[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6C31FK52VCG95KM6CS2M3Y5":3,"skill-same:01M6C31FK52VCG95KM6CS2M3Y5":88},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":25,"publishedAt":56,"linkOnly":57,"markdownContent":58,"markdownBody":59,"frontMatter":60,"refName":67,"commitSha":68,"skillDirectory":69,"forksCount":70,"repositoryHtmlUrl":71,"repositoryHomepageUrl":6,"ownerHtmlUrl":72,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":73,"install":74,"resources":87},"01M6C31FK52VCG95KM6CS2M3Y5","ao-spawn-minimax-worker",null,"Spawn an Agent-Orchestrator (AO) worker that uses the minimax CLI with the M3 model (via the minimax Anthropic-compatible API). Use when: the task should be dispatched to a worker that uses minimax M3 specifically, when the user says 'use minimax', 'use the M3 model', 'use the minimax worker', 'ao spawn minimax', or 'minimax CLI for AO'. This skill verifies the env (MINIMAX_MODEL, MINIMAX_API_KEY, ANTHROPIC_BASE_URL) end-to-end, then runs `ao spawn --agent minimax`. Verified 2026-06-13 against session ao-6355 \u002F PR #678.","jleechanorg\u002Fjleechan-skills","jleechanorg","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F221404966?v=4","01M6C1VJ99XG950X1DWPEBCBGX","01M6C1VJDK5YVDNT3362RT5WSB",6,"MIT","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002Fhermes\u002Fskills\u002Fao-spawn-minimax-worker\u002FSKILL.md",[17,22],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000010","后端与接口","SCENARIO","使用场景",{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发",[26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55],"worker","ao","ao spawn minimax worker","minimax","spawn","orchestrator","2026","when","errors","URL","io","end","time","default","use","session","cli","read","choice","model","src","Bash","shell","api","https","pull","packages","plugin","pr","github","2026-11-10 10:12:30",false,"---\nname: ao-spawn-minimax-worker\ndescription: \"Spawn an Agent-Orchestrator (AO) worker that uses the minimax CLI with the M3 model (via the minimax Anthropic-compatible API). Use when: the task should be dispatched to a worker that uses minimax M3 specifically, when the user says 'use minimax', 'use the M3 model', 'use the minimax worker', 'ao spawn minimax', or 'minimax CLI for AO'. This skill verifies the env (MINIMAX_MODEL, MINIMAX_API_KEY, ANTHROPIC_BASE_URL) end-to-end, then runs `ao spawn --agent minimax`. Verified 2026-06-13 against session ao-6355 \u002F PR #678.\"\nwhen_to_use: \"Use when the user asks to spawn an AO worker on minimax M3, or any time `ao spawn` is needed and the model choice is minimax. Do NOT use for: direct `claude` CLI runs (use shell directly), or for AO workers that should use a different model (use the default `ao spawn` with no `--agent` flag). NOT for fixing 401 errors (use `minimax-401-diagnostic` instead).\"\nallowed-tools: [\"Bash\", \"Read\", \"Write\", \"Edit\"]\ncontext: \"Verified end-to-end on 2026-06-13 with session ao-6355 and PR https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fagent-orchestrator-ts\u002Fpull\u002F678. The agent-minimax plugin (packages\u002Fplugins\u002Fagent-minimax\u002Fsrc\u002Findex.ts) wraps the `claude` CLI with ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic and propagates MINIMAX_MODEL from the tmux env. modelByCli.minimax.model=MiniMax-M3 in ~\u002Fagent-orchestrator.yaml makes this the default. The skill is intentionally a thin pointer — the canonical implementation is the AO plugin source + the user's `~\u002Fagent-orchestrator.yaml`.