/install acpx-codex-playbook
acpx-codex-playbook
Use acpx as a structured control plane for Codex. Prefer persistent sessions, prompt files, and shell-based file generation over fragile one-shot prompts and tool-native file writes.
Quick start
Run this default flow for any non-trivial task:
acpx codex sessions new --name task
acpx codex set-mode -s task full-access
acpx codex -s task -f prompt.txt
Prefer this over acpx codex exec ... when the task needs iteration, file output, validation, or retries.
Workflow
1. Choose session type
Use exec only for small one-shot tasks.
Use a persistent session when the task involves any of the following:
- generating files
- multiple retries
- long prompts
- validation steps
- local installs or virtual environments
- deliverables such as
.pptx,.docx, reports, videos, or scripts
2. Set mode explicitly
For practical work, set mode before prompting:
acpx codex set-mode -s task full-access
Interpretation:
read-only: inspect onlyauto: moderate default behaviorfull-access: broader session capability, including easier file edits and broader path/network freedom
Do not assume full-access means sudo or root. It relaxes the ACP session; it does not guarantee system-level privilege escalation.
3. Use prompt files, not huge shell strings
For long or delicate instructions, always write a prompt file and pass -f:
cat > prompt.txt \x3C\x3C'TXT'
Task: ...
Constraints: ...
Outputs: ...
Validation: ...
TXT
acpx codex -s task -f prompt.txt
This avoids shell quoting failures and makes retries reproducible.
4. Prefer shell/Python file writes over ACP fs writes
If the task must create or rewrite files, instruct Codex to prefer:
- shell heredocs
python - \x3C\x3C'PY' ... PY- direct command-line generation
Prefer these over tool-native fs/write_text_file style edits when prior attempts showed permission failures.
Recommended instruction snippet:
If built-in file-editing tools fail, write files via shell heredoc or Python scripts instead of ACP fs write calls.
5. Use writable output paths first
For fragile generation tasks, write outputs under /tmp first, validate them, then move/copy them into the target workspace.
Recommended pattern:
- generate under
/tmp/... - validate structure and existence
- copy to final destination only after success
This is especially useful for generated binaries like .pptx.
6. Validate before declaring success
Always ask Codex to verify outputs.
Examples:
- file exists
- zip/XML structure parses for
.pptx - image dimensions or PDF page count
- report file with output path, validation result, and model if visible
Practical patterns
Pattern: generated deliverables
For PPT/report/document generation, require all of the following in the prompt:
- exact output path
- exact report path
- validation steps
- final two-line summary with model/path if possible
See references/ppt-playbook.md for a concrete template.
Pattern: local dependency installs
If non-stdlib packages are needed, prefer project-local installs:
python3 -m venv .venv
. .venv/bin/activate
pip install \x3Cpackage>
Avoid assuming global install rights. Use system-level installs only when explicitly intended and actually permitted by the host.
Pattern: troubleshooting failed writes
If touch or shell writes work but ACP file edits fail, treat it as an ACP handler or sandbox-path issue, not proof that Codex itself lacks capability. Switch the generation strategy to shell/Python writes.
Decision rules
- If the task is small and read-heavy:
execis acceptable. - If the task must create deliverables: use a persistent session.
- If the prompt is long: use
-f prompt.txt. - If file editing fails once: switch to shell/Python write strategy.
- If dependencies are missing: try local
.venvinstall before changing system state. - If a binary artifact is required: generate in
/tmp, validate, then move.
Anti-patterns
Avoid these common failure modes:
- stuffing long multilingual prompts directly into one shell string
- assuming
full-accessequals sudo/root - relying only on ACP fs writes for large generated files
- declaring success before validating output structure
- writing the final artifact directly into a path that may be sandbox-restricted
References
- Read
references/ppt-playbook.mdwhen the task is to generate a PPT or similar structured binary deliverable via acpx/Codex. - Read
references/troubleshooting.mdwhen acpx sessions start but file creation, mode behavior, or sandbox boundaries are unclear.
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install acpx-codex-playbook - 安装完成后,直接呼叫该 Skill 的名称或使用
/acpx-codex-playbook触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
ACPX Codex Playbook 是什么?
Practical playbook for running Codex through acpx in persistent sessions, especially when the task needs reliable file creation, local dependency installs, s... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 339 次。
如何安装 ACPX Codex Playbook?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install acpx-codex-playbook」即可一键安装,无需额外配置。
ACPX Codex Playbook 是免费的吗?
是的,ACPX Codex Playbook 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
ACPX Codex Playbook 支持哪些平台?
ACPX Codex Playbook 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 ACPX Codex Playbook?
由 Li Xin(@spyfree)开发并维护,当前版本 v1.0.0。