Change language

I have a tech project!
Zevra School • Legal prompts

20 ChatGPT prompts for lawyers in employment law

(that actually work)

Prompts that are tested, structured and built for the real daily work of an employment lawyer. Not one-liners copied from a LinkedIn thread — working tools, with MCP Factory variants for maximum reliability.

Professional secrecy & AI

No data covered by professional secrecy should be entered into an unsecured generative AI tool. The French bar council (CNB) is explicit: pseudonymize systematically. This is not a precaution, it is an ethical obligation.

🔌

MCP Factory variants

If you subscribe to the MCP Factory: several prompts move up a level when the AI is connected to official French legal sources. The MCP connectors (Légifrance, KALI, JURI, Pappers) let Claude search directly within the legal texts. The prompts concerned are flagged with an 🔌 MCP Factory variant.

Analyze and draft

1

Audit a dismissal letter

You are a lawyer specialized in French employment law, on the employee's side.

Analyze the following dismissal letter. Identify:
- Procedural defects (insufficient statement of grounds, missing mandatory wording, failure to meet statutory deadlines)
- Substantive weaknesses (real and serious cause insufficiently characterized, disproportionate sanction, no attempt at redeployment where applicable)
- Arguments that can be used before the labor tribunal

For each point, cite the applicable articles of the French Labor Code and the case law of the Court of Cassation (social chamber). If you are not certain of a case law reference, flag it explicitly rather than inventing one.

Format: a 3-column table (Point identified | Legal/case law basis | Usable argument).

[Paste the pseudonymized letter]

This prompt works because it assigns a precise role (lawyer on the employee's side — which steers the angle of analysis), imposes a structured output format, and above all contains an explicit anti-hallucination instruction about case law references. Without that last line, ChatGPT will invent ruling numbers with disconcerting confidence.

🔌 MCP Factory variant

The gain: case law references are pulled from Légifrance in real time. No more hallucinated appeal numbers.

Before answering, use the JURI MCP to search recent case law of the social chamber of the Court of Cassation on the keywords related to the dismissal ground identified. Use the Code MCP to verify the articles of the Labor Code you cite. Do not cite any reference you have not verified through these tools.
2

Draft a tailored non-compete clause

You are a lawyer in French employment law. Draft a non-compete clause for a permanent (CDI) employment contract with the following parameters:

- Position: [title]
- Sector: [field of activity]
- Desired geographic scope: [perimeter]
- Desired duration: [duration]
- Financial consideration envisaged: [amount or percentage of salary]

The clause must comply with the validity conditions set by the case law of the Court of Cassation (limitation in time, in space, in light of the specificity of the job, and financial consideration).

Flag if any of the parameters I have indicated risks being deemed disproportionate in light of recent case law. Propose an alternative if necessary.

What makes this prompt valuable is that it doesn't ask the AI to "draft a clause" in a vacuum — it provides the business parameters and asks for critical feedback on their legal viability. The result is a usable draft, not a generic template.

3

Prepare a settlement agreement

You are a lawyer in French employment law, on the employer's side.

Draft a settlement agreement in the following context:
- The employee [pseudonymized role] is contesting their dismissal for [ground]
- The employer wishes to avoid labor tribunal litigation
- The dispute mainly concerns [subject of the disagreement]
- The settlement envelope envisaged is [amount]

The agreement must contain:
1. A reminder of the dispute (without any admission of liability)
2. The mutual concessions (detail them in a way that secures the validity of the settlement within the meaning of article 2044 of the Civil Code and of social case law)
3. The waiver, confidentiality and non-disparagement clauses
4. The payment terms and the social security/tax regime applicable to the settlement payment

Flag the points that require particular vigilance.
4

Analyze an existing employment contract

You are a legal auditor specialized in French employment law. Analyze the following employment contract and produce a structured report identifying:

1. The clauses missing in light of legal and collective-agreement obligations
2. The clauses that are potentially void or unfair
3. The ambiguous clauses that could be interpreted against the employer (or the employee — specify which side)
4. The compliance points with the collective bargaining agreement [name of the CBA if known]

For each point, indicate the risk (low/medium/high) and the associated recommendation.

