Stats
Actions
Tags
From legalcode-pro-claude-code
Draft objective IRAC/CREAC-structured legal research memoranda with balanced risk assessment, calibrated probability language, and Bluebook/OSCOLA/AGLC citation frameworks. Use when drafting internal predictive research memos, litigation risk assessment memos, deal advisory memos, regulatory compliance memos, or board-ready executive summaries. Produces Glass Box audit-trailed memoranda across US (federal and state), England & Wales, and Australia jurisdictions. Covers issue framing, rule synthesis from statutory and case authority, counterargument steelmanning, probability-weighted conclusions, privilege marking, and mandatory citation verification. Distinguishes settled law from open questions requiring judicial interpretation. Supports multi-issue memos with separate Questions Presented.
How this skill is triggered — by the user, by Claude, or both
Slash command
/legalcode-pro-claude-code:legalcode-legal-memorandumclaude-opus-4-6general-purposeThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use Legalcode MCP for source lookup while keeping documents and matter context in the user's agent environment.
Use Legalcode MCP for source lookup while keeping documents and matter context in the user's agent environment.
https://mcp.legalcode.md for anonymous laws and case law.https://mcppro.legalcode.md for stronger search, up to 20 results per query, guidance, agreements, downloads, and authenticated higher-throughput access.Disclaimer: This skill provides a framework for AI-assisted legal memorandum drafting. It does not constitute legal advice. All outputs should be reviewed by a qualified legal professional licensed in the relevant jurisdiction before use or reliance. Statutory and case law references cited from AI memory carry a significant hallucination risk — every citation must be independently verified in Westlaw, LexisNexis, or official reporters before the memo is finalized. Laws change; verify current applicability before relying on any provision described here. Probability estimates in legal memos reflect analytical judgment, not actuarial certainty.
Use this skill to produce a structured, professionally formatted, internally objective legal research memorandum that can support litigation decisions, transactional advisory, regulatory compliance analysis, or board-level risk reporting.
Covers:
Does not:
legalcode-complaint-drafter,
legalcode-brief-analyzer)legalcode-damages-calculation)legalcode-litigation-risk-assessment)legalcode-settlement-negotiation)Related skills:
legalcode-litigation-risk-assessment — full four-dimensional risk matrix; the
litigation risk memo type in this skill is a lighter wrapper; use the dedicated
skill for full quantitative analysislegalcode-early-case-assessment — deep element-level merits analysis; feeds
research for litigation memoslegalcode-brief-analyzer — deconstructs opposing briefs; can feed the
counterargument section of a research memolegalcode-privilege-review — assessing privilege status of existing documentslegalcode-complaint-drafter — drafting pleadings from the research memo's
conclusionsThis skill operates across three primary jurisdiction tracks. Identify the controlling jurisdiction in Step 3 before any rule synthesis begins; the analysis framework and citation format then lock to that jurisdiction.
[JURISDICTION-SPECIFIC] For each jurisdiction, localize at minimum:
United States (Federal or State):
England & Wales:
Australia:
[JURISDICTION-SPECIFIC] Other jurisdictions: For matters outside US/E&W/AU, identify the applicable authority hierarchy, citation standard, and privilege framework before proceeding. Flag that local counsel review is mandatory for jurisdictions not covered by this skill's core three tracks.
This skill uses interactive clarification at key decision points. Rather than assuming context, the workflow pauses and asks when:
Use the ⟁ CLARIFY pattern wherever marked below. If the user has already provided the information, skip the question and proceed with the stated information.
Accept the memo request in any of these formats:
If insufficient information is provided to identify the legal question and controlling jurisdiction, prompt the user before proceeding.
⟁ CLARIFY — Before beginning the memo, ask the following questions. Present as structured options. Skip questions already answered by the user's input.
1. Memo type — What output format is needed?
2. Jurisdiction — What law controls?
3. Side and purpose — Who is the memo for?
4. Scope — How many issues?
5. Depth
Before drafting any legal analysis, establish the controlling legal framework.
