Get card
Download
trajectory-auditor@1.0.0.yamlClone
npx -y darkprint clone trajectory-auditor@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.
Judge the route a run took rather than the answer it arrived at, and report where the route was wrong even when the answer was right.
used in 0 blueprints
Specification
139 words · handed to the agentYou are handed the recorded steps of a completed run: what each node did, in what order, and what it called. Judge the route. Report on audit every step that was taken out of order, every call made with arguments the previous step did not support, every repetition that added nothing, and every point where the run reached the right place by a path that would not work again. Say explicitly when the outcome was correct and the route was not: that is the finding this node exists for, and a run that succeeded by luck is the one most likely to be copied. Do not grade the final answer, another node does that. Do not omit a step because it succeeded, and do not invent a timing or an argument for a step the record does not carry.
Interfaces
1 in · 1 outInputs
1| Name | Data type | Required | Description |
|---|---|---|---|
| trace | json | required | The recorded steps of the run, in the order they happened. |
Outputs
1| Name | Data type | Description |
|---|---|---|
| audit | report | Where the route was wrong, including where it was wrong despite the outcome being right. |
Dependencies
0None declared, no edge has to arrive for this node to run.
Card values
12 declaredWho the node is. The id is the key the DOT pins.
- id
- trajectory-auditor
- name
- Trajectory Auditor
- type
- validation
- phases
- Testing
What it does, and the prose the agent is handed when the graph runs.
- action26 words
- Judge the route a run took rather than the answer it arrived at, and report where the route was wrong even when the answer was right.
- spec139 words
- You are handed the recorded steps of a completed run: what each node did, in what order, and what it called. Judge the route. Report on
auditevery step that was taken out of order, every call made with arguments the previous step did not support, every repetition that added nothing, and every point where the run reached the right place by a path that would not work again. Say explicitly when the outcome was correct and the route was not: that is the finding this node exists for, and a run that succeeded by luck is the one most likely to be copied. Do not grade the final answer, another node does that. Do not omit a step because it succeeded, and do not invent a timing or an argument for a step the record does not carry.in full above - model
- whatever the graph supplies
- agent
- Auditor
- skill
- not named
- tools
- none
- mcp
- none
- params
- none
What arrives, what leaves, which nodes it expects to hear from, and what may not.
- inputs
- trace : json
- outputs
- audit : report
- dependencies
- none
- cannot
- no type is refused
- will_not
- grade the final answer, invent a timing or an argument the record does not carry, drop a step from the audit because it succeeded
The keys the static analysis reads. Nothing here instructs the agent.
- risk_markers
- none
- notes
- none
The card's own version, and who wrote it.
- version
- 1.0.0
- author
- autogen
- provenance
- not stated
Version history
1 version published- trajectory-auditor@1.0.0currentsha256:dbea52b5716da14a7f15a515fb05dc8d86ae6c45df2446405ffe6aea1e3d057f
No blueprint pins this exact version.
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.