Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/oneworks-ai/app/agentsgit clone --depth 1 https://github.com/oneworks-ai/appWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.02018 |
| Opus 5 | $0.00000 | $0.01009 |
| Sonnet 5 | $0.00000 | $0.00404 |
| Haiku 4.5 | $0.00000 | $0.00202 |
Grade A, and why
AGENTS scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Commands Middleware
Channel 指令系统的核心中间件，负责解析用户输入的文本指令并执行对应的操作。
架构概览
commands/
index.ts ← 中间件入口，组装所有指令并分发（含权限拦截）
command-system.ts ← 指令类型系统、chain builder、解析器
cmd.general.ts ← 通用指令：help / whoami / lang（含 help 格式化）
cmd.identity.ts ← 身份指令：identity whoami / link / accounts
cmd.session.ts ← 会话指令：session / search / bind / unbind / reset / stop / get / set / permissionMode
cmd.authorization.ts ← 授权请求指令：auth request / list / grant / deny
cmd.channel-control.ts ← 频道过程控制指令：silent / unsilent / stop / start / ban
cmd.access.ts ← 权限指令：access / admins / admin / allow / block
access.ts ← 权限检查 & 频道配置写入工具
tools.ts ← CommandSpec 到 `channel.*` typed tool definition 的转换
i18n.ts ← 国际化注册 API（defineMessages / t / createT）、LanguageCode 类型、系统级共享翻译
utils.ts ← splitCommand / dedupe / choiceParser
核心机制
Chain-style 指令定义
所有指令使用 command() builder 链式定义，而非对象字面量。支持 .alias() / .description() / .adminOnly() / .argument() / .subcommand() / .action() / .build()。需要 action 的指令以 .action() 结尾返回 CommandSpec；纯分组指令（只有 subcommands 没有 action）以 .build() 结尾。
权限自动拦截
每个 CommandSpec 有 permission: 'everyone' | 'admin'（通过 .adminOnly() 设置）。中间件在 dispatch 时会把解析后的 command、当前 sender、channel admin 配置和 source 交给 services/channel-approval 做最小裁决：若指令标记为 admin-only 且当前发送者不是管理员，直接回复权限错误，不会进入 action。cmd 文件中的 action 无需手动检查权限。
Command Run 审计
解析成功的 fast path 指令会写入 channel_command_runs：
- action 成功：
success - admin-only 被拦截：
denied - action 抛错：
failed
未知指令会继续透传给后续 middleware，不写 command run。解析到已知指令但参数错误时当前仍只回复用法，不写 run；后续若需要完整审计再扩展 parse-error run。
channel_command_runs.metadata.approval 会记录最小 resolver 的裁决摘要，包括 status、reasonCode、capability、actor id 和 authorizationRequestId（如有）。这只是审计摘要，不是 credential secret，也不代表已经获得用户个人登录态。
Typed Tool Registry
tools.ts 将现有 CommandSpec 树转换成 channel.* typed tool definition，并提供 tool input 到 CommandParseSuccess 的解析。registry 描述工具名、command path、slash usage、参数 JSON schema、权限等级和 actorAuthority: 'sender'；执行仍回到 index.ts 的统一 command runner，所以 typed invocation 和 slash fast path 共享权限拦截、action 和 channel_command_runs 审计。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 99 lines · 0 tokens per session scan A 4de1ae1c9dd6
AGENTS is a command published in the GitHub repository oneworks-ai/app (18 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,018 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
translate
Translate the following documentation files to all supported languages.
mem-help
Display localized help for SecondBrain mem- commands.
spec-design
Create comprehensive technical design for a specification.
translation-review
全面审核翻译质量: 对比原版、专项检查、质量评分与发布验证, 发现问题时自动衔接修复流程.
03-gap
Translate the visibility baseline + brand context into a ranked list of content actions that should move the needle. Identifies absent queries, contested queries, and competitor-displacement opportunities. Outputs contentpriorities.json validated against schema. Use after 02-audit; consumed by 04-content-brief.
claude-request
Send a prompt to Claude Code via the LLM gateway with session tracking.