3a. Identify controlling law:
3b. Authority Hierarchy Map — Document the applicable authority hierarchy:
| Tier | Authority Type | Examples | Effect |
|---|---|---|---|
| Mandatory Primary | Highest applicable court + controlling statutes | SCOTUS, Circuit CA, State Supreme Court; applicable statutes | Binding — must apply |
| Persuasive Primary | Other courts, other jurisdictions | Sister circuits, foreign courts, state courts on federal questions | Persuasive — cite with signal (see, cf., accord) |
| Secondary | Treatises, Restatements, Law Review, Guidance | Restatement (Third), ABA guidance, agency no-action letters | Background only — "see also" citations; not dispositive |
3c. Privilege marking selection:
ATTORNEY-CLIENT PRIVILEGED AND CONFIDENTIALATTORNEY WORK PRODUCT — Prepared in Anticipation of LitigationLEGALLY PRIVILEGED AND CONFIDENTIAL — LEGAL ADVICE PRIVILEGECLIENT LEGAL PRIVILEGE — CONFIDENTIAL LEGAL ADVICEOPINION WORK PRODUCT — DO NOT DISCLOSE4a. Identify all discrete legal issues arising from the matter description. For each issue:
Order issues logically: threshold issues first, then substantive issues, then remedies/damages.
4b. Frame each Question Presented (QP) using the "whether-when-and" technique:
Whether [legal conclusion to be tested], when [legally relevant fact(s)], and [additional dispositive fact(s) or legal element that creates the tension].
Well-framed example (Delaware corporate law):
Whether a Delaware corporation's board of directors breached its fiduciary duty of care under the business judgment rule, when the board approved a $50M acquisition without obtaining a fairness opinion, and the CEO had a disclosed financial interest in the target company.
Poorly-framed (too broad): "Whether the contract is enforceable."
Poorly-framed (begs the question): "Whether the defendant wrongfully terminated the agreement, entitling plaintiff to consequential damages."
Rules for QP framing:
4c. Write a Brief Answer for each QP — 1–3 sentences, no new facts or citations. The Brief Answer should tell the reader the conclusion and the single most important reason for it.
Probably yes. The board's failure to obtain a fairness opinion in a self-interested transaction is unlikely to receive business judgment rule protection under Van Gorkom, and the disclosed conflict does not, on its own, satisfy the entire fairness standard.
Write a focused, objective Statement of Facts. Include only facts that are legally relevant to the Issues — do not narrate the entire matter chronology.
Rules:
Structural options:
For each issue, synthesize the applicable rule from the authority hierarchy established in Step 3.
⟁ CLARIFY — If legalcode-mcp is connected, query it for the primary authority on the legal question before drafting the Rule section. If not connected, proceed with in-memory synthesis and mark all authority with [VERIFY CITATION].
Rule synthesis structure (CREAC):
Rule Statement — State the black-letter rule from the highest applicable authority:
Under Delaware law, directors owe a fiduciary duty of care requiring them to act on an informed basis in good faith and in the honest belief that the action taken is in the best interests of the company. Smith v. Van Gorkom, 488 A.2d 858, 872 (Del. 1985) [VERIFY CITATION]. The business judgment rule presumes compliance with this standard unless plaintiff rebuts the presumption by showing self-dealing, bad faith, or waste. Aronson v. Lewis, 473 A.2d 805, 812 (Del. 1984) [VERIFY CITATION].
Synthesis for multi-element rules — Organize by element:
| Element | Statement | Leading Authority |
|---|---|---|
| 1. [element name] | [rule statement] | [citation] [VERIFY] |
| 2. [element name] | [rule statement] | [citation] [VERIFY] |
| … | … | … |
Handling circuit splits / unsettled questions:
The circuits are split on whether X. The Fifth and Seventh Circuits hold [Y]. See [cite] [VERIFY]; [cite] [VERIFY]. The Third Circuit has adopted a narrower rule. See [cite] [VERIFY]. This memo applies [circuit] precedent as controlling.
Statutory text rule — quote, do not paraphrase:
Section 2(2) of the Unfair Contract Terms Act 1977 provides: "In the case of other loss or damage, a person cannot so exclude or restrict his liability for negligence except in so far as the contract term or notice satisfies the requirement of reasonableness." UCTA 1977, s 2(2) [VERIFY current text].