[Paste the pseudonymized contract]

Labor tribunal litigation

5

Structure an argument for the labor tribunal

You are a lawyer pleading before the labor tribunal (conseil de prud'hommes) in France. My client is [employee/employer].

Here are the facts of the case: [pseudonymized summary of the facts]

Build a structured argument plan for the submissions, distinguishing:
- The points of fact (chronology, material evidence)
- The points of law (articles of the Labor Code, applicable collective bargaining agreement, case law of the Court of Cassation social chamber)
- The quantified claims (severance pay, damages, salary arrears — indicate the calculation method for each)

For each point of law, state whether the case law is settled or whether there is a risk of reversal. Cite only references you are certain of.
🔌 MCP Factory variant

This is the kind of prompt where MCPs change everything: the AI no longer merely "knows" the law — it goes to the source, in real time, and builds its argument on verified texts.

For each point of law, use the JURI MCP to search for the relevant rulings of the social chamber of the Court of Cassation. Use the Code MCP to verify the exact wording of the Labor Code articles invoked. Use the KALI MCP to consult the collective bargaining agreement [IDCC] and identify the applicable contractual provisions.
6

Calculate termination indemnities

You are a legal calculator specialized in French employment law. Calculate the following indemnities based on the parameters below:

- Start date: [date]
- Termination date: [date]
- Average gross monthly salary (last 12 months): [amount]
- Average gross monthly salary (last 3 months, pro-rated bonuses included): [amount]
- Applicable collective bargaining agreement: [CBA]
- Termination ground: [dismissal for real and serious cause / gross misconduct / unfitness / mutually-agreed termination / etc.]
- Age of the employee at the termination date: [age]
- Company size: [headcount]

Calculate:
1. The statutory severance pay (articles L.1234-9 and R.1234-2 of the Labor Code)
2. The collective-agreement indemnity (if more favorable — specify the comparison)
3. The compensatory notice indemnity (duration and amount)
4. The compensatory paid-leave indemnity
5. The applicable Macron scale (floor and ceiling) in the event of dismissal without real and serious cause

Detail each calculation step by step. Flag any assumption you had to make.

This prompt is long, and that's normal. A termination-indemnity calculation depends on a dozen variables — if you provide only three, the AI will fill the blanks with assumptions it won't flag. The result will be wrong and you won't know why.

🔌 MCP Factory variant

Without MCP, the AI guesses the collective-agreement indemnity (and often gets it wrong). With KALI, it reads it straight from the text.

Use the KALI MCP to consult the collective bargaining agreement [IDCC] and extract the provisions on the collective-agreement severance pay and on the notice period. Compare with the statutory indemnity and keep the more favorable one. Cite the articles of the agreement you use.
7

Analyze the compliance of a disciplinary procedure

You are a lawyer in French employment law, on the employee's side.

I'm giving you the chronology of a disciplinary procedure. Verify that the statutory steps were respected:
- The 2-month deadline from the date the employer learned of the facts (L.1332-4)
- The summons to a preliminary interview (content, 5 working days' notice)
- The holding of the interview (the employee's right to be assisted)
- The deadline to notify the sanction (minimum 2 working days, maximum 1 month after the interview)
- The proportionality of the sanction to the alleged facts

Chronology:
[Paste the pseudonymized chronology]

For each step, indicate: compliant / non-compliant / to verify. Flag the usable procedural defects and their legal consequence (nullity, reclassification, damages).
8

Prepare a counter-argument

You are opposing counsel in a labor tribunal dispute in French employment law. You are defending [the other party].

Here are the arguments I plan to develop in my submissions:
[List of arguments]

For each of my arguments, identify:
- The most likely counter-argument
- The case law the opposing party could invoke
- The weak point in my reasoning, if there is one

The goal is for me to anticipate objections and strengthen my submissions before filing. Be rigorous and adversarial.

This is one of the most powerful prompts on this list. Asking the AI to play the role of the opponent forces a contradictory analysis that many lawyers skip for lack of time. The keyword here is "adversarial" — it steers the model toward a critical stance, not a complacent one.

