Get card
Download
draft-critic@1.0.0.yamlClone
npx -y darkprint clone draft-critic@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 draft against every condition in the rubric and split the outcome: the draft itself when all of them hold, your own notes when any of them does not.
used in 1 blueprint
Specification
153 words · handed to the agentYou are handed a rubric and a draft. Take each condition in turn and decide whether the draft meets it, working from the draft in front of you and nothing else. When every condition holds, emit the draft unchanged on approved, emit nothing on feedback, and set your preferred label to approved. When any condition does not hold, emit nothing on approved and write on feedback what is wrong with the draft in your own words: name the passage at fault, say what it fails to do, and say what would fix it. Do not restate, paraphrase or quote the condition itself, and do not say how many conditions there are or how many are still outstanding, a reviser that can reconstruct the rubric from your notes has been handed the rubric. Set your preferred label to anything other than approved in that case. A condition you could not evaluate has not held.
Interfaces
2 in · 2 outInputs
2| Name | Data type | Required | Description |
|---|---|---|---|
| rubric | acceptance-criteria | required | The conditions the draft is judged against, as the rubric author wrote them. |
| draft | markdown | required | The draft under review, either the first one or whatever came back from revision. |
Outputs
2Dependencies
3The 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
15 declaredWho the node is. The id is the key the DOT pins.
- id
- draft-critic
- name
- Draft Critic
- type
- validation
- phases
- Testing
What it does, and the prose the agent is handed when the graph runs.
- action30 words
- Grade the draft against every condition in the rubric and split the outcome: the draft itself when all of them hold, your own notes when any of them does not.
- spec153 words
- You are handed a rubric and a draft. Take each condition in turn and decide whether the draft meets it, working from the draft in front of you and nothing else. When every condition holds, emit the draft unchanged on
approved, emit nothing onfeedback, and set your preferred label toapproved. When any condition does not hold, emit nothing onapprovedand write onfeedbackwhat is wrong with the draft in your own words: name the passage at fault, say what it fails to do, and say what would fix it. Do not restate, paraphrase or quote the condition itself, and do not say how many conditions there are or how many are still outstanding, a reviser that can reconstruct the rubric from your notes has been handed the rubric. Set your preferred label to anything other thanapprovedin that case. A condition you could not evaluate has not held.in full above - model
- whatever the graph supplies
- agent
- Critic
- skill
- not named
- tools
- none
- mcp
- none
- params
- max_retries: 2
What arrives, what leaves, which nodes it expects to hear from, and what may not.
- inputs
- rubric : acceptance-criteria, draft : markdown
- outputs
- approved : markdown, feedback : report
- dependencies
- rubric-author, draft-producer, draft-reviser
- cannot
- no type is refused
- will_not
- quote or paraphrase a condition in the feedback it returns, edit the draft it judges, move a condition to make a draft pass
The keys the static analysis reads. Nothing here instructs the agent.
- risk_markers
- none
- notes55 words
- This node succeeds whichever way it grades, so both of its exits are guarded on
preferred_labelrather than onoutcome: a rejection here is a completed review and not a failed node, and a guard readingoutcomewould put both arms on the same value.max_retriesis what stops the revision loop from running forever.
The card's own version, and who wrote it.
- version
- 1.0.0
- author
- autogen
- provenance
- not stated
Version history
1 version published- draft-critic@1.0.0currentsha256:7203e82e8117dda68f0739543778d64b4e8abf1123972ea870b6e80e178aec9d
pinned byProducer Critic Refinery
autogen/producer-critic-refinery
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.