Recent developments note: For any area of law with significant developments after 2023, explicitly flag:
[Note: This area of law has seen significant recent development. Verify whether any post-[date] legislative amendments, regulations, or appellate decisions affect this analysis. The rule as stated reflects law through [knowledge cutoff — state explicitly].]
Apply each element of the synthesized rule to the client's facts. This is the analytical core of the memo.
⟁ CLARIFY — For each element, confirm whether the relevant facts are established, disputed, or unknown. Conditional analysis is required for disputed or unknown facts.
Core analytical disciplines:
7a. Element-by-element analysis: Address each element in order. For elements clearly satisfied (or not), state briefly and move on. Spend analytical depth on contested elements.
Element 1 — [rule standard]: [client fact]. This element is satisfied / not satisfied / genuinely uncertain because [reason with citation to factual record].
7b. The "because" discipline: Every legal conclusion must be followed by "because" and a factual/legal reason:
The limitation clause is likely unenforceable [because] it purports to exclude liability for personal injury caused by negligence — an exclusion categorically prohibited by UCTA 1977 s 2(1), irrespective of reasonableness. [VERIFY citation]
Never: "The clause is likely unenforceable." (conclusory — no reason given)
7c. Fact-citation discipline: Every factual assertion in the Application must cite to a specific source:
Defendant gave written notice on 14 January 2025 (Letter, 14 Jan. 2025, at 2), which is 47 days after the contractual deadline of 28 November 2024 (Agreement s 12(b)).
7d. Counterargument block (mandatory for every Analysis section):
Counterargument: [Opposing party] will argue that [their best argument], relying on [best authority for their position] [VERIFY citation], which held [what the case held].
Rebuttal: However, [case] is distinguishable because [specific factual or legal distinction]. Additionally, [additional rebuttal if available]. Under [controlling authority], [your conclusion] notwithstanding this contrary argument.
The steelman requirement: Present the opposing argument in its strongest form, not a strawman. If you cannot articulate why the opposing argument is reasonable and supported by authority, the analysis is incomplete.
7e. Conditional analysis for uncertain facts:
If the court finds that [Fact A — plaintiff's version], then [legal consequence A] and [conclusion A with probability]. If instead the court finds [Fact B — defendant's version], then [legal consequence B] and [conclusion B with probability]. The more likely factual finding is [A/B] because [factual reasoning].
7f. Adverse authority treatment (mandatory): Never omit adverse controlling authority. Structure as:
It should be noted that [adverse case] [VERIFY citation] appears to support [opposing position]. That decision held [what it held]. However, [case] is distinguishable because [reason] / has been limited by [subsequent case] [VERIFY] / predates [statutory amendment effective date] [VERIFY]. Even applying [adverse case]'s reasoning, [your conclusion holds / carries increased risk] because [reason].
7g. Classification tiers by memo type:
| Memo Type | Classification System |
|---|---|
| Predictive research | STRONG (>75%) / PROBABLE (55-75%) / UNCERTAIN (40-55%) / ADVERSE (<40%) |
| Litigation risk | HIGH RISK / MEDIUM RISK / LOW RISK (tied to expected value ranges) |
| Deal advisory | CRITICAL / MATERIAL / MINOR (by business impact) |
| Regulatory compliance | COMPLIANT / PARTIAL / NON-COMPLIANT |
| Board summary | HIGH / MEDIUM / LOW (plain English risk rating) |
8a. Probability calibration table — Use this table for all probability language. Never use hedging language ("may," "could") when the analysis supports a directional conclusion.
| Language | Approximate Probability | When to Use |
|---|---|---|
| "Will" / "clearly will" | 90%+ | Settled law, unambiguous application to facts |
| "Should" / "very likely" | 75–90% | Strong authority, clear application, minor risk only |
| "Likely" / "probably" | 60–75% | Good authority, reasonable arguments on both sides, one side stronger |
| "More likely than not" | 55–60% | Marginal probability advantage; opposing argument is substantial |
| "Genuinely uncertain" / "may" | 40–55% | Both outcomes are plausible; result depends on fact-finding or unsettled law |
| "Unlikely" / "probably will not" | 25–40% | Adverse outcome more probable; one viable path to favorable result |
| "Very unlikely" / "should not" | 10–25% | Strong adverse precedent; viable only on exceptional facts |
| "Will not" / "clearly will not" | <10% | Controlling adverse authority; no realistic favorable path |
Calibration discipline: Every probability statement must include the primary reason for that confidence level. Example: "We assess a 65–75% likelihood that the clause is enforceable [because the California courts have consistently applied the blue pencil doctrine to preserve severabie provisions, and this clause contains an explicit severability instruction — however, the 5-year restriction exceeds the 3-year maximum approved in the most recent comparable case, creating meaningful uncertainty]."
