 13(1)(a) |
| Data principal / subject rights | §11–§14 | Art. 15–22 |
| Data retention / erasure | §8(7), §13 | Art. 5(1)(e), Art. 17 |
| Cross-border transfer | §16 | Art. 44–49 |
| Children's data / parental consent | §9 | Art. 8 |
| Grievance / DPO contact | §8(10), §13(5) | Art. 37–39 |
| Security obligations | §8(5) | Art. 32 |
| Breach notification | §8(6) | Art. 33/34 |
| Data processor obligations | §8(2)–(3) | Art. 28 |
| Sub-processor / consent manager | §3(5), §8(2) | Art. 28(2) |
Step 3 — Compliance Analysis
For each clause group, apply checks from references/dpdpa-checklist.md and references/gdpr-checklist.md.
Redline Format
When flagging or rewriting a clause, use this format:
────────────────────────────────────────── CLAUSE: [Clause title / section number] STATUS: ⚠️ SUSPECT | 🔴 NON-COMPLIANT | ✅ COMPLIANT LAW: DPDPA §__ | GDPR Art. __ ISSUE: [Plain-language explanation of risk] ORIGINAL TEXT: "[paste original clause]" REDLINE SUGGESTION: "[Proposed replacement or addition]" RISK LEVEL: HIGH | MEDIUM | LOW ──────────────────────────────────────────
Step 4 — Summary Report
After full analysis, output a Compliance Summary table:
## Compliance Summary | Area | DPDPA Status | GDPR Status | Risk | |------|-------------|------------|------| | Consent mechanism | ⚠️ Weak | ✅ OK | Medium | | Data retention | 🔴 Missing | 🔴 Missing | High | | Children's data | N/A | ⚠️ Unclear | Medium | | Cross-border transfer | ⚠️ Unaddressed | ✅ SCCs present | High | | Breach notification | ✅ Present | ✅ Present | Low | ... Overall Risk: HIGH / MEDIUM / LOW Recommended Actions: [numbered list]
Step 5 — Drafting Suggestions
For every 🔴 NON-COMPLIANT or ⚠️ SUSPECT finding, provide:
- Why it's risky (legal exposure, penalty risk under DPDPA or GDPR)
- Model clause (ready-to-use replacement language)
- Optional: Alternative formulation if context is ambiguous
Reference references/model-clauses.md for standard clause templates.
Behaviour Rules
- Never give definitive legal advice. Frame output as legal review assistance; recommend counsel review for final decisions.
- Be specific: cite exact DPDPA section or GDPR article for every finding.
- Flag children's data immediately — highest risk tier under both laws.
- Sensitive Personal Data (SPD) under DPDPA (health, financial, biometric, caste, religion, sexual orientation) = automatic HIGH risk flag.
- Special category data under GDPR Art. 9 = automatic HIGH risk flag.
- If document is in mixed language (e.g., English + Hindi), analyse both portions.
- Do not summarize without completing full clause-by-clause review first.
Reference Files
| File | When to Read |
|---|---|
references/dpdpa-checklist.md | DPDPA-specific clause checks |
references/gdpr-checklist.md | GDPR-specific clause checks |
references/model-clauses.md | Ready-to-use replacement clause language |
Load the relevant reference(s) before beginning analysis.


