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 skills add SanAntonio021/claude-codex-bridge --skill cross-model-orchestrationgit clone --depth 1 https://github.com/SanAntonio021/claude-codex-bridgeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sanantonio021/claude-codex-bridge/cross-model-orchestration)<a href="https://agentmods.dev/skills/sanantonio021/claude-codex-bridge/cross-model-orchestration"><img src="https://agentmods.dev/badge/skills/sanantonio021/claude-codex-bridge/cross-model-orchestration/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sanantonio021/claude-codex-bridge/cross-model-orchestration"><img src="https://agentmods.dev/badge/skills/sanantonio021/claude-codex-bridge/cross-model-orchestration.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00194 | $0.04098 |
| Opus 5 | $0.00097 | $0.02049 |
| Sonnet 5 | $0.00039 | $0.00820 |
| Haiku 4.5 | $0.00019 | $0.00410 |
Grade A, and why
cross-model-orchestration 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 9d ago.
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 — 217 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex / Claude 双向互审
目标与触发
把跨模型复核限定为正式计划的默认质量门。作者先形成将要提交给用户确认的计划草案；对方模型 在 bridge 创建的固定副本中检查、允许范围内修订并运行必要测试。审查者不能直接写作者主项目， 作者必须检查 bridge 同步后的计划状态，再向用户展示最终计划。
只有以下情况自动进入：
- 当前处于 Plan Mode，准备退出并向用户提交正式计划；
- 即将展示一份多步骤实施方案，而且后续执行明确需要用户确认；
- 已有稳定的计划文件或完整计划正文，正在进入计划确认门。
以下情况不自动进入：读取或分析材料、调研、写作、代码/文档修改、运行命令、测试、提交、执行
已经确认的计划、验收和最终汇报。内部 Todo、update_plan、执行清单、简短步骤说明也不算正式
计划。任务复杂、耗时或交付物重要都不是触发理由。边界不清时只判断“是否正准备提交一份需要
用户确认的正式计划”，不能按复杂度扩大触发范围。
用户明确要求“让 Opus/Codex 审查”、明确指定 review_repair、明确要求对方执行，或明确启用跨模型
执行/科研循环等专用流程时，可以显式审查或使用 task；已确认计划中明确约定的对方执行也可以
继续。这些属于用户点名能力，不是全局自动复核门。审查报告、失败报告和用户已裁决的分歧报告不
递归触发。
ExitPlanMode 的拒绝理由只是 Claude 表达“先审查计划”的一种方式，不是唯一触发器，也不是
可信的计划路径来源。没有可靠路径时，使用稳定 artifactName 和完整 artifactContent，
绝不猜测“最新计划”。
统一入口
两端都使用同一个 MCP，但必须连接到与当前作者匹配的 loopback 角色端点。端点身份决定对方模型，
调用方不能在参数中伪造 target、owner、operation、round 或权限字段：
| 当前作者 | MCP URL | 认证环境变量 | 端点推导的 reviewer |
|---|---|---|---|
| Codex Desktop / Codex CLI | http://127.0.0.1:43123/mcp/codex |
CLAUDE_CODEX_BRIDGE_CODEX_TOKEN |
Claude |
| Claude Code CLI | http://127.0.0.1:43123/mcp/claude |
CLAUDE_CODEX_BRIDGE_CLAUDE_TOKEN |
Codex |
正式互审的调用顺序固定为 review_peer 或 review_repair_peer -> await_peer（单次最多 45 秒）
-> peer_result；首次进入或考虑 workspace 模式时，先用 peer_status 读取能力；第三轮用户裁决用
adjudicate_peer_series。不得扫描最新线程、直接调用 /mcp 兼容端点、codex exec、claude -p
或 codex@openai-codex 绕过 bridge。orchestration-control.mjs 与 check-resume-candidate.mjs
只用于历史状态测试和输入诊断，不是运行时入口。
能力门固定如下：peer_status.active=true 和 capabilities.inlineReviews=true 是所有 v2 inline
审查的前提；review_repair_peer artifactMode=workspace 还必须同时满足
capabilities.workspaceRepairs=true 与 workspaceProbeState=available。pending 或 unavailable
只表示当前进程尚未证明 Windows workspace sandbox，不影响零工具 inline 审查。
review_peer 固定为 review_only + inline：只读、零工具、不得声明工作区或测试命令。它是纯文字
正式计划和普通审查在 workspace 能力不可用时的默认入口，作者自行采纳修改并重新计算下一轮身份。
需要完整替换正文时使用 review_repair_peer artifactMode=inline；它也不使用工具，返回完整
repairedArtifact。workspace 修复只允许显式 repairTargets，并由 bridge 检查、同步和回滚。
如果用户明确要求 workspace 修订或结构化测试而能力为 pending/unavailable，不得提交该请求、
不得创建 job、不得重启或改用其他 sandbox；输出 PEER_REVIEW_FAILURE_REPORT，
decisiveError=v2_workspace_capability_unavailable。未经用户重新选择，不把 workspace 请求静默改成
inline 修订。旧 submit_peer(operation=review_repair) 只保留一版兼容，正式流程不得继续使用；兼容调用
缺字段时必须在创建 job 前返回 missing_fields。
What ships with it
18 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- evals/evals.json 8.7 KB
- evals/fixtures/acceptance-criteria.md 116 B
- evals/fixtures/reference.md 96 B
- evals/fixtures/sample-skill-alpha.md 140 B
- evals/fixtures/sample-skill-beta.md 155 B
- evals/fixtures/simulation-results.csv 47 B
- evals/fixtures/technical-note.md 210 B
- evals/integration-cases.md 6.4 KB
- evals/integration-results-2026-07-11.md 9.6 KB
- evals/trigger-evals.json 2.3 KB
- references/model-routing.md 2.9 KB
- references/upstream-sources.md 599 B
- references/workflow-contract.md 14 KB
- scripts/check-resume-candidate.mjs 3.9 KB runs code
- scripts/orchestration-control.mjs 43 KB runs code
- scripts/v2-envelope.mjs 7.6 KB runs code
- test/public-skill.test.mjs 2.0 KB runs code
- tests/orchestration-control.test.mjs 27 KB runs code
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.
- 9d ago First seen · 217 lines · 194 tokens per session scan A ed9c7b39e584
cross-model-orchestration is a skill published in the GitHub repository SanAntonio021/claude-codex-bridge (0 stars, last pushed 24d ago), licensed MIT. It adds 194 tokens to every session and 4,098 once invoked, about $0.0010 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 skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…