8b. Conclusion format by memo type:
Predictive research memo conclusion:
For the reasons set forth above, we conclude that [legal conclusion], with a probability assessment of approximately [X%]. The primary basis for this conclusion is [primary reason]. The primary risk to this conclusion is [adverse authority or uncertain fact]. If [adverse condition], the probability shifts to approximately [Y%].
Litigation risk assessment memo conclusion:
Overall Litigation Risk: [HIGH / MEDIUM / LOW] Exposure range: $[low] – $[high], expected value approximately $[midpoint] Recommended posture: [file/defend/settle at $X / seek mediation] Key uncertainties: [list 2-3 most material unknowns that could shift the analysis]
Deal advisory memo conclusion:
Overall Deal Risk: [CRITICAL / MATERIAL / MINOR] Recommended modifications: [list top 3 in priority order] Acceptable risk if not modified: [yes/no + conditions] Alternative structures considered: [list with relative risk profile]
Regulatory compliance memo conclusion:
Compliance status: [COMPLIANT / PARTIAL / NON-COMPLIANT] Regulatory exposure: [describe risk if non-compliant — penalties, enforcement] Required remediation steps: [prioritized action list with deadlines if applicable] Safe harbor availability: [yes/no + conditions]
Produce a 1–2 page plain-English risk summary derived from the research memo. Strip all citations and legal jargon. Translate every legal risk into a business consequence.
Board summary template:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[ATTORNEY-CLIENT PRIVILEGED AND CONFIDENTIAL]
LEGAL RISK SUMMARY
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
MATTER: [Matter name / description]
DATE: [Date]
PREPARED FOR: [Board / Audit Committee / Executive Team]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
OVERALL RISK RATING: ■ HIGH □ MEDIUM □ LOW
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
RECOMMENDATION:
[One sentence — what the board should approve, decide, or be aware of.
Use business language: "We recommend against proceeding with the proposed
structure because it exposes the company to potential liability of $X–Y."]
KEY FINDINGS:
• [Business consequence in plain English — e.g., "The non-compete clause
is unlikely to be enforced in California, leaving [Business Unit] exposed
to competitive solicitation by departing employees immediately."]
• [Quantified exposure where available — e.g., "Estimated damages exposure
if litigation proceeds: $2.5M–$7M, representing 4–12% of FY25 EBITDA."]
• [Timeline or trigger — e.g., "The statute of limitations expires
[date]. A decision on filing must be made by [earlier date] to allow
for investigation."]
• [Key uncertainty — e.g., "The analysis depends on how the court
characterizes [X]. This is currently unsettled law."]
DECISION REQUIRED:
[What specific authorization, approval, or direction is sought from the board]
RISKS IF NO ACTION TAKEN:
[Plain English consequence of inaction — regulatory penalty, statute of
limitations expiry, default judgment, loss of insurance coverage trigger, etc.]
DETAILED ANALYSIS: See attached legal memorandum dated [date].
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Board summary discipline:
Before delivering any memo, run these quality gates silently and revise if any fail.
Run all five gates before delivering. Fail any gate → revise before output.
| Gate | Rule | Fail Action |
|---|---|---|
| Source | Every legal claim cites a specific statute, regulation, case, or authoritative secondary source | Add citation or mark "[UNVERIFIED]" |
| Format | All citations follow the correct format for the jurisdiction (Bluebook Practitioner / OSCOLA / AGLC) | Fix format |
| Currency | Every cited provision checked for amendments, repeal, or subsequent limiting authority | Flag "[CHECK CURRENCY — [year of cited authority]]" |
| Domain | Analysis stays within the identified controlling jurisdiction; no jurisdiction bleed | Remove or re-flag any jurisdictional bleed |
| Confidence | Uncertainty is explicitly stated, not hidden | Add confidence qualifier or probability statement |
For any issue classified as HIGH RISK or reaching an adverse conclusion for the client, apply this three-pass review before finalizing:
Pass 1 — Legal Chain Integrity: Does the risk conclusion follow logically from the cited authority and the facts as stated? Would a court or regulator actually reach this conclusion on these facts? Is the adverse authority being applied at the correct level of generality?
