Publora Social Publishing Skills

Draft and schedule social posts from AI coding assistants through one Publora MCP integration.

Source repo
publora/skills
Stars
★ 58
Last updated
4d ago
License
MIT
Primary language
Python

At a glance

How it runs
Agent plugin / skillMCP server
Works with
Portable with changesCodex · Claude Code · Claude.ai
Cost
Free tier plus a paid hosted plan
Setup effort
Low · running in minutes
You'll need
Node.jsPublora MCP serverPublora accountShell / CLINetwork accessMCP Server
Typical use
A marketer who wants Claude Code to draft a LinkedIn post and schedule its publication.
Not a fit if
  • Teams that need direct social platform API access without Publora
  • Users who need to publish to personal Facebook profiles

What does this agent do, and when should you use it?

This repository, maintained by the Publora team, contains nine SKILL.md files for Claude Code, Codex, Cursor, OpenClaw, and other skills-aware agents. The skills call publishing and media tools through the Publora MCP server, covering LinkedIn, X, Instagram, Threads, TikTok, YouTube, Facebook, Bluesky, Mastodon, and Telegram. Users can create drafts or schedule posts for a future UTC time; MCP tools also support operations such as updating and deleting posts. The skills are instruction files, while Publora handles platform connections, account authorization, and publishing. The free Starter plan supports every listed platform except X, which requires a paid plan.

After installation, an agent selects a platform skill such as linkedin-post or instagram-post based on the requested content. It uses MCP list_connections to find connected accounts and copy their platformId values, then calls create_post to create a draft or schedule a post; public HTTPS media URLs can be submitted with the post. The agent can also use update_post, get_post, delete_post, list_posts, and media tools; the LinkedIn skill includes engagement operations. Engagement statistics for Mastodon and Bluesky are available only on plans with analytics. The skill files describe platform rules and limits, while the Publora service carries out the operations.

  1. A marketer who wants Claude Code to draft a LinkedIn post and schedule its publication.
  2. A content team using Codex to manage posts for several connected social accounts.
  3. A creator who wants to submit posts and media to multiple platforms through one MCP integration.
  4. An operator scheduling Instagram images or Reels who has an Instagram Business account.
  5. A social editor who wants to draft posts or threads with the X or Threads skill.

How do you install or deploy this agent?

Install the repository with the general skills CLI. The README also documents the Claude Code plugin marketplace. Other clients can connect to Publora through MCP. Clients that require an API key can use one from the Publora dashboard; Claude connections that support browser OAuth do not need a key.

How do you use this agent?

For a skills CLI client, install the repository, then connect the client to the MCP service and authorize a Publora account. Before creating a post, confirm the platform is connected and use the platformId returned by list_connections. Omitting scheduledTime creates a draft; for scheduled publication, provide a future ISO 8601 UTC time. You can also pass up to 10 public HTTPS mediaUrls to create_post.

What are this agent's strengths and limitations?

Pros
  • Nine platform skills cover ten social networks through a shared Publora MCP integration.
  • MCP tools are documented for posts, media, account connections, and selected engagement operations.
  • Supports both drafts and scheduled publishing through scheduledTime, and scheduled or draft posts can be updated.
Limitations
  • Publishing depends on the Publora MCP service, a Publora account, and connected social accounts.
  • X is excluded from the free Starter plan, and X API call costs are passed through to the user.
  • Platforms retain specific constraints: Instagram requires a Business account and JPEG media, while public TikTok posting can be limited for unaudited apps.

How does this agent compare with similar options?

The README distinguishes this repository from content-craft skill bundles maintained for individual platforms: this repository focuses on publishing and scheduling through Publora MCP, while the related bundles focus on platform-specific content creation.

Key facts side by side with the most closely related agents.

Agent Source review Form / cost Stars Updated Language Full support on
Publora Social Publishing Skills This agent 67 · Some gaps Agent plugin / skillFreemium ★ 58 4d ago Python Codex · Claude Code · Claude.ai
Wonda CLI 56 · Major gaps CLIPaid ★ 158 6d ago TypeScript Codex · Claude Code
PaperDebugger 44 · Major gaps Browser extensionFree + model costs ★ 1.5k 3mo ago TypeScript —
Hermes Agent: The Orange Book Guide 31 · Major gaps FrameworkFree ★ 5k 1mo ago Python —

How does FollowAgents rate this agent?

FollowAgents source review · FARS-2.1
Some gaps
67/ 100 5-point scale 3.4 / 5
Trust 13/29
Reliability 8/14
Adaptability 15/18
Convention 15/18
Effectiveness 10/13
Verifiability 6/8
Why each dimension lost points
Trust13 / 29 · 2.2/5

The evidence shows skills use user-configured MCP credentials or an API key to call Publora, with create, update, delete, publish, and scheduling capabilities. The security policy says credentials are not shipped in the repository and describes the CI key's purpose and scope. Deductions: finer-grained permission limits are not shown; the repository explicitly does not guarantee user approval before publishing; external publication can have effects, and no recovery path for already published content is described.

