Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Briefbound Autonomous Collaboration Loop

ASecurity

Use when the user explicitly enables an automated same-project multi-thread development loop that keeps peer Codex tasks progressing, recovers conflicts or stale coordination, integrates verified work into local main, and closes cleanup without repeated gates; do not use before opt-in, for one-off coordination, remote publication without permission, or subagent creation.

6 stars
0 votes
0 copies
1 views
Added 9/19/2026
ai-agentsgogit

Security Analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned 10/5/2026

$npx -y skills add CCDawn/codex-skills --skill briefbound-autonomous-collaboration-loop --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Briefbound Autonomous Collaboration Loop?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Briefbound Autonomous Collaboration Loop
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/ccdawn-briefbound-autonomous-collaboration-loop/badge)](https://www.skillsdirectory.com/skills/ccdawn-briefbound-autonomous-collaboration-loop)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: briefbound-autonomous-collaboration-loop
description: "Use when the user explicitly enables an automated same-project multi-thread development loop that keeps peer Codex tasks progressing, recovers conflicts or stale coordination, integrates verified work into local main, and closes cleanup without repeated gates; do not use before opt-in, for one-off coordination, remote publication without permission, or subagent creation."
license: MIT
---

# Briefbound 自动协作开发闭环

## 目标

一次确认后,驱动同项目平级会话完成、验证、合入 `main`;不建主从或接管任务。

## Briefbound task contract

- Context Boundary: 已对齐目标、授权的现有同项目 thread、各自 scope/branch/worktree、共享契约、目标 `main`/验证门。
- Output Contract: agreement、交付证据、integration queue、本地 `main` 验证和恢复清理。
- Allowed Action: 在授权范围内读取 thread 状态、维护本任务 registry、执行本地写入/提交/合入 `main`、验证和安全清理;当前 `BRT_TRANSPORT_DISABLED`,不得发送 thread 消息。新建会话、远程 push/PR merge/发布、破坏性或高风险动作仍需单独授权。
- Success Evidence: 每个原任务达到验收,目标 `main` 包含预期交付,集成门通过,open coordination、恢复债务和已知开发残留清零。
- Stop Condition: 未获开启确认、共同目标或 target 不明、产品/安全/数据迁移取舍未决、需要远程授权、破坏性动作,或自动恢复连续失败。
- Route Out: 各 Agent 原 owner、`briefbound-multi-agent-orchestration`、`briefbound-thread-coordination`、`briefbound-pr-review`、`briefbound-development-cleanup`、`briefbound-project-memory`、`briefbound-router` 或 BLOCKED。

## 统一调用契约

- 用户可见内容默认中文;只报行动决定、blocker 和最终证据。Route Out 仅以 Briefbound task contract 为准;有自然闸门(需用户裁决、批准或被外部阻塞)时末行 `下一步建议: <一个具体动作>`,否则声明继续已授权工作。
- 各 Agent 仍负责自己的任务、scope、branch;Loop Owner 只维护闭环。
- 开启后不在每个阶段、task、普通冲突或恢复动作后询问是否继续。

## 开启与授权

Briefbound Router 仅在非简单目标确有多会话协作价值时,询问一次明确列出项目、既有参与会话、integration target 和本地集成范围的授权。确认后才可联系这些现有会话、协商、安全开发、验证和限定目标的本地集成,直到完成、取消或自然闸门;泛化“继续/确认”不扩大到创建会话或新的 integration target。

没有合适会话时,只询问一次是否创建。远程 push、PR 或破坏性操作不继承授权。自动循环还必须具备已验证的 `BRT_DIRECT_USER_TURN_V1`,且运行环境提供平级多会话原语(见 harness 能力矩阵);任一不满足保持 `BLOCKED_TRANSPORT`,不启动跨任务循环或 outbox。

## 角色与持久状态

- `Loop Owner`:维护状态、租约、outbox、恢复债务和完成门,不代写 peer 结论。
- `Peer Owner`:完成任务,提供 scope、依赖和 `MERGE_READY`。
- `Integration Owner`:由任一合格 peer 在首个 `MERGE_READY` 出现后可原子认领,维护当前队列;只有目标干净、授权范围匹配且重验通过时才合入 `main`。claim 有效期间不启动新的无关实现任务,变忙时先转交队列。
- `Recovery Dispatcher`:任何活跃参与者在 Loop Owner 租约过期时可 `takeover`,继承 outbox、恢复和集成义务,但不能继承或伪造 peer 的完成声明。

状态为 `DISCOVER -> AGREEMENT -> RUNNING/NEGOTIATING -> MERGE_READY -> INTEGRATING -> INTEGRATED -> CLOSED`。每个 peer 只记 `Task / Scope / Branch or Worktree / Dependency / Checkpoint / Tests / Blocker / Resume Debt`;优先 registry,跨会话恢复才用 memory。

## 自动循环

### 1. 发现与约定

只读最多 3 个相关 thread,确认收益、边界、契约、依赖和 Integration Target,用 `briefbound-multi-agent-orchestration` 建 agreement。Integration Owner 可等首个交付入队后认领;无收益则返回原 owner。