Collective labor law and industrial relations

9

Summarize a collective bargaining agreement

You are a legal expert specialized in French employment law and in collective-agreement law.

Produce an operational summary sheet of the collective bargaining agreement [name and IDCC number] on the following themes:
- Job classification and minimum salary scales
- Working time and working-time arrangement
- Specific leave (beyond the statutory minimum)
- Notice period and severance pay (comparison with the statutory minimum)
- Mandatory provident scheme and health insurance

For each theme, indicate only the provisions more favorable than the Labor Code. Format: one section per theme, with the cited article number of the agreement.

IMPORTANT: if you do not have access to the full text of this collective bargaining agreement in your training data, say so clearly instead of reconstituting approximate provisions.

This prompt illustrates a fundamental rule: on collective bargaining agreements, ChatGPT is often approximate. The final instruction is essential — it forces the model to admit its limits rather than improvise.

🔌 MCP Factory variant

This is probably the prompt where the gap between "with MCP" and "without MCP" is the most spectacular. Without KALI, the AI reconstitutes collective bargaining agreements from memory — with frequent errors. With KALI, it reads the text in force.

Use the KALI MCP to consult the collective bargaining agreement IDCC [number].

Extract the provisions on the following themes:
- Job classification and minimum salary scales
- Working time and working-time arrangement
- Specific leave (beyond the statutory minimum)
- Notice period and severance pay
- Mandatory provident scheme and health insurance

For each theme, compare with the provisions of the Labor Code (use the Code MCP to verify them) and indicate only the agreement provisions that are more favorable. Cite the agreement article number for each point.
10

Prepare a works council (CSE) consultation

You are an advisory lawyer in French employment law, specialized in collective labor relations.

The employer is considering [describe the project: restructuring, change in working conditions, relocation, collective redundancy plan, etc.].

Draft a note for the works council (CSE) that:
1. Identifies the legal basis for the consultation obligation (mandatory or optional information-consultation)
2. Specifies the applicable statutory timeline (consultation deadlines, number of meetings)
3. Lists the documents the employer must provide to the works council
4. Proposes the strategic questions the elected members should ask during the procedure
5. Flags the available means of challenge in the event of an irregular consultation

Context: [describe the pseudonymized situation]
11

Analyze a company-level agreement

You are a legal expert specialized in French employment law. Analyze the following company-level agreement and verify:

- The formal validity (signature conditions: representative unions, 50% threshold or referendum, filing with the DREETS)
- The compliance of the clauses with social public order (the provisions that cannot derogate from the Labor Code even by agreement)
- The clauses potentially less favorable than the industry-wide collective agreement in the areas where the industry prevails (block 1 of article L.2253-1)
- The follow-up and termination clauses

Flag any risk of nullity or unenforceability.

[Paste the pseudonymized text of the agreement]

Advisory work and HR compliance

12

Assess a fixed-term/permanent reclassification risk

You are a lawyer in French employment law. An employer is consulting me on the following situation:

[Describe the situation: succession of fixed-term contracts (CDD), reason for use, waiting period, total duration, etc.]

Analyze the risk of reclassification as a permanent contract (CDI) in light of:
- The authorized grounds for use (L.1242-2)
- Compliance with the maximum durations and the waiting period
- The case law on reclassification in the event of abusive use of fixed-term contracts

Conclude with a risk assessment (low / moderate / high) and an operational recommendation for the employer.
🔌 MCP Factory variant
Use the Code MCP to verify articles L.1242-1 to L.1242-8 of the Labor Code (grounds for use, duration, renewal, waiting period). Use the JURI MCP to search recent rulings of the social chamber on the reclassification of fixed-term contracts as permanent contracts, in particular on the ground for use invoked in this situation.
13

Draft a warning letter

You are a lawyer in French employment law, on the employer's side. Draft a warning letter addressed to an employee based on the following elements:

- Alleged facts: [precise factual description]
- Date(s) of the facts: [dates]
- Consequences of the facts for the company: [impact]
- Any prior disciplinary record: [yes/no, which]

