diff auditor
by Shippers
AI coding agents routinely return a diff that answers a slightly different question than the one asked — extra scope, mi
Free
Works with the AI tools you already use
See it in action
You say
Here is a transcript: I want to build an agent that scans my emails for invoices and saves them to Dropbox. Or maybe Google Drive. We need this by Friday. Should act like a bookkeeper.
Your agent does
Skill brief: invoice-saver-agent
Trigger: When an invoice is detected in email. Input: Raw email body and attachments. Output: File saved to cloud storage.
Open questions
- Should we use Dropbox or Google Drive? Needed to fill: Output.
About this skill
The problem
Founder brainstorming often happens in stream-of-consciousness voice memos or messy transcripts. Converting these "rambles" into actionable technical specs manually is time-consuming and leads to "hallucinated" requirements where the dev fills in gaps without asking.
What it does
- Analyzes voice transcripts to detect if the user needs a PRD, a Decision Memo, or a Skill Brief.
- Extracts raw signals including nouns, verbs, constraints, and contradictions from unstructured text.
- Maps missing information to
⚠️ not statedmarkers within the document. - Generates a dedicated "Open Questions" block to surface contradictions and blocking unknowns.
- Enforces strict artifact templates to ensure consistency across product documentation.
Why this beats prompting it yourself
Generic prompts often try to "fix" a founder's contradictions, leading to wrong assumptions and wasted dev cycles. This skill is hardcoded to surface contradictions and gaps as explicit questions rather than quietly patching them with AI-generated fluff.
Use cases
- Turn a 5-minute Loom transcript into a structured PRD with clear acceptance criteria.
- Convert a "should we use X or Y" rambling Slack message into a formal Decision Memo.
- Draft a prompt engineering brief for a new agent based on a founder's verbal description.
- Audit a recorded meeting to identify conflicting requirements between stakeholders.
Known limitations
Not intended for client-facing polished docs or task prioritization. It focuses on raw spec-shaping for internal founder-to-developer handoffs.
How to install
Drop the file into your AI Agent. Works with Claude, Cursor, ChatGPT, and 20+ more.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean today
- Free to download with an account
Frequently Asked Questions
Popular in Workflow Automation

Session Memory
Automated daily kickoff and shutdown routine with structured session logging and context recovery.

Enterprise Automation Engineering Architect
Designs and upgrades business automation systems into modular, reliable, observable, secure, low-maintenance, enterprise-grade workflows.
Parallel-agent dispatcher: safely run N cloud coding agents without merge collisions
Solo builders trying to run 5-10 Claude Code / Codex agents in parallel keep hitting the same wall: worktrees drift, two
Agent-safe database access kit: harden Postgres before you let an AI tool near it
Builders wiring LLM agents to their DB keep discovering the hard way that a chatty tool call can drop rows, exfiltrate P