Install
$ agentstack add skill-rohasnagpal-legal-ai-skills-compensation-quantifier ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Compensation Quantifier
What this does
Builds a compensation claim as a set of individual heads — cost of the defective good or service, consequential loss, mental agony or harassment, litigation costs, interest, and any other head the claimant wants included — using only the figures actually supplied, with the arithmetic shown for anything calculated rather than given as a flat sum. It does not assess whether the claim succeeds; it quantifies what the claim totals if each head is allowed.
Before you start
The facts and the loss actually suffered, and which heads of claim the user wants included. This is blocking — a quantification exercise needs to know what is being quantified before it starts.
Actual figures for each head. Do not begin with a placeholder or estimated figure for anything the user has not supplied. Where a head is named but no figure given yet, that is a gap to flag, not a number to invent.
Not blocking, ask once and proceed on what is confirmed: the forum or law the claim will be brought under. Certain heads — punitive or exemplary damages, litigation cost recovery, a specific interest rate or basis — are often forum- or law-dependent in whether they are even recoverable. Extract or ask; treat recoverability of any such head as a verification point rather than an assumption.
Method
1. List every head of claim the user wants included before assigning a single figure. Building the list and the figures in the same pass tends to let heads get added ad hoc without the same scrutiny as the ones identified up front.
2. For each head, use only the figure actually supplied. Where a calculation is needed — interest on a principal sum over a period, for instance — show the arithmetic step by step, using only the rate and basis the user has given; do not assume a rate.
3. Separate heads that are precisely quantifiable — a receipt, an invoice, a documented cost — from heads that are inherently a matter of judgment, such as mental agony or harassment. For the latter, do not manufacture a specific figure to make the total look complete. State that the amount is discretionary and ask the user what figure, if any, they want claimed for it.
4. Check for double-counting across heads. Claiming both the cost of replacement and the difference in value for the same defect, for example, can overlap. Flag a potential overlap rather than letting the total absorb it silently.
5. Total the claim, showing every head's individual figure alongside the sum, so the arithmetic is fully auditable rather than presented as a single number to be trusted.
6. Flag any head whose recoverability itself is uncertain or forum-dependent — punitive damages, litigation costs, a specific interest basis — as a verification point. Do not assume a head is recoverable because the user asked for it to be included.
7. Stop at quantification. Do not assess whether the underlying deficiency, defect, or unfair trade practice is actually established — that is a different skill's job. This one quantifies on the assumption the claim succeeds.
Output
1. Header. Claimant, respondent, matter, forum (if known), date.
2. Heads of claim. A table: Head | Basis or description | Figure supplied | Calculation shown (where applicable) | Status — Quantified, Discretionary (needs a figure from the claimant), or Recoverability uncertain.
3. Total claimed. The sum, broken down by head so it can be checked line by line.
4. Discretionary heads. A short list of heads where the amount is a matter of judgment, each flagged for the claimant or counsel to propose a specific figure.
5. Potential overlaps. Any heads that might double-count the same underlying loss.
6. Points requiring verification. Recoverability of specific heads, and the interest rate or basis used, under the applicable law or forum.
Do not
Do not invent a figure the user has not supplied.
Do not assess whether the underlying claim succeeds on the merits. Quantify only.
Do not manufacture a specific number for an inherently discretionary head. Flag it as needing the claimant's own figure.
Do not assume a head of claim is recoverable — punitive damages, litigation costs, a specific interest basis — without flagging it as forum- or law-dependent.
Do not let the total include a double-counted head without flagging the overlap first.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: rohasnagpal
- Source: rohasnagpal/legal-ai-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.