The letter must:
1. State the facts in a precise and detailed manner (no vague wording)
2. Characterize the misconduct without using excessive terms that could be challenged
3. Recall the contractual or regulatory obligations breached
4. Issue the warning, mentioning that it will be added to the file
5. NOT mention any other sanctions contemplated (to preserve proportionality in the event of a repeat offense)

Tone: firm, factual, legally impeccable.
14

Check working-time compliance

You are an auditor in French employment law. A firm is consulting me on the following working-time arrangement:

[Describe: day-rate (forfait jours), overtime, part-time, on-call duty, etc.]

Verify compliance in light of:
- The statutory provisions (maximum durations, mandatory rest periods, overtime quota)
- The collective bargaining agreement [name/IDCC]
- The working-time monitoring and record-keeping obligations (in particular for day-rate arrangements: annual interview, right to disconnect)

Identify the risks in the event of litigation (salary arrears, reclassification of the day-rate arrangement, concealed work).
15

Assess a moral harassment risk

You are a lawyer in French employment law. An employee reports the following situation to me:

[Pseudonymized description of the facts: behaviors, frequency, perpetrator, impact on health]

Analyze these elements in light of the legal definition of moral harassment (L.1152-1 of the Labor Code) and the case law of the social chamber of the Court of Cassation.

Recall the burden-of-proof rules applicable before the labor tribunal (L.1154-1: the employee presents facts suggesting harassment, and the employer must prove that the conduct does not amount to harassment).

Indicate:
1. Whether the facts described are likely to constitute elements suggesting moral harassment
2. The evidence to gather to strengthen the case
3. The recommended actions (alerting the employer, occupational physician, labor inspectorate, referral to the labor tribunal)

Do not minimize the facts and do not rule on their truthfulness — analyze them solely from a legal standpoint.

Client communication and productivity

16

Explain a court ruling in plain language for a client

You are an employment lawyer communicating with your client (a non-lawyer). Rewrite the following ruling in plain language:

[Paste an excerpt of the judgment or ruling]

The summary must:
- Explain who won and who lost
- Specify the amounts awarded (and what they correspond to)
- Indicate the possible avenues of appeal and their deadlines
- Give a clear recommendation (appeal / accept / negotiate)

Tone: accessible, reassuring without being misleading, professional. No unexplained legal jargon.
17

Draft a client letter on a complex situation

You are a lawyer in French employment law. Draft a letter to an employer client setting out the following situation and your recommendations:

Context: [pseudonymized situation]
Legal issue: [question raised]
Analysis: [your analysis in a few points]
Recommendation: [what you advise]

The letter must:
- Be understandable by an SME executive (not a lawyer)
- Present the risks factually (not alarmist, not minimized)
- Propose options with their respective pros and cons
- Conclude with a precise action to take

Length: 1 page maximum.
18

Generate a case law monitoring memo

You are a legal monitoring specialist in French employment law. On the following theme:

[Theme: e.g. day-rate arrangements, Macron scale, work-related unfitness, duty of safety, etc.]

Produce a monitoring memo covering recent case law developments (social chamber of the Court of Cassation). For each ruling:
- Date and appeal number
- Facts summarized in 2 sentences
- The solution adopted by the Court
- The practical scope for lawyers

WARNING: cite only rulings you are absolutely certain exist. If you are unsure of an appeal number, write "reference to verify" next to it. I'd rather have 3 verifiable rulings than 10 invented ones.

The last instruction is critical. This is ChatGPT's Achilles' heel in legal matters: it fabricates ruling numbers with aplomb. A lawyer sanctioned for citing fictitious AI-generated rulings in their submissions — that has already happened, in the United States in 2023, and no one wants to be the first in France.

🔌 MCP Factory variant

With the JURI MCP, the problem of invented rulings disappears: the AI cites only what it actually found in the database.

Use the JURI MCP to search recent rulings of the social chamber of the Court of Cassation on the theme: [theme].