\"\n---\n\n# ao-spawn-minimax-worker\n\n## What it does\n\nSpawns an AO worker that uses the **minimax CLI** (an Anthropic-compatible adapter) with the **M3** model. Verifies env, then dispatches.\n\n## When NOT to use this skill\n\n- The user wants a direct `claude` CLI run (use the shell — no AO)\n- The user wants a different model (just `ao spawn \"\u003Ctask>\"` with no `--agent` flag)\n- The user is debugging a 401 from the minimax API → use `minimax-401-diagnostic` instead\n- The user wants a non-AO worker (e.g., `claude -p`, `codex`, `opencode`) → use those CLIs directly\n\n## Pre-flight checks (do these BEFORE spawning)\n\n```bash\n# 0. CRITICAL — Check that AO will actually source bashrc.\n#    An explicit envSource: [] in agent-orchestrator.yaml overrides the default\n#    of [\"~\u002F.bashrc\"] and silently disables bashrc sourcing, which means\n#    MINIMAX_API_KEY never reaches the plugin (bd-g884 regression \u002F 2026-06-13).\n#    Refuse to spawn if this is wrong; do not work around it.\n#    Regex is indented-anchored (`^\\s*envSource:`) so it matches whether the\n#    key sits at column 0 or nested under `defaults:` \u002F a project block.\nif grep -E '^\\s*envSource:\\s*\\[\\s*\\]\\s*$' ~\u002Fagent-orchestrator.yaml >\u002Fdev\u002Fnull 2>&1; then\n  echo \"BLOCKER: ~\u002Fagent-orchestrator.yaml has 'envSource: []' — AO will NOT\"\n  echo \"         source bashrc, so MINIMAX_API_KEY will be missing. Fix the\"\n  echo \"         config first (replace with [\\\"~\u002F.bashrc\\\"] or delete the line\"\n  echo \"         to restore the default). Aborting spawn.\"\n  exit 1\nfi\n\n# 1. Confirm minimax model is set in bashrc\ngrep \"^export MINIMAX_MODEL\" ~\u002F.bashrc\n\n# 2. Confirm API key resolves in the current shell (sanity check only — the\n#    real check is (0) above, because the spawn happens in a different process)\n[ -n \"$MINIMAX_API_KEY\" ] && echo \"OK: MINIMAX_API_KEY set in current shell\" || echo \"WARN: MINIMAX_API_KEY missing in current shell (bashrc may still provide it)\"\n\n# 3. Confirm endpoint reachable\ncurl -s -o \u002Fdev\u002Fnull -w \"minimax endpoint: HTTP %{http_code}\\n\" -m 5 https:\u002F\u002Fapi.minimax.io\u002Fanthropic\n\n# 4. Resolve MINIMAX_ANTHROPIC_BASE_URL — fall back to the canonical value if\n#    the calling shell didn't inherit it from bashrc. This unblocks shells\n#    that source a minimal env (CI runners, fresh tmux, etc.) where the\n#    extra var is missing even though the standard MiniMax env is valid.\n: \"${MINIMAX_ANTHROPIC_BASE_URL:=https:\u002F\u002Fapi.minimax.io\u002Fanthropic}\"\nexport MINIMAX_ANTHROPIC_BASE_URL\necho \"MINIMAX_ANTHROPIC_BASE_URL=$MINIMAX_ANTHROPIC_BASE_URL\"\n\n# 5. Quick live test of M3 model — uses $MINIMAX_API_KEY from the calling shell.\n#    If (0) is OK but (5) returns 401, the issue is NOT envSource; see\n#    minimax-401-diagnostic skill instead.