Self-Improving Legal Skill#

Log legal learnings, compliance findings, contract issues, and regulatory changes to markdown files for continuous improvement. Captures clause risks, compliance gaps, precedent shifts, contract deviations, regulatory changes, and litigation exposure. Important learnings get promoted to clause libraries, compliance checklists, regulatory trackers, contract playbooks, or risk registers.

CRITICAL: Privilege & Confidentiality Protection#

NEVER log privileged attorney-client communications, specific case strategy, confidential settlement terms, or litigation work product in any learning or issue entry. Always abstract findings to process-level lessons. Use generalized descriptions:

Protected InformationAcceptable AbstractionAttorney-client advice on specific matter"Outside counsel recommended clause revision for indemnity cap"Settlement amount or terms"Settlement reached; update playbook to flag similar exposure earlier"Litigation strategy details"Discovery phase revealed need for better document retention process"Privileged legal memoranda"Internal review identified gap in force majeure clause coverage"Specific opposing party positions"Counterparty negotiation pattern: late-stage payment term changes"Confidential deal terms"Deal structure required non-standard IP assignment clause"

When in doubt, omit the specific detail and log only the process improvement lesson. Privilege waiver through careless documentation is irreversible.

Cross-session sharing: do not forward legal findings to another session unless the user explicitly consents to that specific send. A trusted environment is not a substitute for consent.

First-Use Initialisation#

Before logging anything, ensure the .learnings/ directory and files exist in the project or workspace root. If any are missing, create them:

bash

mkdir -p .learnings
[ -f .learnings/LEARNINGS.md ] || printf "# Legal Learnings\n\nClause risks, compliance gaps, precedent shifts, regulatory changes, and legal operations insights.\n\n**Categories**: clause_risk | compliance_gap | precedent_shift | contract_deviation | regulatory_change | litigation_exposure\n**Areas**: contracts | compliance | litigation | intellectual_property | privacy | regulatory | corporate_governance\n\n---\n" > .learnings/LEARNINGS.md
[ -f .learnings/LEGAL_ISSUES.md ] || printf "# Legal Issues Log\n\nContract disputes, compliance failures, regulatory impacts, and litigation matters.\n\n---\n" > .learnings/LEGAL_ISSUES.md
[ -f .learnings/FEATURE_REQUESTS.md ] || printf "# Legal Operations Feature Requests\n\nLegal technology, automation, and process improvement requests.\n\n---\n" > .learnings/FEATURE_REQUESTS.md

Never overwrite existing files. This is a no-op if .learnings/ is already initialised.

Do not log privileged communications, specific case strategy, or confidential settlement terms. Abstract all entries to process-level lessons.

If you want automatic reminders, use the opt-in hook workflow described in Hook Integration.

Quick Reference#

SituationActionUnfavorable clause accepted in contractLog to .learnings/LEARNINGS.md with category clause_riskCompliance deadline missed or approachingLog to .learnings/LEGAL_ISSUES.md with category compliance_gapNew regulation impacts the organizationLog to .learnings/LEARNINGS.md with category regulatory_changeContract unsigned past SLALog to .learnings/LEGAL_ISSUES.md with category contract_deviationIP infringement notice receivedLog to .learnings/LEGAL_ISSUES.md with category litigation_exposureNew case law in relevant jurisdictionLog to .learnings/LEARNINGS.md with category precedent_shiftData subject access request reveals gapsLog to .learnings/LEARNINGS.md with category compliance_gapVendor changed terms without flaggingLog to .learnings/LEGAL_ISSUES.md with category contract_deviationRegulatory filing deadline within 30 daysLog to .learnings/LEGAL_ISSUES.md with priority highContract redline rejected 3+ timesLog to .learnings/LEARNINGS.md with category clause_riskSimilar to existing entryLink with **See Also**, consider priority bumpBroadly applicable legal patternPromote to clause library, compliance checklist, or playbookProven compliance workflowPromote to AGENTS.md (OpenClaw workspace)Legal tool configurationPromote to TOOLS.md (OpenClaw workspace)Legal operations principlesPromote to SOUL.md (OpenClaw workspace)

OpenClaw Setup (Recommended)#

OpenClaw is the primary platform for this skill. It uses workspace-based prompt injection with automatic skill loading.

