EmmanuelRTM/ScientistOne_claude_code_primitives
In this repo, we explore Google's ScientistOne research pipeline, recreated entirely from Claude Code primitives — subagents as researchers, skills as stages, hooks as integrity guards. Includes a bundled offline demo task and Chain-of-Evid
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
Copy this rig
# review before running: this installs third-party code
$ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude ./rig-scientistone_claude_code_primitives # inspect, then merge into .claude/ MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/citation-provenance .claude/skills/citation-provenance $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/discover .claude/skills/discover $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/distill-feedback .claude/skills/distill-feedback $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/evaluation-protocol .claude/skills/evaluation-protocol $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/evidence-tagging .claude/skills/evidence-tagging $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/investigate .claude/skills/investigate $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/paper-to-latex .claude/skills/paper-to-latex $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/research .claude/skills/research $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/run-status .claude/skills/run-status $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/verify-claims .claude/skills/verify-claims $ npx degit EmmanuelRTM/ScientistOne_claude_code_primitives/.claude/skills/write-paper .claude/skills/write-paper
$ curl -fsSL --create-dirs -o .claude/agents/ablation-analyst.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/ablation-analyst.md $ curl -fsSL --create-dirs -o .claude/agents/auditor.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/auditor.md $ curl -fsSL --create-dirs -o .claude/agents/claim-verifier.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/claim-verifier.md $ curl -fsSL --create-dirs -o .claude/agents/evaluator.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/evaluator.md $ curl -fsSL --create-dirs -o .claude/agents/ideator.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/ideator.md $ curl -fsSL --create-dirs -o .claude/agents/paper-critic.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/paper-critic.md $ curl -fsSL --create-dirs -o .claude/agents/paper-writer.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/paper-writer.md $ curl -fsSL --create-dirs -o .claude/agents/problem-investigator.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/problem-investigator.md $ curl -fsSL --create-dirs -o .claude/agents/refiner.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/refiner.md $ curl -fsSL --create-dirs -o .claude/agents/solver.md https://raw.githubusercontent.com/EmmanuelRTM/ScientistOne_claude_code_primitives/claude/claude-project-multiagent-c9913z/.claude/agents/solver.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Bash(pip install*)",
"Bash(pip3 install*)",
"Bash(rm -rf*)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "Write|Edit",
"hooks": [
{
"type": "command",
"command": "python3 \"$CLAUDE_PROJECT_DIR\"/.claude/hooks/citation_guard.py"
}
]
},
{
"matcher": "Bash",
"hooks": [
{
"type": "command",
"command": "python3 \"$CLAUDE_PROJECT_DIR\"/.claude/hooks/citation_guard.py"
}
]
},
{
"matcher": "AskUserQuestion",
"hooks": [
{
"type": "command",
"command": "python3 \"$CLAUDE_PROJECT_DIR\"/.claude/hooks/run_stop_gate.py"
}
]
}
]
}
} Skills (11)
citation-provenancediscoverdistill-feedbackevaluation-protocolevidence-tagginginvestigatepaper-to-latexresearchrun-statusverify-claimswrite-paper
Subagents (10)
| ablation-analyst | > |
| auditor | > |
| claim-verifier | > |
| evaluator | > |
| ideator | > |
| paper-critic | > |
| paper-writer | > |
| problem-investigator | > |
| refiner | > |
| solver | > |
Hooks (7)
| event | matcher | runs |
|---|---|---|
| SessionStart | * | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/session_start.py |
| PreToolUse | Write|Edit | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/citation_guard.py |
| PreToolUse | Bash | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/citation_guard.py |
| PreToolUse | AskUserQuestion | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/run_stop_gate.py |
| Stop | * | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/run_stop_gate.py |
| SubagentStart | * | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/ledger_log.py start |
| SubagentStop | * | python3 "$CLAUDE_PROJECT_DIR"/.claude/hooks/ledger_log.py stop |
Permissions
deny (3)
Bash(pip install*)
Bash(pip3 install*)
Bash(rm -rf*)
ask (0)
—
allow (7)
Bash(python3 .claude/scripts/*)
Bash(python3 "$CLAUDE_PROJECT_DIR"/.claude/scripts/*)
Bash(python3 workspace/runs/*)
Bash(python3 workspace/tasks/*)
Read
Glob
Grep
Similar rigs
taishi-i/awesome-japanese-nlp-resources
A curated list of resources for Japanese natural language processing (NLP): Python libraries, LLMs, dictionaries, corpora, and datasets. Includes Claude Code and Codex skills to search resources.
Minimalist 338 tok ·
OdradekAI/bundles-forge
An agentic skills framework & bundle-plugin engineering toolkit that works.
Pragmatist 2.4k tok ·
GeiserX/agent-skills
Skills for Claude Code and Codex: durable goal loops, parallel review, research, investigation and implementation, and research grounded in your own knowledge base
Skill Collector 2.0k tok ·
SynaLinks/synalinks
Keras based framework for neuro-symbolic LM systems
Minimalist 840 tok ·