Get card
Download
reply-qa-check@1.0.0.yamlClone
npx -y darkprint clone reply-qa-check@1.0.0There is no repository and no history behind a card. Download hands you the document as it stands, and Clone fetches the same document by name.
Grade the drafted reply against the tone, accuracy and completeness rubric, release it only above the confidence threshold, and emit the score the escalation path branches on.
used in 1 blueprint
Specification
146 words · handed to the agentRead the drafted reply and the citations attached to it on draft, and grade it against the support-reply-rubric@2 rubric on three axes: tone, factual accuracy and completeness. Check every factual claim in the reply against the citation the draft says supports it, and score an uncited or mis-cited claim as an accuracy failure rather than as a style note. Emit the graded score on confidence as a single number between 0 and 1, and put the reply on approved, unchanged apart from your QA stamp, only when that score is 0.72 or above. The customer's personal details ride inside the draft: leave them there, and do not copy names, addresses, order numbers or account identifiers into the score or into any comment you attach to it. You are done when a score is on confidence and approved carries either the stamped reply or nothing at all.
Interfaces
1 in · 2 outInputs
1| Name | Data type | Required | Description |
|---|---|---|---|
| draft | json | required | The composed reply and the citations it claims to rest on. |
Outputs
2Dependencies
1The upstream nodes this card expects to receive from. Whenever a blueprint pins this card, each name is checked against a real edge in that graph. A name without a link is not a published card; it refers to a node inside some graph.
Card values
18 declaredWho the node is. The id is the key the DOT pins.
- id
- reply-qa-check
- name
- Reply QA Check
- type
- validation
- phases
- Testing
What it does, and the prose the agent is handed when the graph runs.
- action27 words
- Grade the drafted reply against the tone, accuracy and completeness rubric, release it only above the confidence threshold, and emit the score the escalation path branches on.
- spec146 words
- Read the drafted reply and the citations attached to it on
draft, and grade it against thesupport-reply-rubric@2rubric on three axes: tone, factual accuracy and completeness. Check every factual claim in the reply against the citation the draft says supports it, and score an uncited or mis-cited claim as an accuracy failure rather than as a style note. Emit the graded score onconfidenceas a single number between 0 and 1, and put the reply onapproved, unchanged apart from your QA stamp, only when that score is 0.72 or above. The customer's personal details ride inside the draft: leave them there, and do not copy names, addresses, order numbers or account identifiers into the score or into any comment you attach to it. You are done when a score is onconfidenceandapprovedcarries either the stamped reply or nothing at all.in full above - model
- claude-sonnet-5
- agent
- QA checker
- skill
- skills/reply-qa-check.md
- tools
- none
- mcp
- none
- params
- rubric_ref: support-reply-rubric@2, confidence_threshold: 0.72
What arrives, what leaves, which nodes it expects to hear from, and what may not.
- inputs
- draft : json
- outputs
- approved : json, confidence : status
- dependencies
- resolution-composer
- cannot
- no type is refused
- will_not
- copy a customer identifier into the score it emits
The keys the static analysis reads. Nothing here instructs the agent.
- risk_markers
- lupo/pii-handling
- notes54 words
- The customer's details ride through this node inside the draft, which is what the
lupo/pii-handlingmarker records, a local term, because doc 3 §4 has no marker for personal data. The threshold is the one dial that trades escalation volume against send quality; raising it moves work to people rather than improving the replies.
The card's own version, and who wrote it.
- version
- 1.0.0
- author
- autogen
- provenance
- not stated
Version history
1 version published- reply-qa-check@1.0.0currentsha256:72af41e2cbfd773043fd473e3e4d8f4038193a56e113771486ff7b369c346e8f
pinned byFrontline Triage
autogen/frontline-triage
A digest is a fingerprint (SHA-256) of the card's content, computed without the author and provenance fields. The same card from two people gets the same digest; any edit gets a new one.
First published version, so there is nothing to compare yet. Versions are never edited in place: the next change arrives as a new version, and the differences between the two documents are listed here.
Community notes (0)
No notes yet.
Nobody has posted about this node card yet.
Sign in to post a note.