Installs into .claude/skills of the current project.
Are you the author of Gate Judgment?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/yogsoth-ai-gate-judgment)
---
name: gate-judgment
description: Evaluate a candidate against gate criteria and render GO/KILL/RECYCLE verdict with evidence.
execution: subagent
prompt: ./prompt.md
input: candidate, gate_criteria
used-by: feasibility-assessment
---
# Gate Judgment
Evaluate a specific candidate against defined gate criteria and render a verdict: GO (proceed to next stage), KILL (terminate), or RECYCLE (rework and re-evaluate). Every verdict is backed by evidence.
## Execution
Spawns a subagent that:
1. Receives candidate data and gate criteria specification
2. Evaluates candidate against each criterion
3. Applies the pass rule to determine overall verdict
4. Documents evidence for each criterion evaluation
5. Specifies conditions for advancement if verdict is not GO
## Why Subagent
Gate judgment must be rigorous and unbiased. A dedicated subagent applies criteria mechanically without emotional attachment to the candidate, ensuring honest evaluation.
## HARD-GATE
Output MUST include: verdict (GO/KILL/RECYCLE), per-criterion evaluation with evidence, and conditions for advancement. Reject if verdict is not supported by criterion-level evaluation.