\ncurl -s -m 30 -X POST \"${MINIMAX_ANTHROPIC_BASE_URL}\u002Fv1\u002Fmessages\" \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -H \"x-api-key: $MINIMAX_API_KEY\" \\\n  -H \"anthropic-version: 2023-06-01\" \\\n  -d '{\"model\":\"MiniMax-M3\",\"max_tokens\":20,\"messages\":[{\"role\":\"user\",\"content\":\"Reply with: PONG_M3_OK\"}]}' \\\n  | python3 -c \"import json,sys; d=json.loads(sys.stdin.read()); print('model:', d.get('model','?')); print('text:', d.get('content',[{}])[0].get('text','?').strip())\"\n```\n\nExpected:\n- Pre-flight (0) must pass — the line must be absent, `[\"~\u002F.bashrc\"]`, or any non-empty array (matches at any indent)\n- `MINIMAX_MODEL=MiniMax-M3` (the user's bashrc already has this — see SOUL.md)\n- HTTP 301 (the API root redirects; that's normal)\n- (4) `MINIMAX_ANTHROPIC_BASE_URL` resolves to `https:\u002F\u002Fapi.minimax.io\u002Fanthropic` (either from bashrc or the default fallback)\n- Direct API call returns `model: MiniMax-M3` and `text: PONG_M3_OK`\n\n## Spawn\n\n```bash\n# From a project root that has an agent-orchestrator.yaml with modelByCli.minimax configured\nao spawn --agent minimax \"\u003Ctask description>\"\n\n# Example: smoke test the minimax worker\nao spawn --agent minimax \"Reply with: PONG_M3_FROM_MINIMAX_WORKER. Do nothing else.\"\n\n# Example: actual coding task\nao spawn --agent minimax \"Fix the bug in src\u002Ffoo.ts where bar() returns null\"\n```\n\nAO will:\n1. Create a worktree (or use existing project worktree)\n2. Launch a tmux session named `953501c04ccc-ao-XXXX`\n3. Inject the `agent-minimax` plugin env: `ANTHROPIC_BASE_URL`, `ANTHROPIC_MODEL=MiniMax-M3`, `ANTHROPIC_API_KEY`, `ANTHROPIC_AUTH_TOKEN`\n4. Run `claude --dangerously-skip-permissions` in the worktree\n5. Auto-create a PR once the worker makes commits\n\n## Verify the worker actually uses M3\n\n```bash\n# After ~10s, check the tmux env (proves the plugin injected M3)\ntmux show-environment -t 953501c04ccc-ao-XXXX 2>&1 | grep -E \"ANTHROPIC_BASE_URL|ANTHROPIC_MODEL|MINIMAX_API_KEY\"\n# Expected: ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic\n#           ANTHROPIC_MODEL=MiniMax-M3\n#           (ANTHROPIC_API_KEY \u002F ANTHROPIC_AUTH_TOKEN present but redacted)\n\n# Read the worker's screen\ntmux capture-pane -t 953501c04ccc-ao-XXXX -p | tail -30\n# Expected: a response from MiniMax-M3, with \"ctx ##-------- XX%\" in the footer\n```\n\n## Common pitfalls (caught in production 2026-06-13)\n\n### Drift: `claudem()` shell function is hardcoded to M2.5\n\n`~\u002F.bashrc` has `claudem() { ... ANTHROPIC_MODEL=\"MiniMax-M2.5\" ... claude --model MiniMax-M2.5 ... }`. If you call `claudem` directly (outside of AO), you get M2.5, not M3.\n\n**For AO workers this doesn't matter** — AO uses the plugin, not `claudem`. But if you're verifying with the shell, use raw `claude --model MiniMax-M3 ...` instead of `claudem`.\n\n### Drift: launchd plists hardcode M2.7\n\n`launchd\u002Fai.agento.lifecycle-all.plist.template` and `ai.agento.health.plist.template` both set `\u003Ckey>MINIMAX_MODEL\u003C\u002Fkey>\u003Cstring>MiniMax-M2.7\u003C\u002Fstring>`. The lifecycle worker spawned from these will use M2.7, NOT M3.\n\n**Workaround:** spawn via `ao spawn --agent minimax` from a shell, which uses `process.env.MINIMAX_MODEL` (= MiniMax-M3 from bashrc).\n\n**Proper fix (not done in this turn):** update the two plist templates to use `MiniMax-M3` or read from a sed-substituted var.