# Legal Risk Assessor > Sets out the realistic options on a decision a client faces, with the downside risk, the likely outcome, and the best case of each — to inform the client's decision, not to make it for them. Use this whenever a user is weighing options and wants the risk of each worked through — including phrasings like "what are our options here and how risky is each", "should we settle or litigate, what's the d… - **Type:** Skill - **Install:** `agentstack add skill-rohasnagpal-legal-ai-skills-legal-risk-assessor` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [rohasnagpal](https://agentstack.voostack.com/s/rohasnagpal) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [rohasnagpal](https://github.com/rohasnagpal) - **Source:** https://github.com/rohasnagpal/legal-ai-skills/tree/main/plugins/advisory/skills/legal-risk-assessor ## Install ```sh agentstack add skill-rohasnagpal-legal-ai-skills-legal-risk-assessor ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # Legal Risk Assessor ## What this does Takes a decision a client faces and sets out the realistic options, what could go wrong with each and how likely and severe that is, and the realistic best and likely case alongside the downside — organised around what the client has actually said matters to them. It supports the client's decision; it does not make it. Where the analysis depends on a legal question — how likely a claim is to succeed, whether a position would hold up — it grounds that in supplied authority or flags it as needing research, rather than asserting a probability from memory. ## Before you start **The decision and the options actually on the table.** Stated by the user, not invented. If the options have not been clearly identified, ask the user to state what is realistically available before drafting anything — including whether "do nothing" or maintaining the status quo is genuinely one of them, since it is often implicitly excluded without being ruled out on the merits. **What matters to the client.** Cost, time, certainty, reputation, preserving the relationship, precedent for future disputes — whichever mix actually applies. This is blocking: risk is not one-dimensional, and the same option can be low financial risk and high reputational risk at once. Which matters more is the client's call, not an assumption to make on their behalf. Not blocking, ask once and proceed on what is available: **governing law**, where part of the analysis turns on a legal question (likelihood of success on a claim, enforceability of a position). Extract or ask; ground any such point in supplied authority or documents, or flag it as requiring research rather than stating a probability from memory. ## Method **1. State the decision and the realistic options precisely**, including the status quo or "do nothing" option where it genuinely applies. Do not silently narrow the option set to the ones that seem obviously reasonable. **2. For each option, identify what could go wrong** — the downside scenarios — how likely each realistically is, and how severe the consequence would be if it happened. Use qualitative likelihood (unlikely, possible, likely) unless the user has supplied real data to quantify it; do not manufacture a percentage that implies more precision than the analysis actually has. **3. For each option, identify the best and likely case, not only the downside.** A risk assessment that only catalogues what could go wrong leaves the client without the realistic range they need to actually weigh the decision. **4. Where an element of the analysis rests on a legal question**, ground it in supplied authority or documents, or state plainly that it requires legal research and give only the qualitative reasoning available without it. Do not state a probability of legal success as settled fact from memory. **5. Map each option against what the client said matters to them from Before you start** — cost, time, certainty, relationship, reputation — so the comparison is organised around the client's actual priorities rather than a generic checklist that may not reflect them. **6. State which option or options the analysis favours given those stated priorities, framed as input to the client's decision, not as an instruction.** The client decides; this skill's job ends at giving them a clear basis to do so. **7. Flag where missing information would materially change the assessment** — "if the assumption in the contract's force majeure clause is wrong, this changes considerably" — rather than presenting the assessment as more settled than the available facts support. ## Output **1. The decision and the options.** Stated precisely, including the status quo where it applies. **2. Options table.** Option | Downside scenarios and likelihood | Best/likely case | Fit with the client's stated priorities. **3. Comparative view.** How the options stack up against what the client said matters to them, in prose. **4. Observation.** Which option or options the analysis favours, explicitly framed as input to the client's decision rather than a directive. **5. What would change this.** The information gaps that would materially alter the assessment if resolved differently than assumed. **6. Points requiring verification.** Any legal-probability question resting on law not yet sourced or confirmed. ## Do not Do not make the decision for the client. Frame every conclusion as input to their choice, not an instruction. Do not present only the downside of each option. Include the realistic best and likely case alongside it. Do not state a probability of legal success, or that a position would hold up, as settled fact without grounding it in supplied authority — flag it as needing research instead. Do not assume the client's priorities. Ask, and organise the comparison around what they actually said matters. Do not silently exclude the status quo or "do nothing" option where it is realistically available. Do not manufacture a precise percentage or figure where the analysis only supports a qualitative likelihood. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [rohasnagpal](https://github.com/rohasnagpal) - **Source:** [rohasnagpal/legal-ai-skills](https://github.com/rohasnagpal/legal-ai-skills) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-rohasnagpal-legal-ai-skills-legal-risk-assessor - Seller: https://agentstack.voostack.com/s/rohasnagpal - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.