For each ruling found, produce a summary sheet:
- Date, appeal number and formation (social chamber, full court, etc.)
- Facts summarized in 2 sentences
- The solution adopted by the Court
- The practical scope for an employment lawyer

Sort the results by date (most recent to oldest). Limit yourself to the rulings actually returned by the MCP — do not add any from memory.
19

Prepare an in-house training plan

You are a trainer in French employment law. Design a half-day (3h30) training plan for the managers of a [sector] company on the theme:

[Theme: e.g. "The legal risks of day-to-day management" / "Preventing harassment" / "Managing an employee in difficulty" / "Disciplinary law for non-lawyers"]

The plan must include:
1. The learning objectives (3 to 5 maximum)
2. The schedule by segment (duration, content, teaching method)
3. Two case studies drawn from real (pseudonymized) situations
4. The key takeaways (in a "5 reflexes" format)
5. A 10-question quiz to assess learning

Target audience: operational managers, not lawyers. The tone should be concrete, not academic.
20

Build a social compliance checklist

You are a social auditor specialized in French employment law. Produce a compliance checklist for a law firm conducting a social audit of a [size] company in the [sector] sector.

The checklist must cover:
1. Employment contracts (mandatory wording, CBA compliance)
2. Working time (monitoring, rest periods, overtime)
3. Pay (minimum wage, collective-agreement minima, professional equality)
4. Health and safety (up-to-date risk-assessment document, mandatory training, medical follow-up)
5. Employee representation (works council in place if the threshold is reached, up-to-date elections)
6. Mandatory postings and registers
7. Professional training (career interviews, CPF top-up)
8. Data protection (HR GDPR register, informing employees)

For each point: reference text, document to check, risk in the event of non-compliance.

Format: a usable table.

What makes the difference: the principles behind the prompts

If these 20 prompts work better than average, it's not magic. They rest on five principles you can apply to any prompt you build yourself.

1. The role

Every prompt starts by assigning the AI a precise role: "You are a lawyer in French employment law, on the employee's side". This is not decoration. The role steers the angle of analysis, the level of detail and the vocabulary used. A prompt without a role produces a generic answer. A prompt with a precise role produces a targeted answer.

2. The context

The more context you provide, the less the AI improvises. The prompts on this list systematically ask for precise parameters: contract type, collective bargaining agreement, chronology of the facts, amounts. A prompt without context is a question put to a lawyer without handing them the file.

3. The constraints

The best prompts impose constraints: "Cite only references you are certain of", "Flag if a parameter risks being deemed disproportionate". Constraints channel the answer and reduce hallucinations.

4. The output format

A 3-column table, a checklist, a structured plan, a summary note... Imposing an output format turns a block of text into a usable deliverable. It's the difference between an answer you read and an answer you use.

5. The anti-hallucination instruction

This is the most important principle for legal use. "If you are not certain of a reference, flag it explicitly rather than inventing one". Without this instruction, the AI fabricates ruling numbers, collective-agreement articles and doctrinal references with a confidence that makes human review indispensable.

These principles are taught in detail in the Zevra School training programs.

The limits to keep in mind

1. Case law references must be verified systematically

Even with the best anti-hallucination instructions, ChatGPT can invent appeal numbers. Human verification remains indispensable — or the use of tools connected to official databases, such as the MCP Factory, which lets the AI search directly within Légifrance.

2. Collective bargaining agreements are a blind spot for generative AI

Language models have not been trained on the full set of up-to-date French collective bargaining agreements. They reconstitute from memory, with frequent approximations on salary scales, notice periods and collective-agreement indemnities. The KALI MCP connector solves this problem by giving the AI direct access to the collective-agreement text in force.

3. Pseudonymization is not optional

The French bar council (CNB) guide on the use of AI by lawyers is unambiguous: no data covered by professional secrecy should be entered into an unsecured tool. Pseudonymize systematically before pasting anything into a prompt. Names, addresses, social security numbers, identifying amounts — everything must be neutralized.

These 20 prompts are only a starting point. The real productivity gain comes when you build your own prompt library, tailored to your practice, your clientele and your recurring cases. Every prompt tested and refined is one more tool in your arsenal.