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 instructions/v833/q-code/agents-mdgit clone --depth 1 https://github.com/v833/q-codeWhat 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 | $0.10314 | $0.10314 |
| Opus 5 | $0.05157 | $0.05157 |
| Sonnet 5 | $0.02063 | $0.02063 |
| Haiku 4.5 | $0.01031 | $0.01031 |
Grade A, and why
q-code AGENTS.md 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 — 215 lines — stays where its author put it; the contents beside it link to each section on GitHub.
q-code 项目协作说明
项目概览
q-code 是一个基于 Vercel AI SDK 的 TypeScript 命令行 Agent 框架。核心能力包括:
- Agent / 任务:Agent Loop、Plan Mode(自然语言审批、智能进入与 Shift+Tab 切换)、Task V2、TodoWrite、上下文压缩、会话持久化与 TUI
/sessions管理、文件历史快照与/rewind按轮次回滚内置写工具改动、TUI/agentsSubAgent Monitor(同步与后台 SubAgent,completed 默认隐藏并可清理,忙碌等待期间可用Ctrl+A查看)、只读同步 SubAgent 动态并行调度、SubAgent 长 final output artifact/preview 回传、TUI 输入历史跨进程持久化、@file文件引用注入与候选索引缓存/监听刷新、@image:<path>/ TUI 图片附件多模态输入、文件派项目记忆(headers 精选、预算化正文注入、年龄提示)、Skills、SubAgent、Agent Teams、Worktree 隔离。 - 工具执行:文件/搜索工具、可配置超时与 spill 的 Shell 工具(Windows 优先 PowerShell7,缺失时回退 Windows PowerShell 5.1)、后台 Shell job(
f_status/f_tail/f_kill/f_list)。 - 集成扩展:MCP server、ACP v1 stdio Agent 接入、Hooks(生命周期事件、pre/post tool-use 决策、prompt/context 注入、退出码协议)、Slash 命令注册表、Output Styles、Markdown User Commands、企业 AI 基建同步(Infra)、GitLab Wiki 知识库。
- 可观测性:NDJSON 审计日志(默认开启)、本地只读 Web Dashboard、模型等待心跳、
ttftMs/elapsedMs/TPS step 诊断、可选 Langfuse/OpenTelemetry trace 导出、崩溃保护(crash guard,默认开启)与 crash report、Usage / Cache / 成本统计、上下文占用预警、启动时版本更新说明(对比~/.q-code/last-version.json与包内changelog.json)。 - 评测:
q-code eval本地优先 Agent 质量平台,覆盖固定任务集、mock/cli/真实模型 runner、LLM judge(opt-in)、工具轨迹、预算/成本、进度、文件副作用、策略安全、JSONL trace、Markdown/JUnit 报告、baseline 对比、趋势看板、定期回归与可选 Langfuse evaluator trace / dataset / scores 导出。 - TUI:基于 Ink 的交互式 TUI(默认)、流式 Markdown 稳定前缀渲染、Markdown 快路径与 LRU parse cache、
--classic经典 readline、可经管道/CI 自动降级;主 Agent 默认人格为「小黄鸭」,可用/ya主动切换到主题鸭「降压鸭」「屁老鸭」。 - CLI 子命令:
q-code help|version|update|audit|init|eval|dashboard|exec|acp(启动前 short-circuit);exec提供 Codex CLI 兼容的无头 JSONL、session resume 和常用参数;acp以 ACP v1 stdio Agent 身份提供 session/prompt;其余参数走主交互循环。
环境与工具
- 运行时:Node.js 22+。
- 包管理器:pnpm。不要混用 npm/yarn 生成新的 lockfile。
- 源码直接通过
tsx运行,项目为 ESM:package.json中"type": "module"。 - TypeScript 严格模式开启,模块解析为
bundler,目标为ES2022。 - 本仓库存在
.env,其中可能包含本地敏感配置;不要在回复、日志或提交中暴露密钥明文。
常用命令
pnpm install
pnpm start # tsx src/index.ts
pnpm continue # tsx src/index.ts --continue
pnpm docs:dev # VitePress 文档站本地预览
pnpm docs:build # 构建 docs/.vitepress/dist
pnpm docs:preview # 预览文档站构建产物
pnpm typecheck # tsc --noEmit
pnpm test # vitest run(unit + integration)
pnpm test:unit # vitest run tests/unit
pnpm test:integration # vitest run tests/integration
pnpm test:acp # ACP 参数、内容转换和 stdio 握手测试
pnpm test:watch # vitest watch
pnpm test:coverage # vitest run --coverage
pnpm test:legacy # 串行:test-mcp + test-skills + test-agents + test-async-agents + test-teams
pnpm test:mcp
pnpm test:skills
pnpm test:agents # test-agents + test-async-agents
pnpm test:teams
pnpm test:infra-candidate
pnpm eval:smoke # 运行 deterministic smoke eval,不导出 Langfuse
pnpm eval:cli # 运行 cli-subprocess fixture eval,不导出 Langfuse
pnpm eval:ci # 运行 smoke + cli eval,并输出 JUnit 报告
pnpm eval:smoke:langfuse # 运行 smoke eval 并按配置导出到 Langfuse
pnpm eval:nightly # 运行定期 deterministic 回归并生成趋势看板
pnpm eval:trend # 从历史 eval runs 生成本地趋势看板
pnpm eval:compare # 对比两个 eval run
pnpm prompt:cache:verify # 本地验证稳定 system prompt hash 与 90%+ 前缀目标
pnpm prompt:quality:verify # 本地审计 Agent prompt 12 维质量基线
pnpm test:all # pnpm test && pnpm test:legacy
pnpm precommit # typecheck + test:unit
pnpm build # 调 scripts/build.mjs:自动生成 changelog 并产出 dist/
pnpm changelog # 手动从 git tag / conventional commit 生成 CHANGELOG(调试用)
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 · 215 lines · 10,314 tokens per session scan A 14a403217059
q-code AGENTS.md is an instructions file published in the GitHub repository v833/q-code (9 stars, last pushed 21d ago), licensed MIT. It adds 10,314 tokens to every session, about $0.0516 per session on Opus 5. 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 2026-08-31.
Other instructions, from other repositories
open-code-review AGENTS.md
Instructions for alibaba/open-code-review, covering agent guidelines for open-code-review, project overview, git commit notes, license headers and code style.
learn-harness-engineering CLAUDE.md
Instructions for walkinglabs/learn-harness-engineering, covering claude.md, project overview, commands, documentation site and run lecture code examples.
open-code-review CLAUDE.md
Instructions for alibaba/open-code-review: See AGENTS.md for all project guidelines.
trueforge AGENTS.md
Instructions for truefoundry/trueforge, a project described as: The open-source agent harness - the runtime layer that turns an LLM into a working agent.
cs CLAUDE.md
Instructions for boyter/cs, covering claude.md, project overview, build & test commands, architecture and query pipeline (pkg/search/).
happyclaw AGENTS.md
Instructions for riba2534/happyclaw, covering happyclaw agent instructions and definition of done.