Use when the user has a piece of legal-jargon-heavy text and wants it rewritten in plain language for a specified non-legal audience. Common scenarios — email to a business stakeholder, legal disclaimer for a customer-facing page, memo for a non-legal audience, explanation of a contract clause for a deal team, summary of a regulatory development for an executive briefing. Produces the rewritten text plus a brief explanation of what was changed and why, including any preservation-of-meaning concerns where the rewrite required interpretation. Does not draft from scratch; the input is text that exists.
Use when the user has a piece of legal-jargon-heavy text and wants it rewritten in plain language for a specified non-legal audience. Common scenarios — email to a business stakeholder, legal disclaimer for a customer-facing page, memo for a non-legal audience, explanation of a contract clause for a deal team, summary of a regulatory development for an executive briefing. Produces the rewritten text plus a brief explanation of what was changed and why, including any preservation-of-meaning concerns where the rewrite required interpretation. Does not draft from scratch; the input is text that exists.
Rewrite legal text in plain language for a specified non-legal audience. The output preserves meaning while changing form: shorter sentences, common words instead of jargon, active voice instead of passive, concrete examples instead of abstract formulations. The skill is transformation, not drafting from scratch.
This skill is for the everyday in-house counsel task of translating between legal precision and non-legal accessibility. Lawyers spend significant time rewriting their own work — or reviewing others' rewrites — for executive readouts, sales communications, customer disclosures, deal-team briefings, and engineering specs. The skill does this rewriting (or supports the user's rewriting) while flagging where translation involved interpretation.
When this skill applies
Apply when the user has legal-jargon-heavy text and wants it rewritten for a specified non-legal audience. Common inputs:
Email drafts to non-legal business stakeholders.
Contract clauses to be explained to a deal team or product team.
Legal disclaimers to be rewritten for customer-facing surfaces.
Memos to be summarized for executive readouts.
Regulatory developments to be communicated to affected business functions.
Compliance obligations to be translated for the operational owner.
Do not apply when:
The user wants to draft text from scratch. Out of scope — the skill rewrites existing text. Drafting from scratch is a different shape (drafting skill candidate; deferred enhancement).
The user wants to translate between languages (e.g., English to Spanish). Out of scope.
The user wants legal review of someone else's plain-language rewrite. Out of scope; that's review, not rewriting. The user can paste both versions and ask whether the rewrite preserves meaning, but the skill is calibrated for the rewriting task.
The text is already plain language and the user wants a different style. Out of scope; that's editing for style, not rewriting for accessibility.
The user wants to "make this sound more legal" or "add legal weight." Out of scope — the skill goes from legal to plain, not the other direction. Adding legal precision to plain text is a different task with different risks (introducing meaning the original didn't carry).
Privilege
Rewriting a privileged legal memo (e.g., in-house counsel advice, outside-counsel work product, attorney-client communication) for a non-lawyer audience can break privilege. Privilege generally protects confidential lawyer-client communications made for the purpose of giving or receiving legal advice; circulating the substance of that advice to people outside the privileged circle — or reformatting it into a deliverable that will be distributed beyond that circle (executive readouts, sales briefings, customer-facing copy) — may waive the protection. Waiver rules vary by jurisdiction and by the specific scope of any "common interest" or work-product doctrine; loss of privilege can have downstream litigation and disclosure consequences.
Before invoking this skill on a privileged source, the user must confirm that the intended audience falls within the privilege scope (i.e., the recipients are within the privileged circle — typically the client organization's personnel with a need to know for the purpose of receiving legal advice — and the communication will remain confidential within that circle). If the audience falls outside that scope, the responsible attorney must approve the rewrite and the contemplated circulation before the rewrite is produced or shared.
When the source text appears to be a privileged legal memo, attorney work product, or confidential lawyer-client communication, the skill:
Asks the user to confirm the rewrite's audience is within the privilege scope.
If the user cannot confirm, routes back to the responsible attorney rather than producing a rewrite that may be shared outside the privileged circle.
Notes in the output that the rewrite, once circulated to a non-privileged audience, will not retain privilege over the underlying advice and that the original memo should not be attached or quoted verbatim in the distribution.
This treatment is the privilege complement to the "substantive review" disclaimer below: the skill does not assess whether privilege actually exists or whether the proposed circulation waives it — those are substantive legal questions for the responsible attorney. The skill's role is to flag the risk and require user confirmation before transforming privileged content for downstream distribution.
