Install
$ agentstack add skill-rohasnagpal-legal-ai-skills-opposite-party-reply-drafter ✓ 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
Opposite Party Reply Drafter
What this does
Drafts the opposite party's reply to a consumer complaint: any threshold objection — jurisdiction, limitation, maintainability — raised prominently, then a response to every allegation, admitting, denying with the client's own account, denying for insufficient knowledge, or qualifying, exactly as the client instructs. It structures what the client actually says; it does not invent a defence or a fact the client has not given.
Before you start
The complaint being replied to. The actual text — there is no reply to draft without seeing exactly what was alleged.
The client's actual position on each allegation. What is true, false, or partially true, from the client. This is blocking; the skill structures the client's account, it does not construct a defence from the allegations alone.
Which forum this reply is being filed before. This affects format, procedural convention, and reply-timing rules. Ask; flag if not confirmed.
Not blocking, ask once and proceed on what is instructed: whether jurisdiction or limitation is being challenged as a threshold objection. If so, this should be raised prominently and early rather than folded into the allegation-by-allegation response — ask so it can be structured that way from the start.
Method
1. Read the whole complaint once before drafting anything. A global point — lack of jurisdiction, limitation, non-joinder or mis-joinder of parties, the complaint not being maintainable in the form filed — can dispose of several allegations at once, and drafting allegation-by-allegation on a first pass misses this.
2. Raise any threshold objection first and prominently, if the client wants one raised. Note that raising a threshold objection does not require abandoning a reply on the merits — a reply can do both, reserving the objection while still addressing the allegations.
3. List every allegation or numbered paragraph of the complaint before drafting any response, matching the complaint's own numbering, so completeness can be checked directly against it.
4. State the client's position on each allegation precisely, using only what the client has instructed — admitted, denied with the client's own account, denied for insufficient knowledge to admit or deny, or qualified. Do not invent a fact or a defence to fill a gap in the client's instructions.
5. Where an allegation is denied, give the client's own account of what happened, not a bare denial. A bare denial carries less weight than a denial coupled with an affirmative account.
6. Keep factual admissions separate from legal characterisation. Admitting that an item was delivered five days late is not the same as admitting deficiency in service — concede the fact if it is true, without conceding the legal conclusion the complainant draws from it.
7. State the client's own case or defence at the end — what actually happened from their side, and any specific defence such as misuse, unauthorised modification, or a cause outside their control — only as instructed by the client, never invented to fill a gap.
8. Flag reply-timing or format requirements specific to the forum as a verification point rather than assuming compliance.
Output
1. Header. Forum, complaint reference, opposite party (respondent) and complainant, date.
2. Preliminary objections, if any. Jurisdiction, limitation, or maintainability, stated prominently and first.
3. Reply to allegations. Numbered to match the complaint: admitted / denied with the client's own account / denied for insufficient knowledge / qualified.
4. Client's own case or defence.
5. Relief sought. Dismissal of the complaint, costs, or other relief, only as instructed.
6. Points requiring verification. Forum-specific reply-timing or format rules, and anything resting on an unconfirmed legal characterisation.
Do not
Do not invent a fact or defence the client has not actually given.
Do not concede a legal conclusion — deficiency, unfair trade practice — while admitting an underlying fact. Keep the two separate throughout.
Do not omit a threshold objection the client wants raised, and do not bury it inside the allegation-by-allegation response.
Do not assume reply-timing or format compliance with the forum's rules. Flag it as a point to verify.
Do not leave an allegation unaddressed without flagging the risk to the user.
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.