\n\n### Drift: autonomous-harness CLI defaults hardcode M2.7\n\n`autonomous-harness\u002Fsrc\u002F{cli,harness-state,orchestrator}.ts` all default `minimax\u002FMiniMax-M2.7`. Anyone running the autonomous-harness without overriding `--generator-model` will get M2.7.\n\n### 401 errors\n\nIf the worker shows `\u002Flogin · API Error: 401` in its tmux pane, the API key is redacted. Use `minimax-401-diagnostic`.\n\n## Verification (proven by this skill)\n\n| Step | Result | Date |\n|------|--------|------|\n| Direct M3 API call | `model: MiniMax-M3, text: PONG_M3_OK` | 2026-06-13 |\n| `ao spawn --agent minimax` | Session `ao-6355` created | 2026-06-13 |\n| Plugin env in tmux | `ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic`, `ANTHROPIC_MODEL=MiniMax-M3`, key resolved | 2026-06-13 |\n| Worker model identity | `Cogitated for 7s` → `PONG_M3_FROM_MINIMAX_WORKER` | 2026-06-13 |\n| PR auto-created | https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fagent-orchestrator-ts\u002Fpull\u002F678 | 2026-06-13 |\n\n## Canonical sources\n\n- **Plugin source:** `~\u002Fproject_agento\u002Fagent-orchestrator\u002Fpackages\u002Fplugins\u002Fagent-minimax\u002Fsrc\u002Findex.ts`\n- **User config:** `~\u002Fagent-orchestrator.yaml` (lines ~119-128: `modelByCli.minimax.model: MiniMax-M3`)\n- **Companion skill (401):** `~\u002F.claude\u002Fskills\u002Fminimax-401-diagnostic\u002FSKILL.md`\n- **AGENTS.md (fork):** `~\u002Fproject_agento\u002Fagent-orchestrator\u002FAGENTS.md` (Provider Agent Plugins section)\n- **This skill (pointer):** `~\u002F.hermes_prod\u002Fskills\u002Fao-spawn-minimax-worker\u002FSKILL.md` (and `~\u002F.hermes\u002Fskills\u002F...`)\n\n## Pitfalls (anti-pattern guards)\n\n- **Don't** use `claudem` to verify M3 — it has M2.5 hardcoded; use raw `claude --model MiniMax-M3` or just check tmux env\n- **Don't** trust the autonomous-harness defaults (M2.7); pass `--generator-model minimax\u002FMiniMax-M3` explicitly\n- **Don't** modify `~\u002FLibrary\u002FLaunchAgents\u002Fai.agento.lifecycle-all.plist` to fix the M2.7 → M3 drift; update the `.template` file and re-run `setup-launchd.sh` instead\n- **Don't** spawn without verifying env first — if `MINIMAX_API_KEY` is missing, the worker will hit 401 and the babysit loop will be wasted\n","# ao-spawn-minimax-worker\n\n## What it does\n\nSpawns an AO worker that uses the **minimax CLI** (an Anthropic-compatible adapter) with the **M3** model. Verifies env, then dispatches.\n\n## When NOT to use this skill\n\n- The user wants a direct `claude` CLI run (use the shell — no AO)\n- The user wants a different model (just `ao spawn \"\u003Ctask>\"` with no `--agent` flag)\n- The user is debugging a 401 from the minimax API → use `minimax-401-diagnostic` instead\n- The user wants a non-AO worker (e.g., `claude -p`, `codex`, `opencode`) → use those CLIs directly\n\n## Pre-flight checks (do these BEFORE spawning)\n\n```bash\n# 0. CRITICAL — Check that AO will actually source bashrc.\n#    An explicit envSource: [] in agent-orchestrator.yaml overrides the default\n#    of [\"~\u002F.bashrc\"] and silently disables bashrc sourcing, which means\n#    MINIMAX_API_KEY never reaches the plugin (bd-g884 regression \u002F 2026-06-13).\n#    Refuse to spawn if this is wrong; do not work around it.\n#    Regex is indented-anchored (`^\\s*envSource:`) so it matches whether the\n#    key sits at column 0 or nested under `defaults:` \u002F a project block.