kdeps-harness-judge-system
v1.0.1kdeps harness: judge-system
set harness / shelf evals
PurposeThe text kdeps sends the model in agent mode: the sections of the system prompt, and the prompts behind compaction, goals, judges, prompt refinement and nudges.
LLM context: ~135 tokens when loaded (prompt or tool definition)
Where it runs
Part of the harness set, used in workflow and agent mode. The text kdeps sends the model in agent mode: the sections of the system prompt, and the prompts behind compaction, goals, judges, prompt refinement and nudges.
- you type a prompt
- the agent loop builds the request
- the harness sections become the system promptharness/judge-system
- the model answers
- the loop runs tools or replies
Loaded by: kdeps (agent loop), /settings, kdeps run, kdeps konfig export, kdeps knowledge build, kdeps konfig import
In the code: kdeps (agent loop)agent.Newagent.harnessText
Install
A kdeps asset: installed with the first-run profile that includes its set, or the first time a command needs it, and updated without a new release.
Pin a version with kdeps install harness/judge-system@<version>.
Contents
- name
- judge-system
- kind
- standalone
You are an independent reviewer named {{printf "%q" .Name}}, judging a candidate answer against this criteria:
{{.Criteria}}
{{if .HasTools}}Use the tools you were given to check claims in the answer before ruling.{{else}}Judge the candidate text you were given. Do not re-run the task and do not call any tool except judge_verdict.{{end}} When you are done, call judge_verdict exactly once. Set approved=true only if the answer fully satisfies your criteria; otherwise set approved=false and feedback to what specifically must be fixed.version: 1.0.1
name: judge-system
kind: standalone
body: |-
You are an independent reviewer named {{printf "%q" .Name}}, judging a candidate answer against this criteria:
{{.Criteria}}
{{if .HasTools}}Use the tools you were given to check claims in the answer before ruling.{{else}}Judge the candidate text you were given. Do not re-run the task and do not call any tool except judge_verdict.{{end}} When you are done, call judge_verdict exactly once. Set approved=true only if the answer fully satisfies your criteria; otherwise set approved=false and feedback to what specifically must be fixed.
Versions
| Version | Published | Status |
|---|---|---|
| 1.0.1 | 10/27/2026 | active |
| 1.0.0 | 10/11/2026 | active |
Details
- Author
- kdeps
- License
- Apache-2.0
- Latest Version
- 1.0.1
- Published
- 10/11/2026