Reliability8 / 14 · 2.9/5

The README, security policy, and CI files are broadly consistent about drafts, scheduling, credentials, and platform limits. The FAQ covers cases such as TikTok private publishing and Instagram media formats, and CI checks for tool and platform-limit drift. Deduction: operation depends on the Publora MCP/REST service, platform APIs, account permissions, and plan availability; general API failure diagnosis and recovery are not adequately described.

Adaptability15 / 18 · 4.2/5

The repository identifies target clients including Claude, Codex, Cursor, and OpenClaw, gives platform scenarios, and defines triggers with a “Use it when / It never” table. Deduction: the boundaries are specific but do not detail every platform combination or behavior in other agent environments; trigger behavior depends on the agent following the skill instructions.

Convention15 / 18 · 4.2/5

The README is organized around installation, the skill directory, operation, REST, FAQ, and contribution guidance. It gives client-specific setup steps, documents limitations, states the MIT license, and points to a changelog and releases. Deductions: migration details and a clearly identified long-term maintainer are absent; evidence for naming stability is limited, and the examples and FAQ are useful but not a complete operating guide.

Effectiveness10 / 13 · 3.8/5

One skill set covers ten platforms and describes creating drafts, scheduling, and handling media through MCP, reducing the need for separate platform integrations. Deductions: practical use depends on platform APIs, account types, audit status, and plan limits; the supplied material does not quantify cost or show a complete end-to-end output example.

Verifiability6 / 8 · 3.8/5

The README names tools, endpoints, and platform limitations; the security policy and workflows add credential and drift-check details, and product claims are generally distinguished from stated limitations. Deduction: the supplied material omits the nine SKILL.md bodies, API documentation contents, and check scripts, so many specific capability claims cannot be cross-checked from these sources.

Risks and how to mitigate them
  • Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
  • The skills can schedule or publish content without a mandatory separate confirmation step; users should check agent and client confirmation settings and verify the target account, platform, and publication time.
  • Media URLs are downloaded server-side by Publora; avoid supplying URLs that are untrusted or should not be accessed by that service.
Evidence confidence: Low Reviewed Oct 16, 2026 Reviewed revision d55f357aa0de
See the full review method →

Recent releases

Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.

  1. v1.1.0 v1.1.0 — Bundled MCP connection and plugin fixes
    Release notes excerpt
    Agents: the plugin bundles Publora's OAuth MCP connection and listing icon; setup covers migration from the old URL, LinkedIn and Instagram metadata is valid YAML, and descriptions reflect actual confirmation behavior. All plugin versions are 1.1.0, with marketplace upgrades verified. API: the skills document existing Publora capabilities, including Threads chains, account context and media attachment; this release updates the skills and plugin package. See the [Publora API changelog](https://docs.publora.com/changelog). …
    View on GitHub ↗
  2. v1.0.3
    Release notes excerpt
    The drift check could not see the bugs it was meant to prevent. Every one of the eight fixes in v1.0.2 passed it clean, because it asked a single question, *does this tool exist*, and answered that correctly while a tool documented with invented parameters, a capability the skill denied, and a limit wrong by three orders of magnitude went unflagged. Parameters are verified against the schema …
    View on GitHub ↗
  3. v1.0.2
    Release notes excerpt
    Eight claims in the skills contradicted the API. Each was verified against the live tools/list schema or publora-api-docs, not against prose. Two came from #6; six predate it. Parameters that do not exist linkedin_create_reshare required parent, while the skill said postedId: ``` required: ['platformId', 'parent'] props: ['platformId', 'parent', 'commentary', 'visibility'] ``` …
    View on GitHub ↗

FAQ

Can I use every platform for free?
The free Starter plan covers every listed platform except X. X requires a paid plan because the X API charges per call.
Is a preview or approval step required before scheduling?
The skills do not enforce a separate preview or approval step. Whether a confirmation prompt appears depends on the agent, client settings, and instructions.
Does installing a skill connect my social accounts automatically?
No. The client must connect to the Publora MCP service and authorize an account, and the target social accounts must be connected in Publora.
Why might a TikTok video publish privately?
TikTok forces SELF_ONLY for apps that have not passed its audit, regardless of the requested viewerSetting.
Can I publish to a personal Facebook profile?
No. The README says the API forbids posting to personal Facebook profiles; the social-post skill targets YouTube, Facebook Pages, and Mastodon.
View on GitHub ↗ Install ↓

Compare agents like this one

The same FARS review applied across the shortlist this agent qualifies for.

Related agents