\nif grep -E '^\\s*envSource:\\s*\\[\\s*\\]\\s*$' ~\u002Fagent-orchestrator.yaml >\u002Fdev\u002Fnull 2>&1; then\n  echo \"BLOCKER: ~\u002Fagent-orchestrator.yaml has 'envSource: []' — AO will NOT\"\n  echo \"         source bashrc, so MINIMAX_API_KEY will be missing. Fix the\"\n  echo \"         config first (replace with [\\\"~\u002F.bashrc\\\"] or delete the line\"\n  echo \"         to restore the default). Aborting spawn.\"\n  exit 1\nfi\n\n# 1. Confirm minimax model is set in bashrc\ngrep \"^export MINIMAX_MODEL\" ~\u002F.bashrc\n\n# 2. Confirm API key resolves in the current shell (sanity check only — the\n#    real check is (0) above, because the spawn happens in a different process)\n[ -n \"$MINIMAX_API_KEY\" ] && echo \"OK: MINIMAX_API_KEY set in current shell\" || echo \"WARN: MINIMAX_API_KEY missing in current shell (bashrc may still provide it)\"\n\n# 3. Confirm endpoint reachable\ncurl -s -o \u002Fdev\u002Fnull -w \"minimax endpoint: HTTP %{http_code}\\n\" -m 5 https:\u002F\u002Fapi.minimax.io\u002Fanthropic\n\n# 4. Resolve MINIMAX_ANTHROPIC_BASE_URL — fall back to the canonical value if\n#    the calling shell didn't inherit it from bashrc. This unblocks shells\n#    that source a minimal env (CI runners, fresh tmux, etc.) where the\n#    extra var is missing even though the standard MiniMax env is valid.\n: \"${MINIMAX_ANTHROPIC_BASE_URL:=https:\u002F\u002Fapi.minimax.io\u002Fanthropic}\"\nexport MINIMAX_ANTHROPIC_BASE_URL\necho \"MINIMAX_ANTHROPIC_BASE_URL=$MINIMAX_ANTHROPIC_BASE_URL\"\n\n# 5. Quick live test of M3 model — uses $MINIMAX_API_KEY from the calling shell.\n#    If (0) is OK but (5) returns 401, the issue is NOT envSource; see\n#    minimax-401-diagnostic skill instead.\ncurl -s -m 30 -X POST \"${MINIMAX_ANTHROPIC_BASE_URL}\u002Fv1\u002Fmessages\" \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -H \"x-api-key: $MINIMAX_API_KEY\" \\\n  -H \"anthropic-version: 2023-06-01\" \\\n  -d '{\"model\":\"MiniMax-M3\",\"max_tokens\":20,\"messages\":[{\"role\":\"user\",\"content\":\"Reply with: PONG_M3_OK\"}]}' \\\n  | python3 -c \"import json,sys; d=json.loads(sys.stdin.read()); print('model:', d.get('model','?')); print('text:', d.get('content',[{}])[0].get('text','?').strip())\"\n```\n\nExpected:\n- Pre-flight (0) must pass — the line must be absent, `[\"~\u002F.bashrc\"]`, or any non-empty array (matches at any indent)\n- `MINIMAX_MODEL=MiniMax-M3` (the user's bashrc already has this — see SOUL.md)\n- HTTP 301 (the API root redirects; that's normal)\n- (4) `MINIMAX_ANTHROPIC_BASE_URL` resolves to `https:\u002F\u002Fapi.minimax.io\u002Fanthropic` (either from bashrc or the default fallback)\n- Direct API call returns `model: MiniMax-M3` and `text: PONG_M3_OK`\n\n## Spawn\n\n```bash\n# From a project root that has an agent-orchestrator.yaml with modelByCli.minimax configured\nao spawn --agent minimax \"\u003Ctask description>\"\n\n# Example: smoke test the minimax worker\nao spawn --agent minimax \"Reply with: PONG_M3_FROM_MINIMAX_WORKER. Do nothing else.