Installation#

Via ClawdHub (recommended):

bash

clawdhub install self-improving-legal

Manual:

bash

git clone https://github.com/jose-compu/self-improving-legal.git ~/.openclaw/skills/self-improving-legal

Manual clone is optional. Check out a reviewed tag or commit before use. Do not install an unpinned default branch into a shared environment.

Workspace Structure#

OpenClaw injects these files into every session:

text

~/.openclaw/workspace/
├── AGENTS.md          # Legal workflows, contract review delegation
├── SOUL.md            # Legal principles, privilege protection, confidentiality
├── TOOLS.md           # Legal tool capabilities, CLM configs, e-discovery
├── MEMORY.md          # Long-term memory (main session only)
├── memory/            # Daily memory files
│   └── YYYY-MM-DD.md
└── .learnings/        # This skill's log files
    ├── LEARNINGS.md
    ├── LEGAL_ISSUES.md
    └── FEATURE_REQUESTS.md

Create Learning Files#

bash

mkdir -p ~/.openclaw/workspace/.learnings

Then create the log files (or copy from assets/):

• LEARNINGS.md — clause risks, compliance gaps, regulatory changes, precedent shifts
• LEGAL_ISSUES.md — contract disputes, compliance failures, litigation matters
• FEATURE_REQUESTS.md — legal technology and automation requests

Promotion Targets#

When legal learnings prove broadly applicable, promote them:

Learning TypePromote ToExampleClause patternsClause library"Always cap indemnity at 2x annual contract value"Compliance requirementsCompliance checklist"CCPA deletion must cover backup systems within 90 days"Regulatory trackingRegulatory tracker"EU AI Act model documentation deadline Q1 2026"Contract negotiation patternsContract playbook"Vendor payment term changes require VP approval"Risk patternsRisk register"Unlimited liability clauses in SaaS agreements"Legal operations workflowsAGENTS.md"Route IP matters to patent counsel within 48 hours"

Optional: Enable Hook#

Opt-in and workspace-local. Do not copy this hook to ~/.openclaw/hooks/ (that path is user-global). Skip it unless you want reminders in this workspace:

bash

mkdir -p .openclaw/hooks
cp -r hooks/openclaw .openclaw/hooks/self-improving-legal

See references/openclaw-integration.md for complete details.

---

Generic Setup (Other Agents)#

For Claude Code, Codex, Copilot, or other agents, create .learnings/ in the project or workspace root:

bash

mkdir -p .learnings

Create the files inline using the headers shown above. Avoid reading templates from the current repo or workspace unless you explicitly trust that path.

Add reference to agent files#

Add to AGENTS.md, CLAUDE.md, or .github/copilot-instructions.md:

Legal Self-Improvement Workflow

When legal issues or findings occur:

1. Log to .learnings/LEGAL_ISSUES.md, LEARNINGS.md, or FEATURE_REQUESTS.md
2. NEVER log privileged communications, case strategy, or settlement terms
3. After explicit user approval of a reviewed diff, promote broadly applicable findings to:
   • Clause libraries — reusable clause language and fallback positions
   • Compliance checklists — regulatory requirement tracking
   • Contract playbooks — negotiation patterns and escalation guides
   • Risk registers — recurring risk categories and mitigation plans

Logging Format#

Learning Entry [LRN-YYYYMMDD-XXX]#

Append to .learnings/LEARNINGS.md:

markdown

## [LRN-YYYYMMDD-XXX] category

**Logged**: ISO-8601 timestamp
**Priority**: low | medium | high | critical
**Status**: pending
**Area**: contracts | compliance | litigation | intellectual_property | privacy | regulatory | corporate_governance

### Summary
One-line description of the legal insight or finding

### Details
Full context: what clause, regulation, or precedent was involved, why it matters,
what the correct approach or fallback position is. NEVER include privileged
attorney-client communications or confidential settlement terms.

