- Home
- Skills
- Marketing & SEO
- QuoteRound — Source-Backed Customer Quote Reviews
More screenshots
Works with the AI Agents you already use
QuoteRound — Source-Backed Customer Quote Reviews
Turn a supplied interview and proposed customer quotes into a finished review packet your customer can return to your agent.
$9
QuoteRound — Source-Backed Customer Quote Reviews
Example session with this skill installed
Prepare these proposed quotes from our fictional Cedar interview for customer wording review. Include the supplied pilot context, flag the unsupported 40% claim in the author notes and keep the internal budget line out of the client packet. Do not send or publish anything. Give us a review page the customer can return to our agent.
- Read your context and instructions
- Compiled the quoteround
- Generated the data export
Fictional maintainer demonstration
No customer was contacted and nothing was published. The included Cedar interview and plan produced this complete review handoff:
| ID | Proposed wording | Type | Source context |
|---|---|---|---|
| Q1 | In our three-week pilot, the queue felt easier to manage. | Original wording | Lines 3–6: Qualitative observation from a four-person, three-week pilot. No measured saving. |
| Q2 | Seeing the next ticket without asking around helped me stay focused. | Edited proposal | Lines 7–7: Proposed wording shortens a personal experience; reviewer should check that meaning is retained. |
| Q3 | We need a longer trial before making that decision. | Original wording | Lines 8–9: The rollout decision remained open. This limits the story; do not imply company-wide adoption. |
Author-only held item H1: “We cut handling time by 40%.” Held because no measurement was supplied; the transcript explicitly says time saved was not measured. The unrelated internal budget note is absent from the client HTML. Q2 shows its exact original quotation next to the proposed edit; Q1 retains the nearby pilot and measurement qualifications.
Built client-review.html, printable-review.html, packet.json, author-ledger.json and held-items.csv. The HTML lets the reviewer choose wording feedback for each quote and export a reply; the static fallback supports manual return.
Actual browser reply and reconciled result
A simulated reviewer used the browser page and downloaded its reply:
| ID | Wording feedback | Suggested replacement | Comment / next step |
|---|---|---|---|
| Q1 | wording-ok | None | Response recorded; publication permission remains separate. |
| Q2 | suggest-edit | I could see the next ticket without asking around. | Please keep this closer to my original wording. Review the replacement and create a new wording round. |
| Q3 | do-not-use | None | Exclude this quotation from the proposed use. |
Respondent: Fictional Cedar reviewer (self-reported).
Packet: e7fd6c3bc0fe02b3149e0de8d36998ac3d164cf306df840c208ccf36fe0ec199
Generated response.html, response-ledger.json and drafting-handoff.csv. The CSV carries the packet ID and self-reported reviewer as well as each decision. The lossless JSON preserves exact reviewed/replacement wording and records identity_verified: false and publication_cleared: false.
A separate changed-packet test rejected the old reply with no output directory. An independent fresh fictional buyer run also exercised changed wording and held unsupported/restricted/wrong-speaker candidates. These are local synthetic/browser checks, not customer approval, consent, time-saving, adoption or income evidence. Browser qualification used localhost; direct file-URL opening and recipients’ corporate browser policies were not tested.
client-review.html
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Turn a supplied interview and proposed customer quotes into a finished review packet your customer can return to your agent.
QuoteRound puts the exact proposed wording, attribution and selected transcript context side by side. The recipient can mark Wording OK, suggest a replacement, choose Do not use or leave an item unanswered. They download a reply JSON or copy its contents and return it through your usual channel. Your agent reconciles that reply against the exact packet to produce a readable response ledger and drafting handoff.
Bring: readable supplied transcript text, proposed quotes/attribution and an intended-use note. Tell your connected agent what you need; it prepares the structured inputs for you. Source files stay unchanged. PDF/Word/audio extraction is not built into the helper; supply readable text or use your agent's existing extraction tools and review any gaps.
Get: a self-contained client-review.html; a printable review fallback; an author ledger with held candidates; and, after a reply, response.html, response-ledger.json and drafting-handoff.csv. Only selected excerpts enter the client page. Keep full transcripts and author-only held items in your workspace.
Why this is more than an approval prompt: it includes original readable code for the working review page, exact line/excerpt checks, visibly labeled edited proposals, validation of missing/conflicting reply fields and exact-packet reply matching. Changed wording, attribution, use notes or excerpts produce a new packet identity. A reply to an old packet is rejected instead of silently applied. Your agent still checks speaker/context and expected respondent.
What it needs: a connected agent with local file/shell tools, Python 3.10+ and a modern browser preview. The single Python helper uses the standard library and runs independently from a temporary file. No processing subscription, API key, custom MCP server or external Python dependency. Your agent/model may process supplied material under its provider's policies.
Where it stops: wording feedback is not a legal release or publication permission. Respondent names and times are self-reported; there is no identity/authority verification, signature or tamper-proof approval. Exact text occurrence does not prove facts, meaning, representative results or transcription accuracy. No email is sent, no reminder runs, and no content is published or inserted into your draft automatically. The HTML has no access control; anyone you give it to can read its included excerpts. It saves nothing automatically. Some organizations block interactive HTML; use the printable fallback and your normal review process.
Built for content agencies, customer marketers and case-study writers handling a focused quote wording round. Shared documents and free testimonial tools may be enough for your workflow; this one-time skill provides a repeatable portable artifact and reply compiler. No customer adoption, time savings or paid demand is claimed.
Checks exact supplied text occurrence and packet/version matching only. No factual, semantic, identity, authority, transcription, consent or legal clearance guarantee. Returned JSON is editable and unauthenticated. No hosting, access control, autosave, reminders or automatic sending. Some recipients block HTML; printable fallback available. Up to 100 quotes, 100 held items and 30 text sources per round; 6MB input files. Direct file-URL execution was not browser-tested; localhost preview and download/copy roundtrip passed.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 15 days ago
Requires
Permissions
File Scopes
Original source; fictional examples. One standalone temporary script, no sibling imports/dependencies/network/environment reads. Python excluding CSS/JS: 16660 bytes; executable JS: 3325 bytes; combined executable excluding CSS: 19985 bytes. Full source 23276 bytes includes 3291 bytes CSS. ZIP SHA256 153cfb2212c81a8f37a6dc6d0f00daaeecb023083c7e209802a96b5e4b36a85b. Six CLI behavioral test groups and independent fresh buyer run; localhost browser reply download/reconcile and 390px layout tested. Responses unauthenticated; no semantic/identity/legal guarantee. No outside checkout. User-selected transcript/plan/packet/reply reads and fresh output writes. Browser is for local review artifact preview. No helper network access or environment variables; no account services to connect.