Pass 2 — Completeness: Have all relevant statutes, regulations, and case law been considered? Is there a safe harbor, exception, or defense that has not been analyzed? Has the most recent authority been checked?
Pass 3 — Challenge (Steelman of Client Position): What is the strongest argument that the client's preferred outcome is correct? Under what circumstances might a reasonable practitioner advise that the risk is manageable? Incorporate this analysis into the counterargument section if not already present.
Every delivered memo must include this verification checklist appended to it, listing every citation that requires independent verification before the memo is finalized:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
MANDATORY VERIFICATION CHECKLIST
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
The following items MUST be independently verified
before this memo is finalized or relied upon:
CITATIONS REQUIRING VERIFICATION:
□ [Case citation 1] — verify in [Westlaw/LexisNexis/
official reporter]; confirm no overruling or adverse
subsequent history
□ [Statute reference 1] — verify current text; confirm
no amendments effective after [date]
□ [Regulation reference 1] — verify current CFR/SI/
Commonwealth Register text; confirm no amendment
or revocation
[… list all citations]
FACTUAL MATTERS REQUIRING CONFIRMATION:
□ [Disputed fact 1 — confirm with client]
□ [Date/deadline 1 — verify against underlying documents]
RECENT DEVELOPMENTS — VERIFY FOR CHANGES AFTER [cutoff date]:
□ [Area of law 1 — check for post-[year] developments]
□ [Agency guidance 1 — check for updated guidance]
SCOPE CONFIRMATION:
□ The attorney supervising this matter has confirmed
that this memo covers all material legal issues.
□ Any issues outside the scope of this memo have been
separately identified and are being addressed.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
The final delivered memo should follow this structure precisely:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[PRIVILEGE MARKING — see Step 3c for applicable marking]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
MEMORANDUM
TO: [Recipient — attorney name / GC / client]
FROM: [AI-Assisted Draft — Requires Attorney Review and Adoption]
DATE: [Date]
RE: [Matter name — brief descriptive title]
[Matter number if applicable]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CONTROLLING JURISDICTION: [Jurisdiction]
CITATION FORMAT: [Bluebook Practitioner / OSCOLA / AGLC]
GOVERNING LAW: [Statute(s) / Common law / Regulatory framework]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
QUESTION(S) PRESENTED
I. Whether [Issue 1 — using "whether-when-and" structure].
II. Whether [Issue 2].
[… additional Issues as needed]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
BRIEF ANSWER(S)
I. [Probable / Likely / Uncertain / etc.] [1–3 sentences
stating the conclusion and primary reason. No citations.
No new facts.]
II. [Answer to Issue II]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
STATEMENT OF FACTS
[Focused, objective narration of legally relevant facts.
Cite to specific documents, depositions, or pleadings for
each factual assertion. Flag disputed facts. Flag
unverified facts as [UNVERIFIED — CONFIRM WITH CLIENT].]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DISCUSSION
I. [Heading restating Issue I in affirmative or question form]
A. [Sub-issue or element heading, if applicable]
[CONCLUSION UP FRONT — CREAC structure]
[State the conclusion on this issue first, then prove it.]
Rule: [Rule statement with citations [VERIFY CITATION]]
Explanation: [How courts have applied the rule — case
illustrations with citations [VERIFY CITATION]]
Application: [Element-by-element application to facts.
"Because" discipline throughout. Fact citations to record.]
Counterargument: [Opposing party will argue... Rebuttal:]
Conclusion: [Restate conclusion with probability calibration
and any conditions.]
B. [Sub-issue B if applicable — same structure]
II. [Heading for Issue II — same structure]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CONCLUSION
[Consolidated conclusion addressing all Issues.