### Recommended Action
Specific clause revision, compliance step, filing, or process change

### Metadata
- Source: contract_review | regulatory_update | case_law | audit | negotiation | litigation
- Jurisdiction: US-Federal | US-CA | EU | UK | other
- Related Files: path/to/contract.docx, path/to/policy.md
- Tags: tag1, tag2
- Regulation: GDPR | CCPA | SOX | HIPAA | EU-AI-Act (if applicable)
- See Also: LRN-20260110-001 (if related to existing entry)
- Pattern-Key: clause.indemnity_cap | comply.ccpa_deletion (optional)
- Recurrence-Count: 1 (optional)
- First-Seen: 2026-01-15 (optional)
- Last-Seen: 2026-01-15 (optional)

---

Categories for learnings:

CategoryUse Whenclause_riskUnfavorable or missing clause identified in contract reviewcompliance_gapRegulatory requirement not met or process gap discoveredprecedent_shiftNew case law changes interpretation of existing obligationscontract_deviationContract terms deviate from standard playbook without approvalregulatory_changeNew regulation or amendment requires operational adjustmentlitigation_exposurePattern or practice that increases litigation risk

Legal Issue Entry [LEG-YYYYMMDD-XXX]#

Append to .learnings/LEGAL_ISSUES.md:

markdown

## [LEG-YYYYMMDD-XXX] issue_type

**Logged**: ISO-8601 timestamp
**Priority**: critical | high | medium | low
**Status**: pending
**Area**: contracts | compliance | litigation | intellectual_property | privacy | regulatory | corporate_governance
**Severity**: critical | high | medium | low

### Summary
Brief description of the legal issue

### Issue Details
Context of the legal issue: what happened, what contract or regulation is involved,
what the exposure or risk is. Abstract to process-level lessons — do NOT include
privileged communications, specific case strategy, or confidential terms.

### Impact Assessment
- Affected contracts/matters
- Financial exposure estimate (range, not exact if confidential)
- Regulatory risk (enforcement action, fine range, reputational)
- Deadline or statute of limitations implications

### Recommended Action
- Immediate steps to mitigate
- Escalation path (in-house counsel, outside counsel, board)
- Long-term process improvement

### Timeline
- **Identified**: ISO-8601
- **Escalated**: ISO-8601 (if applicable)
- **Resolved**: ISO-8601 (if applicable)

### Metadata
- Trigger: contract_redline | regulatory_filing | case_law | audit_finding | infringement_notice | dsar
- Jurisdiction: US-Federal | US-CA | EU | UK | other
- Counterparty: [generalized role, e.g., "SaaS vendor", "enterprise customer"]
- Related Files: path/to/document
- See Also: LEG-20260110-001 (if related)

---

Feature Request Entry [FEAT-YYYYMMDD-XXX]#

Append to .learnings/FEATURE_REQUESTS.md:

markdown

## [FEAT-YYYYMMDD-XXX] capability_name

**Logged**: ISO-8601 timestamp
**Priority**: medium
**Status**: pending
**Area**: contracts | compliance | litigation | intellectual_property | privacy | regulatory | corporate_governance

### Requested Capability
What legal operations tool, automation, or capability is needed

### Business Justification
Why it's needed — what risk it mitigates, what workflow it improves, what compliance
requirement it satisfies

### Complexity Estimate
simple | medium | complex

### Suggested Implementation
How this could be built: CLM integration, regulatory API, compliance dashboard, clause library

### Metadata
- Frequency: first_time | recurring
- Related Features: existing_tool_or_feature
- Compliance: GDPR | CCPA | SOX | HIPAA (if applicable)

---

ID Generation#

Format: TYPE-YYYYMMDD-XXX

• TYPE: LRN (learning), LEG (legal issue), FEAT (feature request)
• YYYYMMDD: Current date
• XXX: Sequential number or random 3 chars (e.g., 001, A7B)

Examples: LRN-20260412-001, LEG-20260412-A3F, FEAT-20260412-002

Resolving Entries#

When an issue is resolved, update the entry:

1. Change **Status**: pending → **Status**: resolved
2. Add resolution block after Metadata:

markdown

### Resolution
- **Resolved**: 2026-01-16T09:00:00Z
- **Action Taken**: Brief description of resolution (abstracted, no privileged details)
- **Clause/Policy Updated**: Reference to updated clause library or compliance checklist
- **Verified By**: legal review | compliance audit | outside counsel