Inputs
The skill requires the text and the audience. If audience is not provided:
"Before I rewrite, who is this for? The audience determines tone, length, and detail level. Examples:
'Executive readout for CEO/CFO — single paragraph'
'Sales team — they need to understand what they can say to prospects'
'Deal team — commercial counsel plus product manager'
'Engineering team — explaining a privacy obligation that affects design'"
Optional inputs (purpose, length_constraint, tone, preserve_specific_terms) refine the output. The most consequential is preserve_specific_terms — without it, the skill may simplify legal terms whose precise wording carries legal weight. The skill flags where simplification occurred so the user can verify.
Workflow
The workflow has three steps.
Step 1: Read the original carefully
Before rewriting:
Identify what the text is doing — informing, advising, requiring, warning, summarizing, etc.
Identify the legal terms of art, defined terms, and regulatory language that may have precise meaning. These are the candidates for preserve_specific_terms treatment.
Identify ambiguities, conditional structures, and qualifications that require interpretation to translate. ("subject to applicable law and regulatory guidance" might mean something specific in context, or might be CYA language that adds little.)
Identify the bottom-line point. Plain-language rewrites lead with the bottom line; legal-jargon-heavy text often buries it.
Use reference/legal_to_plain_patterns.md for common transformations.
Step 2: Rewrite for the audience
Apply transformations appropriate to the audience:
Sentence length: legal writing often has long compound sentences; plain language uses shorter sentences (typically 15-20 words average).
Voice: prefer active over passive ("we must notify the customer" vs. "the customer shall be notified").
Word choice: replace legal terms with everyday equivalents where the precise meaning is preserved (for terms where precise meaning matters, see "preserve specific terms" below).
Concrete over abstract: "if a customer asks for their data, we have 30 days to respond" vs. "data subject access requests must be fulfilled within applicable statutory timeframes."
Direct over indirect: "you can't share customer data with third parties" vs. "third-party data sharing is impermissible absent the requisite authorizations."
Lead with the bottom line: the audience's first read should answer their primary question.
Calibrate the level of transformation to the audience:
Executive briefing audiences want the bottom line in one sentence and the reasoning in two-three more. Detail goes in an appendix or follow-up if requested.
Operational owners (sales, engineering, support) want enough detail to act on the obligation. Include the specific actions and the boundary conditions ("yes you can do X; no you can't do Y; if you encounter Z, escalate to legal").
Customer-facing audiences want the substance without the legalese. Use second person ("you can request your data") and direct verbs ("we delete your data within 30 days of your request").
Deal-team audiences (commercial counsel, product) often need the legal substance preserved but in efficient prose. Less compression than executive briefing, more compression than the original legal text.
When preserve_specific_terms is provided, those terms appear verbatim in the rewrite. Without preserve_specific_terms, the skill applies judgment about which terms to preserve and flags choices in the explanation.
Step 3: Produce the rewrite plus explanation
Produce the rewritten text first, followed by a brief explanation of what was changed and why. The explanation:
Identifies the structural changes (sentence length, ordering, lead with bottom line, etc.).
Notes any terms that were simplified and confirms whether the simplification preserved meaning.
Flags any preservation-of-meaning concerns — places where the original was ambiguous, where simplification required interpretation, or where the user should verify the rewrite captures their intent.
Suggests follow-up rewrites if the audience pushes back or asks questions.
The explanation should be substantially shorter than the rewrite. The rewrite is the deliverable; the explanation is the show-your-work.
Output
Produce the output in markdown with this structure:
# Comms Improver: [Brief description of the text]**Audience:** [from input]
**Purpose:** [from input, or inferred]
**Length constraint:** [from input, or "matched to original"]
**Tone:** [from input, or "neutral businesslike"]
## Rewritten text
[The rewrite. Set off as a code block or quote block so the user can copy it directly. Multiple variants if the user wants alternatives — e.g., "one-sentence version" and "one-paragraph version" for executive briefing audiences.]
## What was changed
[Brief explanation of structural changes — sentence shortening, voice changes, ordering, etc. 2-4 bullets typical.]
## Terms simplified or preserved
[For terms that have legal weight: which were preserved verbatim, which were simplified and why. If `preserve_specific_terms` was provided, confirm those terms appear verbatim. If not provided, flag which terms might need preservation depending on the use case.]
