Oracle Evaluator — Development agent for Claude Code
Runs the evaluation harness on candidate artifacts via pp_harness best-of-N and Borda.
How to install Oracle Evaluator
Installs to ~/.claude/agents/lebobo88-agentsmith-oracle-evaluator.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/lebobo88/AgentSmith/HEAD/.claude/agents/oracle-evaluator.md -o ~/.claude/agents/lebobo88-agentsmith-oracle-evaluator.md Restart Claude Code, or start a new session, for it to be picked up.
What Oracle Evaluator does
name: oracle-evaluator description: Runs the evaluation harness on candidate artifacts via pp_harness best-of-N and Borda. Invoke after a draft has cleared Inspector and is ready for comparative judgment before promotion. model: opus tools: Read, mcp__agentsmith__oracle_evaluate, mcp__pp_harness__start_best_of_stage, mcp__pp_harness__borda_count, mcp__eights__evolution_propose skills: evolution-handoff, agent-factory-recipes color: magenta
Oracle-Evaluator
I do not tell you what to c
Alternatives in Development
- Harvest Curator — Draft session learnings from harvest staging into lizard-ready adoption notes 1k ★
- Inspector 230 ★
- Reverse Parity Inspector — Inspecteur de parité comportementale (Phase 3.8, optionnel) 193 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Smith Inspector
Schema and invariant validator for AgentSmith artifacts. Fail-closed. Invoke whenever a draft agent/skill/comm
Harness Checker
Fresh-context artifact evaluator for goal loops. Reads only final artifacts — never Maker reasoning or PROGRES
Evaluator Teammate
Code evaluation teammate. Works in Agent Teams mode, receives Review Dispatch payloads, reviews scoped code ch
Bully Evaluator
Evaluates a single bully semantic-evaluation payload against a diff and returns a structured violation list. I
Smith Quarantine
Isolates rogue artifacts and opens HITL tickets. Invoke when the sentinel classifies an anomaly as actionable,
Cleanliness Inspector
File organization inspector for cleanup tasks. Detects misplaced docs, duplicates, temp files, build artifacts
Related Skills
D5 Comparative
D5: Comparative (With/Without) Evaluation
Kagemusha
影武者 / kagemusha: a Markdown governance kit for Claude Code and AI agents—approval queues, correction promotion
Agent Oracle Codex
Invoke the Claude-native Codex oracle for explicit cross-harness QAQC or implementation delegation.