Works with the AI Agents you already use
knowledge response drafting
This skill automates the matching of support queries to verified knowledge base articles to generate grounded, cited response drafts.
$1.99
knowledge response drafting
Example session with this skill installed
I have 2 questions about our refund policy. Match them against our knowledge base articles KB001 and KB002 and give me drafted replies.
- Read your context and instructions
- Compiled the knowledge response drafting
I have processed the 2 questions against the provided knowledge base articles. The matching script has completed with grounding for both queries.
Knowledge-based Response Drafting — Support Team, 2024-05-20
1. Summary
- Total questions: 2
- Grounded responses: 2
- Approval required: 2
2. Knowledge base coverage
- Total articles: 2
- Approved articles: 2
- Usage distribution: KB001 (50%), KB002 (50%)
3. Response drafts
Question Q1: What is the refund window for software subscriptions?
- Status: Grounded
- Primary Article: KB001 (Refund Policy Overview)
- Relevance Score: 0.92
- Citations: [KB001, Section 2.1]
Draft: Thank you for reaching out. According to our policy, software subscriptions are eligible for a full refund if requested within 14 days of the initial purchase.
- Approval: Management decision required.
Question Q2: Do you offer pro-rated refunds for cancellations?
- Status: Grounded
- Primary Article: KB002 (Subscription Cancellations)
- Relevance Score: 0.88
- Citations: [KB002, Section 4.5]
- Draft: We do not typically offer pro-rated refunds for mid-cycle cancellations, as outlined in our service terms.
- Approval: Management decision required.
4. Warnings and items for agent review
- No warnings reported.
- All drafted responses require final agent validation for appropriateness and tone.
Next steps
- Review the generated drafts for tone alignment.
- Verify the 14-day window against the customer's specific purchase date.
- Approve and send the grounded responses via the CRM.
Connects securely to your tools. The creator never sees your data.
About this skill
Support teams often struggle to maintain consistency and accuracy when drafting customer replies. This skill automates the matching of support queries to verified knowledge base articles to generate grounded, cited response drafts.
What it does
- Knowledge Matching uses keyword-based retrieval to link support questions to approved internal documentation.
- Relevance Scoring calculates exact confidence intervals for every match to ensure factual grounding.
- Draft Generation produces customer-ready responses that include direct citations to source articles.
- Gap Detection flags questions that cannot be answered by the current knowledge base as information gaps.
- Safety Filtering identifies and excludes unapproved articles or unauthorized questions from the drafting process.
How it works
- Source Registration: The agent catalogs the provided support questions and knowledge articles for full traceability.
- Request Processing: It builds a structured JSON request containing the raw text data and metadata.
- Execution: A deterministic retrieval script computes the matching logic and relevance scores.
- Report Generation: The agent delivers a structured Markdown report with response drafts, citations, and coverage metrics.
Frameworks & tools
The skill utilizes a self-contained Python retrieval script (scripts/run.py) and follows strict JSON schema validation for inputs and outputs. It is designed to work with structured knowledge base exports and raw support ticket data.
Why this beats prompting it yourself
Standard prompting often results in hallucinations or "creative" answers that deviate from company policy. This skill uses a deterministic script to enforce keyword-grounding and exact relevance scoring, ensuring every draft is backed by an approved source.
Use cases
- Drafting customer support replies based on a verified product wiki.
- Identifying holes in documentation where support questions have no matching articles.
- Auditing existing support responses to ensure they align with the latest approved content.
Known limitations
Requires both a list of questions and a list of knowledge articles to function. It does not perform semantic or vector-based search, relying instead on keyword retrieval.
Requires structured knowledge articles and questions. Relies on keyword-based retrieval rather than semantic search. Human review is mandatory for all drafts.
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 1 day ago