agent-content-pipeline

What This Skill Does

The agent-content-pipeline skill provides a structured, human-in-the-loop workflow for managing social media content. It transforms the chaotic process of drafting, refining, and publishing content into a deterministic state machine: Draft → Reviewed → Revised → Approved → Posted. By utilizing specific directory structures and CLI commands, it ensures that an AI agent can assist with high-quality content generation while strictly deferring final approval and publishing authority to a human user.

Installation

To integrate this skill into your environment, follow these steps:

1. Ensure you have Node.js installed.
2. Run npm install -g agent-content-pipeline.
3. Navigate to your project directory and initialize the pipeline using content init . (or content init . --secure for encrypted approvals).
4. Install the skill into OpenClaw using clawhub install openclaw/skills/skills/larsderidder/agent-content-pipeline.

Use Cases

This skill is ideal for content creators, agencies, and developers managing technical blogs or social media presence. Use it when:

• You need to draft a series of technical posts for LinkedIn or X (Twitter) consistently.
• You want to ensure every post undergoes a mandatory human review cycle to prevent misinformation.
• You need to archive content automatically after it has been published to keep your workspace organized.
• You require a clear audit trail of feedback loops via the .content-pipeline/threads/ system.

Example Prompts

1. "I need to draft a new LinkedIn post about our latest API update. Please create a new file in the drafts folder following the naming convention and keep the tone professional but human-centric."
2. "I've reviewed the latest feedback in the reviewed/ folder. Please apply the requested changes, move the file to the revised/ folder, and notify me with a summary of the edits made."
3. "List all current content pending in the pipeline and tell me how many items are currently waiting for human approval in the approved folder."

Tips & Limitations

• Safety First: The agent lacks the permission to move files to approved/ or trigger the content post command, ensuring you never accidentally publish unvetted material.
• File Management: Strictly adhere to the YYYY-MM-DD-<platform>-<slug>.md naming format to prevent CLI errors.
• Platform Specifics: Keep X content under 280 characters to avoid truncation and remember that LinkedIn performs best with 3-5 hashtags at the end of the post.