メニュー
Configure which model and effort level your proxyme proxy uses. Shows current config and lets you pick a new one. /proxyme:model → show current; /proxyme:model set → interactive picker; /proxyme:model reset → restore defaults.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | proxyme-model |
| description | Configure which model and effort level your proxyme proxy uses. Shows current config and lets you pick a new one. /proxyme:model → show current; /proxyme:model set → interactive picker; /proxyme:model reset → restore defaults. |
| argument-hint | [set | reset] |
| allowed-tools | Bash, Read, Write |
Configure the model and effort level used when /proxyme spawns your proxy agent.
Config is saved to ~/.claude/skills/proxyme/config.json and read every time /proxyme activates.
/proxyme:model → show current configuration
/proxyme:model set → interactive picker (model + effort)
/proxyme:model reset → restore defaults (opus, xhigh)
showRead and display current config:
cat ~/.claude/skills/proxyme/config.json 2>/dev/null || echo '{"model":"opus","effort":"xhigh"}'
Display as:
Current proxy config:
Model: <model>
Effort: <effort>
Run /proxyme:model set to change.
set — interactive pickerStep 1 — Show model options and ask:
Choose proxy model:
1. opus — Most capable. Best judgment, highest cost. (default)
2. sonnet — Balanced. Fast and capable, lower cost.
3. haiku — Fastest. Good for simple delegation tasks.
Wait for user to pick 1, 2, or 3 (or type the name directly).
Step 2 — Show effort options and ask:
Choose effort level:
1. xhigh — Maximum reasoning depth. (default)
2. high — Strong reasoning, slightly faster.
3. medium — Standard effort.
4. low — Minimal reasoning. Fast, cheap.
Wait for user to pick 1–4 (or type the name directly).
Step 3 — Save config:
mkdir -p ~/.claude/skills/proxyme
Write ~/.claude/skills/proxyme/config.json:
{"model": "<chosen>", "effort": "<chosen>"}
Step 4 — Confirm:
Proxy config updated:
Model: <model>
Effort: <effort>
Takes effect on next /proxyme activation.
resetWrite defaults to ~/.claude/skills/proxyme/config.json:
{"model": "opus", "effort": "xhigh"}
Confirm: "Proxy config reset to defaults: opus / xhigh."
/proxyme defaults to opus + xhigh/proxyme activationAnalyzes your Claude Code memories and session history to synthesize your digital identity file (~/.claude/skills/proxyme/${LOGNAME}-identity.md). Run once to bootstrap, then refresh when your preferences or active projects change significantly. Requires Claude Code session history (JSONL files in ~/.claude/projects/).
Activate a read-only, ephemeral, one-shot digital proxy. While active, any question Claude would normally ask the user is answered by a FRESH proxy that reads (read-only) this workdir, replies once with your full authority, and terminates — purely reactive, no proactive advice. /proxyme --off deactivates for this session. --except "<carve-out>" registers a session carve-out. An optional positional instruction is answered immediately as a one-shot. Runs /proxyme-identity automatically if no identity file exists yet.
Adversarial actor/critic validation of your digital proxy. Generates analogous, held-out questions from the info collected by /proxyme-identity, queries the proxy hypothesis (actor), scores each answer on a documented scorecard (critic), and iterates the GENERAL identity until the scorecard averages 8.5/10 — never inserting the specific case. Run after /proxyme-identity and before trusting /proxyme.