Include overall probability assessment or risk rating.
State the primary recommendation.]
[For litigation risk memos — expected value range and
strategic recommendation.]
[For deal advisory memos — top risk modifications
recommended.]
[For compliance memos — compliance status and
remediation priorities.]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[BOARD EXECUTIVE SUMMARY — if requested; see Step 9 template]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[MANDATORY VERIFICATION CHECKLIST — always appended]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Apply this hierarchy in strict order:
Erie doctrine [VERIFY citation]: In diversity jurisdiction cases, federal courts apply state substantive law. Identify whether the question is governed by state or federal law before citing any authority.
Statutory interpretation canons (apply per circuit's preferred methodology):
Cases (Practitioner):
First citation: Smith v. Jones, 123 F.3d 456, 460 (5th Cir. 1997).
Short form: Smith, 123 F.3d at 461.
Immediately preceding: Id. at 461.
Unpublished: Smith v. Jones, No. 23-1234, slip op. at 3 (5th Cir. Jan. 15, 2025).
Federal statutes:
17 U.S.C. § 101 (2018).
Sherman Antitrust Act, 15 U.S.C. §§ 1–7 (2018).
Federal regulations:
17 C.F.R. § 240.10b-5 (2024).
Signal hierarchy (before citation in Practitioner format):
Internal research memos prepared in anticipation of litigation are opinion work product if they reflect the attorney's mental impressions and legal theories. Near-absolute protection from discovery. Hickman v. Taylor, 329 U.S. 495 (1947) [VERIFY citation]; Fed. R. Civ. P. 26(b)(3).
Waiver risk: Sharing memo with third parties outside the attorney-client relationship can waive privilege. Common interest doctrine preserves privilege when shared with co-defendants under a written common interest agreement.
AI-generated content and work product [VERIFY — actively developing area]: Conservative practice requires supervising attorney to review, annotate, and materially adopt AI-generated analysis before treating it as protected opinion work product. The skill's output should always be identified as "AI-Assisted Draft — Requires Attorney Review and Adoption" in the memo heading.
Apply in order:
Statutory interpretation (E&W):
Privilege in E&W:
Marking: LEGALLY PRIVILEGED AND CONFIDENTIAL — LEGAL ADVICE PRIVILEGE
Cases:
First citation: R v Brown [1994] 1 AC 212 (HL).
Modern neutral: Smith v Jones [2023] EWCA Civ 456, [12].
Short form: Brown (n 3) 215. [n 3 = footnote where full citation appears]
Immediately preceding: ibid 215.
Statutes:
Unfair Contract Terms Act 1977, s 2(1).
Consumer Rights Act 2015, s 62(1).
Statutory instruments:
Consumer Contracts (Information, Cancellation and Additional Charges)
Regulations 2013, SI 2013/3134, reg 9.
EU retained legislation (post-Brexit UK):
UK GDPR, art 6(1)(a). [distinguish from EU GDPR, art 6(1)(a) — [VERIFY
current citation convention in UK practice]]
Footnote structure: First full citation in footnote; subsequent citations
use short form with cross-reference: (n X). Use ibid for immediately
preceding footnote authority.
Federal matters:
Persuasive:
Statutory interpretation (Australia): Apply Acts Interpretation Act 1901 (Cth) s 15AA: "In interpreting a provision of an Act, the interpretation that would best achieve the purpose or object of the Act is to be preferred to each other interpretation." [VERIFY current text]
High Court methodology:
Privilege in Australia: Legal professional privilege / client legal privilege under Evidence Act 1995 (Cth) ss 118–119 [VERIFY]: confidential communications for dominant purpose of legal advice. Esso Australia Resources Ltd v Commissioner of Taxation (1999) 201 CLR 49 [VERIFY] established the dominant purpose test.
Marking: CLIENT LEGAL PRIVILEGE — CONFIDENTIAL LEGAL ADVICE
Cases:
High Court: Mabo v Queensland (No 2) (1992) 175 CLR 1 (HCA).
[round brackets when volume number present]
NSWCA: Smith v Jones [2023] NSWCA 123, [45].
