Install
$ agentstack add skill-legalquants-lq-skills-uk-witness-statement-review ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
uk-witness-statement-review
When to Use
- A user has a draft witness statement for England and Wales civil proceedings.
- The statement needs checking against documents, chronology, pleadings, or disclosure.
- The user wants issues flagged before signing, service, exchange, or filing.
- The draft may contain argument, speculation, hearsay, unsupported recollection, or inconsistent dates.
This skill is a review workflow. It does not certify truth, settle admissibility, or replace solicitor/counsel sign-off.
Audience and Work Shape
Audience: England and Wales litigation lawyers, pupils/trainees, and paralegals reviewing witness evidence under supervision before service or filing.
Work shape: pattern-matched review with accretive judgment. The skill classifies paragraph types, flags support/rule issues, and prepares questions for the responsible lawyer/witness; it does not certify the evidence.
Legal Failure Modes
- Legal support, not legal advice: the review flags drafting, evidential, and procedural issues; it does not certify truth, admissibility, or compliance.
- Privilege/confidentiality: witness drafts, exhibits, and reviewer comments may be privileged or confidential. Use approved environments and keep outputs within the privilege group unless a responsible lawyer approves circulation.
- Accountability: the witness owns the truth of the evidence, and the responsible lawyer owns service, filing, and strategic use.
Access Modes
This skill works in three modes:
- Live source mode - use browser, web search, MCP, API, or other configured access to retrieve public rules, orders, filings, or source documents.
- User-supplied source mode - use the draft statement plus uploaded or pasted exhibits, pleadings, correspondence, chronology, disclosure, prior statements, rules, or orders supplied by the user.
- No-source mode - review the statement's internal drafting issues and produce a source-request checklist, but do not mark factual assertions as supported.
If exhibits, documents, rules, or orders cannot be retrieved or supplied, mark the relevant check source_missing or verify_current_rule. Do not verify facts or procedural compliance from model memory.
How It Works
1. Confirm context
Identify:
- Court or tribunal, claim type, and procedural stage.
- Witness role: party, employee, expert-adjacent fact witness, third party, or litigant in person.
- Purpose: interim application, trial, summary judgment, costs, or pre-action evidence.
- Governing directions, page limits, exhibit rules, or court-specific order.
- Source set: pleadings, chronology, exhibits, correspondence, disclosure, and prior statements.
If the governing order or forum is unknown, flag CPR/PD points as verify_current_rule.
If the forum is a tribunal or specialist court not governed by the CPR witness statement regime, identify the applicable procedural rules and mark CPR/PD checks as not_applicable_or_verify.
2. Build a statement map
For each paragraph, classify:
first-hand facthearsay or reported factbelief or inferenceargument or submissionlegal conclusionprocedural backgroundexhibit referenceneeds_source
Keep paragraph numbers stable. The output should let a drafter jump straight to the problem paragraph.
3. Check source support
For each factual assertion:
- Tie it to a source document, exhibit, correspondence item, chronology entry, or witness knowledge basis.
- Quote or pinpoint the supporting material.
- Mark unsupported assertions as
needs_source. - Mark document conflicts as
source_conflict. - Preserve uncertainty: a witness can say what they recall, but the review must show where documents differ.
4. Check drafting and compliance risks
Flag:
- Argument dressed as evidence.
- Legal submissions or conclusions better left to skeleton argument.
- Speculation about another person's state of mind.
- Hearsay that needs a clear source or notice analysis.
- Collective phrases such as "we knew" without identifying who knew what.
- Dates, amounts, names, and defined terms inconsistent with documents or pleadings.
- Missing or vague exhibit references.
- Overlong narrative, irrelevant background, or duplicated chronology.
- PD 57AC trial witness statement requirements, certificate of compliance, and confirmation of compliance where applicable; otherwise mark
not_applicable_or_verify. - Statement of truth, signature, date, interpreter, translation, or exhibit formatting issues where applicable.
Do not invent procedural requirements from memory. If current CPR/PD wording matters, retrieve it or mark verify_current_rule.
Confidence Bands
- High: issue is directly supported by the statement text and supplied exhibit/source.
- Medium: issue is visible on the draft but depends on current rule, order, or witness confirmation.
- Low: source is missing, forum/rule status is unclear, or the point turns on admissibility/weight.
Escalation
Stop and route to the responsible lawyer when evidence is hearsay-critical, a source conflict affects a material fact, PD 57AC or a court order is unclear, the statement contains argument or legal conclusions that drive the case, or the user asks whether the statement is ready to sign/file.
5. Produce a review report
Recommended output:
| Paragraph | Issue | Severity | Source / Pinpoint | Rule / Source Checked | Suggested Fix | |---|---|---|---|---|
Severity values:
critical- likely service/signing/accuracy problem.major- material evidential or drafting risk.minor- clarity, formatting, or housekeeping issue.verify- rule or source needs checking.
Then include:
- Key inconsistencies with documents.
- Unsupported factual assertions.
- Argument/legal-conclusion paragraphs.
- Exhibit and statement-of-truth checklist.
- Questions for the witness or solicitor.
Example
Review this draft witness statement against the documents and chronology. Flag unsupported paragraphs, argument, hearsay, date conflicts, and CPR/PD compliance issues. Do not rewrite the statement unless I ask.
For a compact output pattern, see examples/output.md. For paragraph classifications and witness-evidence review axes, see references/witness-review-playbook.md.
Limitations
- The witness, not the agent, owns the truth of the evidence.
- Current CPR, Practice Direction, court order, and tribunal rules must be checked where compliance matters.
- Privileged comments and drafts require an approved processing environment.
- This skill flags evidential risk; it does not make final admissibility rulings.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: LegalQuants
- Source: LegalQuants/lq-skills
- License: Apache-2.0
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.