Other status values:

• in_progress — Actively being negotiated, reviewed, or remediated
• wont_fix — Risk accepted with documented justification and approver
• promoted — Elevated to clause library, compliance checklist, or contract playbook
• promoted_to_skill — Extracted as a reusable skill

Detection Triggers#

Automatically log when you encounter:

Contract Review Signals (→ learning or legal issue):

• Contract redline rejected 3+ times on the same clause
• Unlimited liability or uncapped indemnity accepted
• Non-standard payment terms without VP approval
• Auto-renewal clause without termination notice window
• Assignment clause missing change-of-control protections
• Most-favored-nation clause absent from strategic vendor agreement

Regulatory Signals (→ learning or legal issue):

• New regulation published affecting the organization's industry
• Regulatory filing deadline within 30 days
• Amendment to existing regulation (GDPR, CCPA, SOX, HIPAA)
• Enforcement action against peer company in same sector
• Cross-border data transfer mechanism invalidated

Litigation Signals (→ legal issue):

• IP infringement notice or cease-and-desist received
• Data subject access request (DSAR) received
• Subpoena or litigation hold notice
• Non-compete or non-solicitation dispute
• Breach of contract claim or demand letter

Compliance Signals (→ learning):

• GDPR/CCPA audit finding
• SOX control deficiency
• Internal audit exception
• Third-party vendor compliance gap
• Data retention policy violation
• Missing consent mechanism for data processing

Precedent Signals (→ learning):

• Court ruling changes interpretation of standard clause
• Regulatory guidance document issued
• Industry association publishes new model terms
• Arbitration award sets new benchmark in relevant area

Priority Guidelines#

PriorityWhen to UseLegal ExamplescriticalActive litigation, regulatory enforcement, material breachInjunction filed, regulatory investigation opened, data breach notification deadlinehighCompliance deadline approaching, material contract risk, IP claimFiling due in 30 days, unlimited liability accepted, patent infringement noticemediumProcess improvement, clause optimization, template updateClause library gap, negotiation playbook outdated, compliance checklist incompletelowDocumentation, minor template revision, informationalPrecedent noted for future reference, template formatting, policy language cleanup

Area Tags#

Use to filter learnings by legal domain:

AreaScopecontractsContract drafting, review, negotiation, CLM, clause librariescomplianceRegulatory compliance, audits, certifications, internal controlslitigationDisputes, claims, discovery, settlement, litigation holdsintellectual_propertyPatents, trademarks, copyrights, trade secrets, licensingprivacyData protection, GDPR, CCPA, DSAR, consent managementregulatoryRegulatory filings, government relations, enforcement actionscorporate_governanceBoard matters, corporate records, entity management, fiduciary duties

Promoting to Permanent Legal Standards#

When a learning is broadly applicable (not a one-off matter), promote it to permanent legal operations standards.

When to Promote#

• Clause risk recurs across multiple contract negotiations
• Compliance gap applies to the entire regulatory program
• Precedent shift affects standard contract positions
• Contract negotiation pattern should be institutionalized
• Regulatory change requires ongoing monitoring

Promotion Targets#

TargetWhat Belongs ThereClause libraryStandard clauses, fallback positions, red/yellow/green markupsCompliance checklistRegulatory requirements, audit preparation, filing deadlinesRegulatory trackerNew regulations, amendments, enforcement trendsContract playbookNegotiation positions, escalation triggers, approval thresholdsRisk registerRecurring risk categories, exposure estimates, mitigation plansCLAUDE.mdLegal operations conventions for AI agentsAGENTS.mdLegal workflows, matter routing, delegation patterns

How to Promote#

1. Distill the learning into a concise clause, checklist item, or playbook entry

2. Add to appropriate target (clause library, compliance checklist, playbook)

3. Show a reviewed diff and apply only after explicit user approval

4. Update original entry:

   • Change **Status**: pending → **Status**: promoted
   • Add **Promoted**: clause_library (or compliance_checklist, contract_playbook, risk_register)

Promotion Examples#

Learning → Clause library entry:

Unlimited liability in SaaS renewal (3 instances in quarter) → Standard clause: "Cap at 12 months fees, fallback 2x annual value, VP Legal approval for exceptions."