### 2. 并行推进

各 peer 继续非冲突任务。`BRT_TRANSPORT_DISABLED` 期间不得通信、轮询、入队或催促;`MERGE_READY` 只由原任务交付或用户直接传达。未来恢复后,消息默认只发差量:`From/Task / Changed Fact / Action Impact / Evidence Pointer / Reply Needed`;背景、方案和测试用指针,首次契约或安全语义才展开。

### 3. 冲突协商

普通重叠先缩 scope、调顺序或交换接口,由 `briefbound-thread-coordination` 复用 discussion。仅无法拆分且继续写会立即覆盖/回归时暂停冲突 surface,记录 `resumePendingAgentIds`;解决后发 `CONFLICT_RESOLVED` 并确认恢复。

`main` 变化、排队或 gate stale 不暂停。Peer 聚焦验证并提交 `MERGE_READY` 后结束执行,由 Integration Owner 集成;不等待 stable main,不反复 closeout。

### 4. 故障与接管

每次收发在未来恢复后均先通过 `briefbound-thread-coordination` 的 `BRT_DIRECT_USER_TURN_V1` 验收门:只允许同项目、fresh `read_thread` 返回精确 `threadId` 的平级顶层 thread;原生子 agent 不得加入 outbox 或 relay。当前不得调用 `send_message_to_thread`,不得保存 outbox,也不得依据 `/root`、agent 标签或旧 registry 重新寻址。Loop Owner 失活时只记录交接事实,不自动唤醒其他任务。

自动恢复先分类为任务、依赖、集成或环境失败;证据发给 owner,其他 peer 继续。只有高风险取舍或同一 blocker 经两轮不同恢复策略仍失败才询问用户。

接管者只继承推进义务,不能继承证据所有权。失活 peer 没有正式 `MERGE_READY` 时:

- Git commit/artifact、scope 和新鲜验证足以独立重验:接管者可标记 `MERGE_READY_RECOVERED`,必须注明原 owner 未确认及重验证据。
- 只有聊天中的采纳、计划、进度或草稿:标记 `RECOVERY_PENDING_EVIDENCE`;接管者自行完成缺口或恢复原 owner,不能代其宣告完成。
- 无法安全完成且两轮恢复失败:保持明确 owner/blocker 后再询问用户,不把缺证据项目合入。

### 5. 集成队列

每个 peer 用 `MERGE_READY` 入队,完整证据字段规范见 `briefbound-multi-agent-orchestration` 的 `team-protocol.md`。`MERGE_READY_RECOVERED` 也必须满足相同证据门。Integration Owner 以 Git 和新鲜测试为事实源:

首个交付入队时检查一次 `lane=integration/<target-key>`:无 claim 则认领队列并发 `INTEGRATION_CLAIMED`。baseline/dirty main 时进入 `WAITING_FOR_CLEAN_TARGET`,保留证据但不应用、变基或合并;已有负责人则其他 peer 停止合并。Loop Owner 对空缺自行认领或只联系一个合适 peer,不轮询或询问用户。细节交 `briefbound-thread-coordination`。

1. 无依赖且不重叠的交付可成组;共享契约或依赖项串行。
2. Peer 只需聚焦验证;Integration Owner 对过期 base 变基或应用提交,重验受影响窄面。机械冲突可处理,语义冲突才回相关 peer。
3. 全部预期交付合入本地 `main` 后只运行一次完整集成 gate;失败才重开责任任务和受影响依赖。
4. 不用远程 CI 替代本地证据,也不在未授权时 push 或操作远程 PR。

gate 期间若无关任务推进 `main`,不冻结全项目:比较新增提交与本轮 scope;无重叠则在最终 HEAD 补受影响的窄验证,有语义重叠才重跑对应 gate。不得因无关漂移重跑整套门禁。

可重建的 ignored/untracked 测试缓存不阻塞 `MERGE_READY`,统一留给 final cleanup;删除受阻且不影响 tracked 交付时保留,不更换多种命令追查。

### 6. 收尾

验收和集成 gate 通过后,广播一次 `INTEGRATED`,释放 integration claim 并关闭 merge coordination;债务清零后用 `briefbound-development-cleanup` 清理安全残留,最后只做一次整体汇报。

## 完成门

以下条件必须同时成立:

- 每个 peer 的原任务已完成或由用户明确取消;“消息已发送”“已采纳”“接近完成”和恢复生成的总结都不能代替交付证据;
- 本地 `main` 包含所有预期交付且集成验证通过;
- 没有有效或遗留 integration claim、未处理 outbox、open coordination、stale owner 或 `resumePendingAgentIds`;每个 `TRANSPORT_EXPIRED` 都有 fresh list/read 证据;
- 清理没有删除未合并工作,远程动作仍保持未执行;
- 失败和 Deferred 项有明确 owner、证据和触发条件。

未满足时由当前 Loop Owner 或接管者继续闭环,不把普通协调债务抛回用户。

Attribution

CCDawnCCDawn
View sourceSee grades on GitHubMore from CCDawn →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →