brainqub3/RLM
Claude code setup as an RLM scaffhold Implemented by Brainqub3
ARCHETYPE
Minimalist
Lean instructions, few tools, tiny context tax. Lets the model think.
Copy this rig
# review before running: this installs third-party code
$ npx degit brainqub3/RLM/.claude ./rig-rlm # 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 brainqub3/RLM/.claude/skills/rlm .claude/skills/rlm Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Read(./rlm_vs_agent_experiment/_archive/**)",
"Read(./.claude/skills/rlm/eval/data/contexts_with_labels/**)",
"Read(./.claude/skills/rlm/eval/data/oolong_trec_coarse.jsonl)",
"Read(./.claude/skills/rlm/eval/_cache/**)",
"Read(./.claude/skills/rlm/eval/README.md)",
"Read(./rlm_vs_agent_experiment/_sealed/**)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "Read|Grep|Glob|Bash",
"hooks": [
{
"type": "command",
"command": "grep -qiE '_archive|contexts_with_labels|_cache|verify_eval|oolong_trec_coarse|eval..?readme|gold_label_stats' && printf '%s' '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"permissionDecision\":\"deny\",\"permissionDecisionReason\":\"Block"
},
{
"type": "command",
"command": "grep -qE '_sealed/' && printf '%s' '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"permissionDecision\":\"deny\",\"permissionDecisionReason\":\"Blocked: sealed prior-arm eval results (cross-arm leak guard, issue #6). A finished arm is moved"
},
{
"type": "command",
"command": "[ -n \"$RLM_CONTROL_SESSION\" ] && grep -qE 'rlm_repl|llm_query|rlm_query' && printf '%s' '{\"hookSpecificOutput\":{\"hookEventName\":\"PreToolUse\",\"permissionDecision\":\"deny\",\"permissionDecisionReason\":\"Blocked: a RLM-OFF control arm may not use "
}
]
}
]
}
} Skills (1)
Hooks (3)
| event | matcher | runs |
|---|---|---|
| PreToolUse | Read|Grep|Glob|Bash | grep -qiE '_archive|contexts_with_labels|_cache|verify_eval|oolong_trec_coarse|eval..?readme|gold_label_stats' && printf '%s' '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","permissionDecisionReason":"Block |
| PreToolUse | Read|Grep|Glob|Bash | grep -qE '_sealed/' && printf '%s' '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","permissionDecisionReason":"Blocked: sealed prior-arm eval results (cross-arm leak guard, issue #6). A finished arm is moved |
| PreToolUse | Read|Grep|Glob|Bash | [ -n "$RLM_CONTROL_SESSION" ] && grep -qE 'rlm_repl|llm_query|rlm_query' && printf '%s' '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","permissionDecisionReason":"Blocked: a RLM-OFF control arm may not use |
Permissions
deny (6)
Read(./rlm_vs_agent_experiment/_archive/**)
Read(./.claude/skills/rlm/eval/data/contexts_with_labels/**)
Read(./.claude/skills/rlm/eval/data/oolong_trec_coarse.jsonl)
Read(./.claude/skills/rlm/eval/_cache/**)
Read(./.claude/skills/rlm/eval/README.md)
Read(./rlm_vs_agent_experiment/_sealed/**)
ask (0)
—
allow (0)
—
Similar rigs
multica-ai/andrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
Minimalist 723 tok ·
shareAI-lab/learn-claude-code
Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1
Minimalist 156 tok ·
tt-a1i/archify
Agent skill for beautiful, verifiable architecture, workflow, sequence, data-flow, and lifecycle diagrams—self-contained HTML with motion and crisp export.
Minimalist 484 tok ·
ayghri/i-have-adhd
A skill to stop your coding agent from burying the answer. ADHD-friendly output.
Minimalist 1.4k tok ·