2.5k
GitHub Stars
3
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill agent-content-pipeline- _meta.json302 B
- skill.json747 B
- SKILL.md4.5 KB
Overview
This skill provides a safe content workflow with human-in-the-loop approval for drafting, reviewing, revising, approving, and posting content. It includes a CLI to initialize a project, list and move items through the pipeline, manage review threads, and post approved items to LinkedIn or X. The agent can create and revise drafts but cannot approve or post — humans retain final control.
How this skill works
The skill sets up a folder structure (drafts, reviewed, revised, approved, posted, templates, and a thread log) and enforces a Draft → Review → Revise → Approve → Post lifecycle. Agents write one post per file with YAML frontmatter; humans run review commands that either attach feedback (moving files to reviewed/) or approve (moving files to approved/). Posting is executed manually via the CLI after human confirmation.
When to use it
- Setting up a content pipeline for a team or solo creator
- Automating initial draft generation while keeping human approval
- Managing review threads and iterative revisions
- Publishing approved posts to LinkedIn or X with controlled auth
- Archiving posted content and keeping an auditable thread log
Best practices
- One post per file; name files YYYY-MM-DD-<platform>-<slug>.md and include frontmatter
- Keep drafts idiomatic to the target platform (tone, length, hashtags) before review
- When you receive feedback, apply edits, move to revised/, and add a brief change note to the thread
- Use dry-run posting to preview network output before confirming live publish
- Do not attempt to set status: approved or move files into approved/ — leave approval to a human
Example use cases
- Generate a week of LinkedIn posts as drafts, then run review cycles with an editor before posting
- Draft X posts that respect the 280-character limit and let a human approve threading and auth tokens
- Create Reddit experimental drafts with subreddit frontmatter and have moderators review compliance
- Use the CLI to list pending items and move revised files through the pipeline during a content sync
- Archive posted content automatically to posted/ for future reference and audits
FAQ
No. Agents can draft and revise but cannot approve or post. Posting requires a human-run content post command and confirmation.
How do I handle authentication for X if Firefox auth fails?
You can extract auth_token and ct0 from Firefox cookies manually and supply them to the CLI or the encrypted auth store, following the documented steps.