\"\n\n# Example: actual coding task\nao spawn --agent minimax \"Fix the bug in src\u002Ffoo.ts where bar() returns null\"\n```\n\nAO will:\n1. Create a worktree (or use existing project worktree)\n2. Launch a tmux session named `953501c04ccc-ao-XXXX`\n3. Inject the `agent-minimax` plugin env: `ANTHROPIC_BASE_URL`, `ANTHROPIC_MODEL=MiniMax-M3`, `ANTHROPIC_API_KEY`, `ANTHROPIC_AUTH_TOKEN`\n4. Run `claude --dangerously-skip-permissions` in the worktree\n5. Auto-create a PR once the worker makes commits\n\n## Verify the worker actually uses M3\n\n```bash\n# After ~10s, check the tmux env (proves the plugin injected M3)\ntmux show-environment -t 953501c04ccc-ao-XXXX 2>&1 | grep -E \"ANTHROPIC_BASE_URL|ANTHROPIC_MODEL|MINIMAX_API_KEY\"\n# Expected: ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic\n#           ANTHROPIC_MODEL=MiniMax-M3\n#           (ANTHROPIC_API_KEY \u002F ANTHROPIC_AUTH_TOKEN present but redacted)\n\n# Read the worker's screen\ntmux capture-pane -t 953501c04ccc-ao-XXXX -p | tail -30\n# Expected: a response from MiniMax-M3, with \"ctx ##-------- XX%\" in the footer\n```\n\n## Common pitfalls (caught in production 2026-06-13)\n\n### Drift: `claudem()` shell function is hardcoded to M2.5\n\n`~\u002F.bashrc` has `claudem() { ... ANTHROPIC_MODEL=\"MiniMax-M2.5\" ... claude --model MiniMax-M2.5 ... }`. If you call `claudem` directly (outside of AO), you get M2.5, not M3.\n\n**For AO workers this doesn't matter** — AO uses the plugin, not `claudem`. But if you're verifying with the shell, use raw `claude --model MiniMax-M3 ...` instead of `claudem`.\n\n### Drift: launchd plists hardcode M2.7\n\n`launchd\u002Fai.agento.lifecycle-all.plist.template` and `ai.agento.health.plist.template` both set `\u003Ckey>MINIMAX_MODEL\u003C\u002Fkey>\u003Cstring>MiniMax-M2.7\u003C\u002Fstring>`. The lifecycle worker spawned from these will use M2.7, NOT M3.\n\n**Workaround:** spawn via `ao spawn --agent minimax` from a shell, which uses `process.env.MINIMAX_MODEL` (= MiniMax-M3 from bashrc).\n\n**Proper fix (not done in this turn):** update the two plist templates to use `MiniMax-M3` or read from a sed-substituted var.\n\n### Drift: autonomous-harness CLI defaults hardcode M2.7\n\n`autonomous-harness\u002Fsrc\u002F{cli,harness-state,orchestrator}.ts` all default `minimax\u002FMiniMax-M2.7`. Anyone running the autonomous-harness without overriding `--generator-model` will get M2.7.\n\n### 401 errors\n\nIf the worker shows `\u002Flogin · API Error: 401` in its tmux pane, the API key is redacted. Use `minimax-401-diagnostic`.\n\n## Verification (proven by this skill)\n\n| Step | Result | Date |\n|------|--------|------|\n| Direct M3 API call | `model: MiniMax-M3, text: PONG_M3_OK` | 2026-06-13 |\n| `ao spawn --agent minimax` | Session `ao-6355` created | 2026-06-13 |\n| Plugin env in tmux | `ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic`, `ANTHROPIC_MODEL=MiniMax-M3`, key resolved | 2026-06-13 |\n| Worker model identity | `Cogitated for 7s` → `PONG_M3_FROM_MINIMAX_WORKER` | 2026-06-13 |\n| PR auto-created | https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fagent-orchestrator-ts\u002Fpull\u002F678 | 2026-06-13 |\n\n## Canonical sources\n\n- **Plugin source:** `~\u002Fproject_agento\u002Fagent-orchestrator\u002Fpackages\u002Fplugins\u002Fagent-minimax\u002Fsrc\u002Findex.ts`\n- **User config:** `~\u002Fagent-orchestrator.yaml` (lines ~119-128: `modelByCli.minimax.model: MiniMax-M3`)\n- **Companion skill (401):** `~\u002F.claude\u002Fskills\u002Fminimax-401-diagnostic\u002FSKILL.md`\n- **AGENTS.md (fork):** `~\u002Fproject_agento\u002Fagent-orchestrator\u002FAGENTS.md` (Provider Agent Plugins section)\n- **This skill (pointer):** `~\u002F.hermes_prod\u002Fskills\u002Fao-spawn-minimax-worker\u002FSKILL.md` (and `~\u002F.hermes\u002Fskills\u002F...