Workspace/Skills/Contract Clause Reviewer
production
Catalog/Legal Ops/Tier 1

Contract Clause Reviewer

T1Production
Reviews vendor contracts for non-standard clauses and flags departures from the master playbook.
Lifecycle
Repository
wh/legal-skills
github
Reference
v2.4.0-rc.2
8a31cf2 · main
Evaluation
248 cases · 6 flagged
Baseline 91.4 → cand 92.0
Approval
2 of 3 approved
awaiting compliance
Release
Production · v2.3.7
Pinned 2d ago
Evaluation6 regressions
Versions18
Access
Activity
Candidate introduces 2 regressions and 4 improvements over baseline.
Eval set: contract-corpus-v9.eval · 248 cases · ran 58m ago · 24s
Baseline
production
v2.3.7shipped 6d ago
Overall
91.4
Pass rate
91.3%
Latency p95
2.8s
{ "summary": "Standard 3-year SaaS NDA. Clauses align with master playbook v6.", "risk_flags": [], "recommend": "approve" }
Candidate
awaiting approval
v2.4.0-rc.2by ari.chen · 1h ago
Overall
92.0 0.6
Pass rate
93.5% 2.2%
Latency p95
3.1s 0.3s
{ "summary": "Standard 3-year SaaS NDA with non-standard mutual indemnity clause (§7.4).", "risk_flags": [ "non-standard indemnity scope — deviates from master playbook v6", "auto-renewal language uses 60-day instead of 30-day notice" ], "recommend": "review" }
Rubric breakdown
baseline candidate
Clause extraction precision
0.91(0.03)
Tone compliance
0.95(0.02)
Playbook adherence
0.93(+0.05)
Risk flag recall
0.89(+0.08)
First-pass acceptance
0.82(+0.08)
Edit distance reduction
0.71(+0.09)
Flagged cases
6 regressions
CaseDescriptionRubricBaselineCandidateΔ
#0117Mutual NDA with non-standard indemnityClause extraction precision1.000.5545.0
#0142MSA with custom termination clauseClause extraction precision0.940.7123.0
#0089Vendor agreement with auto-renewal languageTone compliance0.970.889.0
#0203Procurement DPA — schedule 3 omittedRisk flag recall0.810.747.0
#0226Non-standard governing lawClause extraction precision0.900.7911.0
#0234Indemnity carve-out for IP infringementTone compliance0.960.915.0
Approval timeline
awaiting compliance
ari.chen submitted candidate v2.4.0-rc.2
Skill owner
1h ago
eval-runner completed eval run · 248 cases · 6 regressions flagged
Automation
58m ago
kalia.b left 3 comments on rubric breakdown
Peer reviewer
44m ago
sasha.gw approval — security review
Security
30m ago
compliance awaiting approval — clause-extraction regression review
Compliance
now
Required approvals
2 of 3
Skill owner
ari.chen
Approved1h ago
Security
sasha.gw
Approved30m ago
Compliance
Awaiting reviewer assignment
Pending
Reviewer notes
2
KAkalia.b39m ago
The clause-extraction regression looks contained to non-standard NDAs. Worth narrowing the new rule to NDA templates only?
SAsasha.gw30m ago
Security review approved. Audit log additions cover the new tool-call paths.