[square brackets when year identifies volume]
Short form: Mabo (No 2) (n 12) 66. [n 12 = footnote number]
Australian statutes (italicized):
Competition and Consumer Act 2010 (Cth) s 18.
Corporations Act 2001 (Cth) s 181(1).
Fair Work Act 2009 (Cth) s 385.
Environmental Planning and Assessment Act 1979 (NSW) s 4.15.
Key AGLC difference from OSCOLA: Statute names are italicized in AGLC. Jurisdiction in round brackets after statute name.
| Level | Range | Meaning | Action |
|---|---|---|---|
| Definite | 0.95–1.0 | Settled law, clear statute, unambiguous application | State with full confidence; no caveat needed |
| High | 0.80–0.94 | Strong authority, minor factual uncertainty only | State with brief caveat ("subject to confirmation of [fact]") |
| Probable | 0.60–0.79 | Good arguments for the conclusion; meaningful contrary arguments | State conclusion with probability calibration and contra-indicators noted |
| Uncertain | 0.40–0.59 | Genuinely uncertain outcome; both sides have substantial arguments | Flag for professional review; present both sides with probability ranges |
| Adverse | 0.0–0.39 | Weak basis; adverse authority controls; favorable outcome requires exceptional facts | Do not assert favorably; clearly flag adverse risk; recommend alternative strategies |
What NOT to do when drafting legal memoranda:
Conclusory analysis without element-by-element work — Writing "the clause is likely unenforceable because it is overbroad" without applying each element of the applicable test to the specific facts. Every conclusion must be earned through analysis, not asserted.
Omitting adverse controlling authority — The most common failure identified by supervising attorneys. An internal research memo that omits adverse authority is not an honest analysis — it is advocacy masquerading as research.
Jurisdiction bleed — Citing California law on a New York matter, applying federal circuit authority to a state court case, or applying English law to a Scottish law issue. Every citation must be verified as being from a court with authority over the controlling jurisdiction.
Paraphrasing statutory text — Summarizing the statute in your own words instead of quoting the operative text verbatim. Paraphrasing introduces error and masks ambiguity that may be central to the analysis.
Citation hallucination (AI-specific) — Generating plausible-sounding but non-existent case citations, incorrect reporter volumes, or fabricated holdings. Every AI-generated citation must be independently verified before delivery. The verification checklist is non-negotiable.
Hedging every sentence — Using "may," "could," or "might" throughout to avoid taking a position. A memo that hedges every conclusion is useless for decision-making. Calibrate probability language using the probability table in Step 8a and take a reasoned position.
The passive voice problem — "Notice was given" instead of "The contractor gave written notice by email at 3:47 PM on 14 January 2025 (Email, 14 Jan. 2025, Exh. C)." Legal analysis requires precision about who did what, when, and how. Passive voice obscures legally material agency.
Issue framing that begs the question — Framing the QP as: "Whether the defendant wrongfully terminated the contract" prejudges the conclusion. The issue should be framed neutrally using the "whether-when-and" technique.
Scope creep — Addressing legal issues not raised in the Questions Presented. Memos should be scoped carefully; addressing unrequested issues creates potential professional responsibility exposure by creating advice the client did not request and may not have wanted.
Outdated authority without verification — Citing a statute that has been amended or a case that has been overruled. For any authority more than 3 years old in a rapidly developing area, always flag [CHECK CURRENCY].
Treating persuasive authority as mandatory — Citing a district court opinion from another circuit as if it controls, or citing a law review article as establishing the rule. Authority tier must be explicitly stated.
String citations without explanation — Citing six cases in a footnote without explaining what each adds. Each citation must have a reason; if you cannot articulate it, cut the citation.
No specific record citation in the Application section — Every factual assertion in the Analysis must cite to a specific document, deposition page, or pleading paragraph. "The contract provides for a 5-year restriction" without citing "Agreement § 12(c)" is sloppy practice.
Board summary that repeats legal jargon — The board executive summary must be completely free of legal terminology and citations. Every sentence must answer "So what?" from a business perspective.
Failing to distinguish the client's facts from cited case facts — Spending two paragraphs reciting the facts of a cited case without explicitly connecting them to the client's facts through comparison or distinction.