`)\n\n## Pitfalls (anti-pattern guards)\n\n- **Don't** use `claudem` to verify M3 — it has M2.5 hardcoded; use raw `claude --model MiniMax-M3` or just check tmux env\n- **Don't** trust the autonomous-harness defaults (M2.7); pass `--generator-model minimax\u002FMiniMax-M3` explicitly\n- **Don't** modify `~\u002FLibrary\u002FLaunchAgents\u002Fai.agento.lifecycle-all.plist` to fix the M2.7 → M3 drift; update the `.template` file and re-run `setup-launchd.sh` instead\n- **Don't** spawn without verifying env first — if `MINIMAX_API_KEY` is missing, the worker will hit 401 and the babysit loop will be wasted\n",{"name":5,"context":61,"description":7,"when_to_use":62,"allowed-tools":63},"Verified end-to-end on 2026-06-13 with session ao-6355 and PR https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fagent-orchestrator-ts\u002Fpull\u002F678. The agent-minimax plugin (packages\u002Fplugins\u002Fagent-minimax\u002Fsrc\u002Findex.ts) wraps the `claude` CLI with ANTHROPIC_BASE_URL=https:\u002F\u002Fapi.minimax.io\u002Fanthropic and propagates MINIMAX_MODEL from the tmux env. modelByCli.minimax.model=MiniMax-M3 in ~\u002Fagent-orchestrator.yaml makes this the default. The skill is intentionally a thin pointer — the canonical implementation is the AO plugin source + the user's `~\u002Fagent-orchestrator.yaml`.","Use when the user asks to spawn an AO worker on minimax M3, or any time `ao spawn` is needed and the model choice is minimax. Do NOT use for: direct `claude` CLI runs (use shell directly), or for AO workers that should use a different model (use the default `ao spawn` with no `--agent` flag). NOT for fixing 401 errors (use `minimax-401-diagnostic` instead).",[47,64,65,66],"Read","Write","Edit","main","42ccd02cbf5e8e47839dc3eab6fc754dc6b339cf","hermes\u002Fskills\u002Fao-spawn-minimax-worker",0,"https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills","https:\u002F\u002Fgithub.com\u002Fjleechanorg","2026-11-11 22:19:36",{"source":8,"skillName":5,"commands":75,"prompt":85,"zipAvailable":86},[76,79,82],{"tool":77,"command":78},"npx","npx skills add jleechanorg\u002Fjleechan-skills --skill ao-spawn-minimax-worker",{"tool":80,"command":81},"bunx","bunx skills add jleechanorg\u002Fjleechan-skills --skill ao-spawn-minimax-worker",{"tool":83,"command":84},"pnpm","pnpm dlx skills add jleechanorg\u002Fjleechan-skills --skill ao-spawn-minimax-worker","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002Fhermes\u002Fskills\u002Fao-spawn-minimax-worker\u002FSKILL.md",true,[],[89,104,126,140,154,169],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":56},"01M716Y5MTJ6GYJZ2ATXCJK8G4","instinct","Use when the user invokes \u002Finstinct or \u002Fin, or asks to \"message Instinct\", \"ask Instinct\", \"tell Instinct\", \"check Instinct\", or read replies from the configured peer assistant. iMessage first, WhatsApp Web as fallback.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Finstinct\u002FSKILL.md",[],[91,96,33,97,98,40,99,100,101,43,102,103],"user","replies","check","identity","whatsapp","gmail","message","ask",{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":115,"publishedAt":56},"01M716Y4HF7FVVMJSA2CCVN5VF","hr","Compatibility alias for history-resume; produce the same fresh-state, provenance-aware resume briefing.