一键在 Manus 中运行任何 Skill

paude-spec

星标7

分支0

更新时间2026年6月23日 13:53

Write a task spec file for a paude container agent. Use when delegating work to paude, preparing a paude prompt-file, writing fire-and-forget agent tasks, or scoping submodule work for harvest.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

hhellbusch

hhellbusch/my-ai-workspace

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

Paude Spec

Produce a task spec that a container agent can execute autonomously and hand off for harvest without ambiguous commits or wrong submodule discipline.

<essential_principles>

Spec states outcomes, not step-by-step implementation (unless safety-critical).
Every spec names session-name, harvest/<slug>, and repos likely touched.
Spec ends with harvest-prep instructions — uncommitted work is lost.
Use absolute output paths when the spec will be passed to --prompt-file.

</essential_principles>

- Template: `templates/task-spec.md` - Harvest prep (for container): `templates/harvest-prep-prompt.md` - Also copied to: `devops/paude/harvest-prep-prompt.md` for human visibility - Launch: `.agents/skills/paude-launch/SKILL.md` - Planning specs: `.planning/paude-integration/task-specs/`

1. Gather intent

From conversation and $ARGUMENTS:

Task goal and boundaries
--project <name> — planning directory (default: infer or paude-integration)
--slug <slug> — kebab-case task id
--output <path> — override default path

If scope is unclear, ask one sharp question before writing.

2. Choose output path

Default:

.planning/<project>/task-specs/<slug>.md

3. Write spec

Copy structure from templates/task-spec.md. Fill all sections.

Required fields:

Session name (match planned paude create name)
Harvest branch harvest/<slug>
Repos table
Commit discipline block
"When finished" → harvest prep template reference

4. Copy harvest-prep for visibility

Ensure devops/paude/harvest-prep-prompt.md exists and matches templates/harvest-prep-prompt.md (update if drift).

5. Present to user

Show path and suggest:

paude-launch <session-name> --slug <slug> --spec <absolute-path>

<success_criteria>

Spec file exists at agreed path
Session name and harvest branch are explicit
Submodule defaults and no-push rule included
User has launch command with absolute spec path </success_criteria>

同仓库更多 Skills

同仓库

review

hhellbusch/my-ai-workspace

Review all pending changes against repo conventions before committing

2026-07-237

validate

hhellbusch/my-ai-workspace

Mark files as human-reviewed with specific validation types

2026-07-227

paude-harvest

hhellbusch/my-ai-workspace

Harvest and locally merge changes from a paude container session. Use when pulling agent work from paude, running paude harvest, merging session commits, importing submodule changes from a container, or fixing failed harvest/submodule fetch.

2026-06-237

paude-launch

hhellbusch/my-ai-workspace

Start a paude container session for fire-and-forget agent work. Use when creating a paude session, delegating a task to paude, running work in an isolated container, setting up a paude worktree, or launching pi/claude/gemini in podman.

2026-06-237

paude-triage

hhellbusch/my-ai-workspace

Diagnose a running or stalled paude container session. Use when paude seems stuck, harvest returned empty, checking paude status, debugging 403 proxy blocks, or deciding wait vs harvest vs reset.

2026-06-237

brief

hhellbusch/my-ai-workspace

Create or update a project BRIEF.md — scaffolds planning structure for work that crosses the 5-file / multi-session threshold

2026-06-237

name	paude-spec
description	Write a task spec file for a paude container agent. Use when delegating work to paude, preparing a paude prompt-file, writing fire-and-forget agent tasks, or scoping submodule work for harvest.
argument-hint	[--project <name>] [--slug <slug>] [--output <absolute-path>]
allowed-tools	Read Write Shell Glob Grep

Paude Spec

Produce a task spec that a container agent can execute autonomously and hand off for harvest without ambiguous commits or wrong submodule discipline.

<essential_principles>

Spec states outcomes, not step-by-step implementation (unless safety-critical).
Every spec names session-name, harvest/<slug>, and repos likely touched.
Spec ends with harvest-prep instructions — uncommitted work is lost.
Use absolute output paths when the spec will be passed to --prompt-file.

</essential_principles>

1. Gather intent

From conversation and $ARGUMENTS:

Task goal and boundaries
--project <name> — planning directory (default: infer or paude-integration)
--slug <slug> — kebab-case task id
--output <path> — override default path

If scope is unclear, ask one sharp question before writing.

2. Choose output path

Default:

.planning/<project>/task-specs/<slug>.md

3. Write spec

Copy structure from templates/task-spec.md. Fill all sections.

Required fields:

Session name (match planned paude create name)
Harvest branch harvest/<slug>
Repos table
Commit discipline block
"When finished" → harvest prep template reference

4. Copy harvest-prep for visibility

Ensure devops/paude/harvest-prep-prompt.md exists and matches templates/harvest-prep-prompt.md (update if drift).

5. Present to user

Show path and suggest:

paude-launch <session-name> --slug <slug> --spec <absolute-path>

<success_criteria>

Spec file exists at agreed path
Session name and harvest branch are explicit
Submodule defaults and no-push rule included
User has launch command with absolute spec path </success_criteria>