## Preservation-of-meaning concerns
[Any places where the rewrite required interpretation, where the original was ambiguous, or where the user should verify the rewrite captures intent. If none, this section can be a single sentence: "No preservation-of-meaning concerns; the rewrite is a straightforward transformation of the original."]
## Suggested follow-ups
[Brief — typically 1-3 items. What to do if the audience pushes back, how to handle common follow-up questions, whether a longer or shorter version might be useful.]
The output should be proportional to the input. A one-paragraph original produces a brief output (rewrite plus a few-sentence explanation). A multi-page memo produces a longer output. Don't pad short rewrites; don't truncate substantial ones.
Edge cases and refusals
Original text is itself unclear or ambiguous. The skill rewrites what's there, flags the ambiguity, and notes that the rewrite cannot be clearer than the source. If the user wants disambiguation, that requires substantive understanding of the source's intent — the skill can ask, or recommend the user clarify the original first.
Audience is too vague to calibrate ("just make it simpler"). The skill produces a default audience calibration (neutral businesslike, single-paragraph, mid-detail) and notes the assumption.
Original text is very short (a sentence). The skill rewrites and notes that for short text the explanation may be longer than the rewrite — that's appropriate; the explanation surfaces what was preserved.
Original text contains specific legal language the user is going to send to a lawyer counterparty. Plain-language rewrite may be inappropriate; the original may have been drafted with that audience in mind. The skill notes this concern and asks the user to confirm the audience.
Original text contains text the user shouldn't be sending at all (legally inadvisable, factually inaccurate, or exposing the user's organization to risk). The skill rewrites for plain language but flags substantive concerns separately. The skill is not a substantive review; if substantive concerns are visible, they're flagged but the user is responsible for substantive review.
Audience is hostile or adversarial (counterparty in a dispute, regulator, opposing counsel). Plain-language rewrites for adversarial audiences require careful handling — simplification can lose hedging that's protective. The skill notes this consideration and recommends user verification.
Source text is a privileged legal memo, attorney work product, or confidential lawyer-client communication. Rewriting privileged content for a non-lawyer audience may waive privilege depending on jurisdiction and on whether the audience falls within the privileged circle. The skill asks the user to confirm the intended audience falls within privilege scope; if the user cannot confirm, the skill routes back to the responsible attorney rather than producing the rewrite. See the "Privilege" section above for the full treatment.
What this skill does not do
Draft text from scratch. The skill rewrites existing text.
Substantive review. Plain-language rewrite is a transformation, not a substantive check on whether the underlying advice is correct.
Translation between languages. Out of scope.
Style editing for already-plain text. Out of scope.
Generate "more legal" versions of plain text. Out of scope; that's a different skill with different risks.
Multiple-audience optimization in a single rewrite. Each rewrite targets one audience. If the user has multiple audiences, run the skill multiple times.
Reference materials
reference/legal_to_plain_patterns.md — common transformations from legal jargon to plain language.
reference/audience_calibration.md — how to calibrate tone, length, and detail for different audience types.
examples/example_executive_briefing.md — worked example: contract concern translated for CEO/CFO.
examples/example_sales_team.md — worked example: regulatory restriction translated for sales team.
examples/example_customer_disclaimer.md — worked example: legal disclaimer rewritten for customer-facing page.
QA Remediation (LegalQuants, 2026-05)
This skill was reviewed against the Legal Skill Design Framework on 2026-05-11 (verdict: SOME CONCERN). The principal gap identified was that the skill did not address privilege risk when rewriting privileged legal memos for non-lawyer audiences — a textbook waiver scenario for the executive-briefing, sales-team, and customer-disclaimer worked examples.
This remediation adds:
A "Privilege" section to scope and legal use, requiring the user to confirm the intended audience falls within the privilege scope before the skill is invoked on a privileged source. When the source appears to be a privileged legal memo, attorney work product, or confidential lawyer-client communication, the skill routes back to the responsible attorney if the user cannot confirm audience scope.
A privilege failure-mode bullet in "Edge cases and refusals" cross-referencing the Privilege section.
version, last_reviewed, and last_reviewed_by fields in the frontmatter to surface the review cadence for forks and downstream users.
Technical content (inputs, workflow, output format, transformations, audience calibration) is unchanged. The skill remains a bounded text-in / text-out transformation; the remediation tightens the legal-use scope rather than the technical surface.
Other QA observations (work-shape labelling, confidence bands, consolidated escalation triggers, delegation-threshold tightening) are recorded for a future revision and are not addressed in this targeted remediation.
