jleechanbrain

Use when dispatching work through the Hermes gateway with /claw, especially when the task may resolve slash commands or hand off into AO worker orchestration.

12-hour autonomous evolution loop — observe, measure, diagnose, fix, repeat every 10min. Drives zero-touch rate up by finding friction and dispatching fixes.

Custom evolve loop for jleechanbrain orchestrator — drains dropped Slack thread backlog via /claw, fixes hermes issues, proposes new work items. Max 50 items, newest-first.

jleechanbrain / Hermes-specific harness — gateway, canary, deploy, lane backlog, staging vs prod. Use with user-scope harness-engineering.

Delegate coding change tasks to Agent-Orchestrator (AO). Triggered by the keyword "agento" anywhere in a message, or by default for any coding task. Never use mctrl unless explicitly requested.

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 / PR #678.

Enforce headless browser automation by default for Hermes — Playwright MCP and superpowers-chrome. Use when opening browsers, scraping, UI tests, localhost verification, or any chrome_use_browser / Playwright call. Never headed unless Jeffrey explicitly requests visible browser.

Use the browserclaw CLI to capture browser traffic, infer API endpoints, generate Python clients, OR decrypt + inject Chrome cookies for session reuse. Use when asked to reverse-engineer a website, generate an API client from a HAR, or reuse a logged-in Chrome session in a Playwright browser.

Read and report token usage stats from Hermes gateway logs. Use when asked "how many tokens", "token breakdown", "context size", "what's using my context", or any question about token/component breakdown in the current session or across recent sessions.

Diagnose Hermes gateway health, Hermes monitor issues, and launchd service failures. Run when the user shares a monitor report, says "is everything ok", "check Hermes", "hermes is down", or when a launchd service exits with code 127 (command not found).

Diagnose why a Slack reply "didn't go to the right thread" — covers FIVE known failure modes (gateway self-threaded post, runtime tool-surface gap, gateway scratch-leak during probing, tool-call narration leak, AND wrong-thread_ts from a stale session context header) and the durable direct-HTTP fallback path. Use when a Slack reply lands as a top-level channel message instead of a thread reply, when the bot's thinking trace leaks into the wrong thread, when mcp__slack__conversations_add_message is missing from the runtime tool list, when the session context header reports a thread_ts that turns out not to be the user's actual thread, or when the user says "I thought we fixed this?" / "isn't this patched?". NOTE: the `send_message` thread_ts-drop bug (Failure 1) was FIXED by hermes-agent PR #29 on 2026-06-14 — see the STATUS UPDATE banner; the still-open mode is Failure 5 (wrong thread_ts in the session context header).

Dispatch a bead-tracked task via ao spawn/ao send, register the mapping, and ack in Slack thread.

End-to-end finish protocol for any Slack thread, CLI invocation, or cron task where the user has handed off a goal. Routes to /fs (spec gen) → /f (Dark Factory loop) → drives to a verifiable conclusion (green PR with non-unit-test evidence, finished code change, or dry-run to local machine state). Never stops halfway. Loads automatically when the SOUL.md `finish-the-job` commit fires.

Diagnose and trace unwanted/automated comments on GitHub PRs to their source. Use when a PR is getting spammed by bot comments, AI automation directives, or cron-triggered messages, and you need to identify which system is responsible and how to disable it.

Delegate coding change tasks to Agent-Orchestrator (AO). Triggered by the keyword "agento" anywhere in a message, or by default for any coding task. Never use mctrl unless explicitly requested.

Dispatch a bead-tracked task via ao spawn/ao send, register the mapping, and ack in Slack thread.

Stay responsive to new high-priority user messages in the same thread while delegate_task workers run in parallel. Replies to the same thread steer the orchestrator instead of interrupting sibling work.

Thin pointer — canonical /repro workflow lives in WorldArchitect repo at .claude/skills/repro-twin-clone-evidence/SKILL.md. File gh issue, create draft PR, copy campaign, reproduce, verdict.

When a Python requests-based LLM provider silently corrupts multi-byte UTF-8 (em-dash, curly quotes) in streaming SSE responses, the cause is iter_lines(decode_unicode=True) using ISO-8859-1 default. Use when user reports "weird â\x80\x94" characters in LLM output, or when adding new streaming providers.

Drive any PR I own (or am asked to) all the way to a green PR state (CI green, required review status clear, MERGEABLE/CLEAN) and hand off to skeptic-cron.yml for the auto-merge when Jeffrey says "do it directly if its easy" or "next time don't ask, just finish the work". Never stop at local commits, never stop at "ready for review", never run `gh pr merge` directly (skeptic-cron owns the merge), never ask "want me to merge?" when the user has already given the go-ahead. Stop-halfway is the exact violation this skill exists to prevent.

为 Hermes 配置 MiniMax 作为模型源。MiniMax 提供两种接入方式：API Key 直连（openai-completions 协议）和 OAuth 门户（anthropic-messages 协议）。包含 provider 注册、模型定义、别名配置、fallback 链接入和验证的完整流程。当管理员说想"加 MiniMax"、"配 minimax"、"接入 MiniMax 模型"、"加海螺模型"、"配 M2.1"时使用此 skill。

Use when asked to control, automate, click/type in, or manage Google Antigravity from Hermes. This Hermes shim skill delegates canonical behavior to the Claude skill at ~/.claude/skills/antigravity-computer-use/SKILL.md.

Delegate coding to OpenAI Codex CLI (features, PRs).

Pixel art w/ era palettes (NES, Game Boy, PICO-8).

Control a running TouchDesigner instance via twozero MCP — create operators, set parameters, wire connections, execute Python, build real-time visuals. 36 native tools.

Webhook subscriptions: event-driven agent runs.

Diagnose and recover dropped Jeffrey messages — threads or standalone messages that got no response within 30 min. Use when Jeffrey asks about missed messages, the script reports drops, or you need to understand why a message was unanswered.

GitHub auth setup: HTTPS tokens, SSH keys, gh CLI login.

Review PRs: diffs, inline comments via gh or REST.

Create, triage, label, assign GitHub issues via gh or REST.

GitHub PR lifecycle: branch, commit, open, CI, merge.

Clone/create/fork repos; manage remotes, releases.

Prepare briefings for today's meetings — attendee research, email history, past meeting notes, LinkedIn, and company context. Use when running the daily meeting prep cron, or when user asks to prepare for meetings, review who they're meeting with, or get context on upcoming calls.

Route tasks to the launchd-managed Symphony daemon when the user says "sym".

Run multi-hour autonomous coding tasks using a 5-agent harness: Researcher, Planner, Executor (with Opus advisor), Evaluator, and Orchestrator. Combines Anthropic's advisor strategy, GAN-inspired generator/evaluator design, and Boris Tane's annotation-cycle workflow.

MCP client: connect servers, register tools (stdio/HTTP).

Search/download GIFs from Tenor via curl + jq.

Build, test, and debug Hermes Agent RL environments for Atropos training. Covers the HermesAgentBaseEnv interface, reward functions, agent loop integration, evaluation with tools, wandb logging, and the three CLI modes (serve/process/evaluate). Use when creating, reviewing, or fixing RL environments in the hermes-agent repo.

Read, search, create, and edit notes in the Obsidian vault.

Airtable REST API via curl. Records CRUD, filters, upserts.