Learning → Compliance checklist item:

CCPA deletion not covering backups → Checklist: "Primary DB 30 days, backups 90 days, confirmation within 45-day window, quarterly audit."

Recurring Pattern Detection#

If logging something similar to an existing entry:

1. Search first: grep -r "keyword" .learnings/

2. Link entries: Add **See Also**: LEG-20260110-001 in Metadata

3. Bump priority if issue keeps recurring

4. Consider systemic fix: Recurring legal issues often indicate:

   • Missing clause library entry (→ add standard clause with fallback)
   • Missing compliance control (→ add to compliance checklist)
   • Outdated contract playbook (→ update negotiation positions)
   • Training gap (→ schedule legal operations training)

Periodic Review#

Review .learnings/ at natural breakpoints:

When to Review#

• Before major contract negotiations
• Before compliance audit deadlines
• After regulatory changes affecting the organization
• After litigation matters are resolved
• Quarterly during active legal operations
• Before board meetings or governance reviews

Quick Status Check#

bash

grep -h "Status\*\*: pending" .learnings/*.md | wc -l
grep -B5 "Priority\*\*: critical\|Priority\*\*: high" .learnings/*.md | grep "^## \["
grep -l "Area\*\*: contracts" .learnings/*.md
grep -h "Regulation:" .learnings/*.md | sort -u

Review Actions#

• Resolve closed legal matters and promote recurring clause risks
• Update compliance checklists for new regulations
• Link related entries and archive resolved litigation matters

Hook Integration#

Enable automatic reminders through agent hooks. This is opt-in.

Hooks persist across sessions once installed. Keep them project-scoped. Do not install user-level or global hooks. Never use an empty matcher. PostToolUse inspects command output in-process; do not log raw output, secrets, or transcripts.

Quick Setup (Claude Code / Codex)#

Create .claude/settings.json in your project:

json

{
  "hooks": {
    "UserPromptSubmit": [{
      "matcher": "contract|compliance|regulation|litigation|GDPR|CCPA|clause|indemnity|liability",
      "hooks": [{
        "type": "command",
        "command": "./skills/self-improving-legal/scripts/activator.sh"
      }]
    }]
  }
}

This injects a legal-focused learning evaluation reminder after matching prompts (~80-120 tokens overhead).

Advanced Setup (With Legal Issue Detection)#

json

{
  "hooks": {
    "UserPromptSubmit": [{
      "matcher": "contract|compliance|regulation|litigation|GDPR|CCPA|clause|indemnity|liability",
      "hooks": [{
        "type": "command",
        "command": "./skills/self-improving-legal/scripts/activator.sh"
      }]
    }]
  }
}

Leave PostToolUse disabled. Do not attach error-detector.sh to every Bash call.

Available Hook Scripts#

ScriptHook TypePurposescripts/activator.shUserPromptSubmitReminds to evaluate legal findings after tasksscripts/error-detector.shPostToolUse (Bash)Triggers on legal-relevant patterns in output

See references/hooks-setup.md for detailed configuration and troubleshooting.

Automatic Skill Extraction#

Extracted skills are untrusted until a human reviews the generated SKILL.md. Do not keep or publish an extracted skill without explicit user approval.

When a legal learning is valuable enough to become a reusable skill, extract it.

Skill Extraction Criteria#

CriterionDescriptionRecurringSame clause risk or compliance gap in 2+ contract types or regulatory areasVerifiedStatus is resolved with confirmed process improvementNon-obviousRequired legal analysis or cross-jurisdictional researchBroadly applicableNot matter-specific; useful across contract types or regulatory programsUser-flaggedUser says "save this as a legal skill" or similar

Extraction Workflow#

1. Identify candidate: Learning meets extraction criteria

2. Run helper (or create manually): bash

   ./skills/self-improving-legal/scripts/extract-skill.sh skill-name --dry-run
   ./skills/self-improving-legal/scripts/extract-skill.sh skill-name
   