No explicit conclusion — Writing analysis that leads to an implicit conclusion without explicitly stating it. Every Question Presented must receive an explicit, stated answer in the Conclusion section.
Opinion letter format for internal memo — Using first-person opinion language ("It is our opinion that") in an internal predictive memo. Opinion letters create professional responsibility obligations and are only appropriate when signed by a partner with explicit reliance qualifications.
Failing to mark privilege — Delivering a legal memo without privilege markings. Legal memos are targets in litigation discovery; every memo should carry the appropriate privilege marking from the moment of creation.
Over-confident AI output — Presenting AI-generated probability assessments as equivalent to calibrated professional legal judgment without disclosing the AI-assisted nature of the draft. The memo heading must always note "AI-Assisted Draft — Requires Attorney Review and Adoption."
No verification checklist — Finalizing a memo without appending the mandatory verification checklist that lists every citation and factual matter requiring independent confirmation. This is the primary safeguard against citation hallucination in AI-assisted drafting.
Apply these quality gates before delivering the memo:
Precision:
Objectivity:
Clarity:
Completeness:
Format integrity:
With legalcode-mcp connected (preferred):
/tmp/legal-memo-research-[matter].md) before draftingWithout legalcode-mcp:
legalcode_mcp: "Not connected"WebSearch / WebFetch (fallback): If legalcode-mcp is unavailable and the matter involves a recent legislative development (post-2023 statutes, new regulations, recent enforcement actions), use WebSearch to confirm the current legal landscape before drafting the Rule section. Cite web sources with full URL and access date; note they require verification against official sources.
Append this audit section to every produced memo for traceability:
glass_box:
skill_name: "legalcode-legal-memorandum"
memo_type: "[Predictive Research / Litigation Risk / Deal Advisory /
Regulatory Compliance / Board Summary]"
controlling_jurisdiction: "[Jurisdiction — court system]"
governing_law: "[Primary statute or doctrine]"
citation_format: "[Bluebook Practitioner / OSCOLA / AGLC]"
privilege_marking: "[Marking applied]"
issues_addressed: "[Number of Questions Presented]"
analysis_depth: "[Standard / Deep]"
legalcode_mcp: "[Connected — N citations verified / Not connected]"
citations_generated: "[N total]"
citations_verified: "[N verified / N requiring independent verification]"
counterarguments_addressed: "[Yes — N counterarguments steelmanned]"
adverse_authority_addressed: "[Yes / None identified — [VERIFY]]"
probability_calibration_applied: "[Yes — probability table used]"
verification_checklist_appended: "[Yes]"
quality_score: "[X]/40"
completeness: "[X]/18 elements"
confidence: "HIGH / MEDIUM / LOW — [rationale]"
limitations:
- "All citations require independent verification — [N] citations flagged"
- "Knowledge cutoff: [date] — recent developments require separate verification"
- "[Any scope limitations, assumptions, or caveats specific to this memo]"
reviewer: "AI-Assisted Draft — requires review and adoption by qualified
legal professional licensed in [jurisdiction]"
This skill covers US, England & Wales, and Australia as primary jurisdiction tracks. For other jurisdictions, it provides the structural framework (IRAC/CREAC, privilege marking, probability calibration) but cannot provide reliable primary authority.
For jurisdictions outside US/UK/AU:
For Canada, Singapore, New Zealand, Ireland, and other common law jurisdictions, the CREAC structure and analytical disciplines apply directly; the citation format and authority hierarchy require localization by counsel admitted in that system.
Created by Legalcode (2026-03-01). Legalcode original synthesis. Research conducted via 2-agent parallel research pipeline covering: IRAC/CREAC structural methodology, US/E&W/AU authority hierarchies, Bluebook 21st/OSCOLA 4th/AGLC 4th citation formats, attorney-client privilege and work product doctrine across three jurisdictions, probability language calibration standards, board executive summary format, AI-specific risks in legal memo drafting, and anti-patterns from legal writing pedagogy and law firm practitioner guidance. No third-party skill inputs. All legal authority marked [VERIFY CITATION] per AI hallucination risk protocol.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub roberthh-is/legalcode-plugin --plugin legalcode-pro-claude-code