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhr\u002FSKILL.md",[110],{"categoryNo":111,"categoryName":112,"taxonomyType":113,"taxonomyTypeDesc":114},"01JBSEEDCAT000000000000038","人力资源","PROFESSION","适用职业",[106,116,117,118,119,120,121,122,123,124,125],"resume","compatibility","provenance","for","history","produce","alias","briefing","state","fresh",{"skillNo":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":133,"publishedAt":56},"01M716Y1FFNFKFJYYKF8HQA6T8","history-resume","Resume briefing for the current repository branch or worktree using fresh state and admitted history sources.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhistory-resume\u002FSKILL.md",[132],{"categoryNo":111,"categoryName":112,"taxonomyType":113,"taxonomyTypeDesc":114},[120,116,134,135,96,136,123,137,119,138,139,124,125],"history resume","branch","analysis","repository","worktree","sources",{"skillNo":141,"name":142,"nameZh":6,"description":143,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":144,"categories":145,"keywords":146,"publishedAt":56},"01M716XWXJ8EPHK4RH9T14TR85","grok","Alias for the canonical grokbot browser conversation workflow; \u002Fgrok talks to the Grok bot the same way \u002Fdot talks to Dot.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fgrok\u002FSKILL.md",[],[142,147,148,149,119,150,122,151,152,153],"dot","bot","workflow","canonical","conversation","browser","grokbot",{"skillNo":155,"name":156,"nameZh":6,"description":157,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":158,"categories":159,"keywords":160,"publishedAt":56},"01M70N9Q07DTE0EW5KFXMMMSJ0","codexa-bench","Compare `codexa` and normal `codex` on a fair split of comparable work by default; use `--strict` for duplicate runs of the exact same task.","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fcodexa-bench\u002FSKILL.md",[],[161,162,163,40,39,119,164,165,166,167,168],"codexa bench","work","split","duplicate","task","compare","normal","codex",{"skillNo":170,"name":171,"nameZh":6,"description":172,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":173,"licenseFileUrl":174,"sourceHtmlUrl":175,"categories":176,"keywords":186,"publishedAt":56},"01M6C24S2X60SP8BBFGAEKPEGW","frontend-design","Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML\u002FCSS layouts, or when styling\u002Fbeautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ffrontend-design\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fjleechanorg\u002Fjleechan-skills\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ffrontend-design\u002FSKILL.md",[177,180,183],{"categoryNo":178,"categoryName":179,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000009","前端开发",{"categoryNo":181,"categoryName":182,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":184,"categoryName":185,"taxonomyType":113,"taxonomyTypeDesc":114},"01JBSEEDCAT000000000000032","设计师",[187,188,189,40,190,191,192,96,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,33,211],"frontend","design","frontend design","generic","ui design","react","html","components","code","artifacts","ui","aesthetics","Production","web components","landing","applications","quality","web","css","create","examples","with","react components","web ui","pages"]