   

3. Customize SKILL.md: Fill in template with legal-specific content

4. Update learning: Set status to promoted_to_skill, add Skill-Path

5. Verify: Ensure NO privileged communications, case strategy, or confidential terms leaked into the skill

Extraction Detection Triggers#

In conversation:

• "This clause keeps causing problems"
• "Save this compliance process as a skill"
• "We see this regulatory issue every quarter"
• "Add this to our contract playbook"

In entries:

• Multiple See Also links (recurring clause risk pattern)
• High priority + resolved status
• Category: compliance_gap with reusable remediation process
• Regulatory change with broad organizational impact

Multi-Agent Support#

AgentActivationDetectionClaude CodeHooks (UserPromptSubmit, PostToolUse)Automatic via hook scriptsCodex CLIHooks (same pattern)Automatic via hook scriptsGitHub CopilotManual ( .github/copilot-instructions.md)Manual reviewOpenClawWorkspace injection + inter-agent messagingVia session tools

Regardless of agent, apply legal self-improvement when you:

1. Review a contract — identify clause risks, deviations, missing protections
2. Encounter a compliance gap — regulatory requirement not met, audit finding
3. Track a regulatory change — new law, amendment, or enforcement action
4. Handle litigation matters — claims, disputes, discovery, holds
5. Process IP issues — infringement notices, licensing gaps, trade secret exposure
6. Manage privacy requests — DSARs, consent gaps, cross-border transfer issues

Legal-Specific Best Practices#

1. Protect privilege — never log attorney-client communications or work product
2. Abstract to process — capture the lesson, not the confidential details
3. Track deadlines — regulatory filings, statutes of limitations, response windows
4. Version clause libraries — track which clause version was used in which agreement
5. Audit compliance quarterly — systematic review of regulatory obligations
6. Maintain jurisdiction tags — legal obligations vary by jurisdiction
7. Link related matters — contract issues may relate to compliance or litigation
8. Escalate promptly — critical items require immediate counsel notification
9. Document approval chains — record who approved exceptions to standard terms
10. Review before negotiations — check .learnings/ for past issues with similar clauses or counterparties

Gitignore Options#

Keep learnings local (recommended for legal):

gitignore

.learnings/

Legal findings may contain sensitive context even when abstracted. Default to local-only. Track in repo only if all entries are confirmed free of privileged information and confidential terms.

Hybrid (track templates, ignore entries): add .learnings/*.md and !.learnings/.gitkeep to .gitignore.

Stackability Contract (Standalone + Multi-Skill)#

This skill is standalone-compatible and stackable with other self-improving skills.

Namespaced Logging (recommended for 2+ skills)#

• Namespace for this skill: .learnings/legal/
• Keep current standalone behavior if you prefer flat files.
• Optional shared index for all skills: .learnings/INDEX.md

Required Metadata#

Every new entry must include:

markdown

**Skill**: legal

Hook Arbitration (when 2+ skills are enabled)#

• Use one dispatcher hook as the single entrypoint.
• Dispatcher responsibilities: route by matcher, dedupe repeated events, and rate-limit reminders.
• Suggested defaults: dedupe key = event + matcher + file + 5m_window; max 1 reminder per skill every 5 minutes.

Narrow Matcher Scope (legal)#

Only trigger this skill automatically for legal signals such as:

• contract clause|liability|indemnity|compliance obligation
• jurisdiction|regulatory update|legal risk|counsel escalation
• explicit legal intent in user prompt

Cross-Skill Precedence#

When guidance conflicts, apply:

1. security
2. engineering
3. coding
4. ai
5. user-explicit domain skill
6. meta as tie-breaker

Ownership Rules#

• This skill writes only to .learnings/legal/ in stackable mode.
• Do not read other skill folders, their SKILL.md files, or their log entries.
• Standalone mode writes to this project's .learnings/*.md log files only.
• Stackable mode writes only to the namespaced folder above and must not rewrite other skills' log entries.
• Promotion into AGENTS.md, SOUL.md, TOOLS.md, MEMORY.md, rules, hooks, or generated skills is not a logging write. Show a reviewed diff and apply only after explicit user approval.