This skill stress-tests a proposed settlement position before an offer goes out or comes back: the assumptions it depends on, your leverage and the opponent’s, the evidential weaknesses, the likely opponent response, and the timing and costs pressures around it. It structures settlement judgment for a better-informed decision; it does not advise whether to settle.
SkillScore73Maintained
Automatic
◆Legal & Contracts pack
Last verified 8/16/2026
What it changes
Passed 16 of 22 test cases with the skill, and 8 of 22 without it.
Without skill36%
With skill73%
Checks it passed that the model alone missed
Refrains from redrafting the offer letter text in section 12
Limits section 1 (Settlement position in one paragraph) to a single paragraph
Classifies each litigation risk in section 6 using low, medium, high, or unclear
and 9 more in the full report
10 fixed6 passed either way4 still failing2 broke
Pass-rate lift
+36 pp
Tokens
+92%
Agent turns
0%
Response time
+30%
Model gemini-3.6-flashMeasured 8/16/2026v116 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 10
Prompt
We represent a UK-based software supplier in a dispute over late delivery and allegedly defective code. The claimant customer demands £600,000 in damages. We are considering making an informal settlement offer of £180,000 to resolve all claims before expert reports are formally exchanged. The draft offer letter states: 'Our client offers £180,000 in full and final settlement of all claims, subject to formal contract, without prejudice save as to costs.' Please conduct a structured pressure test on our proposed offer covering: 1) position summary, 2) strength of case assessment, 3) key assumptions and failure impacts, 4) our leverage, 5) opponent leverage, 6) litigation risk factors, 7) BATNA/WATNA/MLATNA, 8) whether the offer is aggressive, conservative, or defensible, 9) anticipated opponent response, 10) timing and procedural pressures, 11) costs/proportionality, 12) risks in our draft wording, 13) essential pre-offer questions, 14) prioritized action items, and 15) a concise strategy summary for our internal case file.
What was checked, without and with the skill
Checked
Without skill
With skill
Classifies overall case strength in section 2 using one of the bolded terms: strong, moderately strong, mixed, weak, or unclear on the information provided
Missed
Passed
Classifies each litigation risk in section 6 using low, medium, high, or unclear
Missed
Passed
Prioritises recommended next steps in section 14 using urgent, important, or optional/later
Missed
Passed
case-01 · gemini-3.6-flash · 8/16/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md215 linessingle file
Settlement Pressure Tester
Purpose
Settlement positions often sound commercial, but they still rest on assumptions
about proof, leverage, timing, and risk. A number that feels reasonable can be
built on a liability theory the evidence will not carry, or a confidence about the
opponent's appetite that has never been tested. This skill stress-tests a proposed
settlement position so the team can see what it actually depends on before an
offer goes out or comes back.
You are acting as an experienced litigation solicitor advising internally. Your
job is to find the pressure points, not to cheerlead.
Core discipline:
Do not simply summarise the position back.
Do not assume the user's case is as strong as presented.
Do not assume settlement is either desirable or undesirable.
Do not invent facts, evidence, procedural steps, quantum, costs, dates, or
opponent motivations.
Where something is speculative, label it speculative.
Where information is missing, say what is missing before giving firm
conclusions.
Be candid about weakness. A pressure test that flatters the position is useless —
its whole value is telling the team what they would rather not hear before the
opponent says it for them.
When the user has not given you enough
If you cannot test the position meaningfully on what is provided, open by asking
for it: