SkillsForge

SkillsForge is described as an AI CLI plugin that acts as a "Work OS for Agent Skills", aiming to improve agent productivity by routing skills and workflows, shrinking context, and enabling multi-host installation across various AI CLIs (e.g., Claude, Cursor, Codex). It includes a local skill library with 499 skills, 28 packs, and 100 dry-run workflows, along with compact CLI operators like sf map, sf slim, and sf digest. The system supports validation, packaging, and receipts to enhance safety and trust. It is built as a Node.js 20 ESM developer tool with a local HTML UI and MCP server.

The description states that SkillsForge is not just another prompt folder but a structured system for managing agent skills and workflows. It emphasizes local-first design, no cloud dependencies, and cross-host compatibility. The project was submitted to the OpenAI 2026 hackathon by one developer (Tlkh201313).

Key open question: Is there evidence of real-world usage or adoption of SkillsForge beyond its author's claims? The description does not indicate any revenue, customers, or traction.

The description states that SkillsForge is a Work OS for Agent Skills, designed to help AI agents:

• Pick the right skill or workflow for a task
• Use a local skill library with 499 skills, 28 packs, 11 profiles, and 100 dry-run workflows
• Reduce noisy terminal output using compact CLI commands (sf map, sf slim, sf digest, etc.)
• Operate across multiple AI CLIs (Codex, Claude, Code, Cursor, OpenCode, ZCode, Hermes, Gemini)
• Stay safer with validation, packaging, and receipts
• Access a local HTML library UI and AI-readable index

It is built as a Node.js 20 ESM developer tool, including:

• A bundled CLI skill catalog
• Workflow engine
• Local library UI
• MCP server
• Multi-host installer

The system uses skillsforge.json sidecars for routing capability and trust metadata, and supports local library building that indexes repo skills, user skills, plugin-cache, and host-specific roots.

Inference: The product is a developer tool aimed at improving AI agent workflows through structured skill management and context reduction. It is not a commercial SaaS offering but a CLI-based system for developers to use in their own environments.

The description states that SkillsForge was built because "agent productivity isn’t just about throwing more prompts at the model". The author positions it as an alternative to existing tools like Claude, Code, or Superpowers-style plugins, which they claim "just dump more skills into one host."

The core positioning is:

• A cross-CLI Work OS
• A system that routes the right skill and workflow
• A tool that shrinks context with map/slim/digest
• A system that ships with validate→package→receipts

It is described as a local-first, no-cloud dependency solution that enables agents to be faster without blowing up prompts or wasting tokens.

Inference: The positioning evolved from a hackathon project into a structured tool for agent productivity, but the description does not indicate any prior version or evolution beyond this self-reported iteration.

The description states that SkillsForge is built for AI developers and agent users, particularly those working with AI CLIs like Claude, Cursor, Codex, etc. It targets:

• Developers who use AI coding agents
• Users who want to improve agent performance by reducing context noise and increasing skill routing accuracy
• Teams or individuals looking for safer, reusable, and organized skill packs

It is described as a developer tool, not a commercial product for end-users.

Inference: The ICP is likely early-stage AI developers or advanced users of AI agents who are interested in systematizing their workflows. No evidence of specific personas or customer segments beyond this general description.

The description does not provide any information about pricing, monetization, or business model.

Not evidenced

The project is built using:

• Node.js 20 ESM
• A bundled CLI skill catalog
• A workflow engine
• A local library UI
• An MCP server
• A multi-host installer

It supports:

• Local-first design
• No heavy frontend framework
• No cloud dependencies
• A single-file local HTML dashboard plus JSON index

The system includes:

• 499 skills, 28 packs, 11 profiles
• 100 dry-run workflows
• Compact CLI operators (sf map, sf slim, sf digest, etc.)
• Trust layer with validation, packaging, and receipts
• A thin MCP server for routing, lookup, settings, quality checks, and workflows

Inference: The technical stack is minimal and developer-focused. It appears to be a prototype or early-stage tool, not a commercial-grade product.

The description states that:

• It was submitted to the OpenAI 2026 hackathon
• It includes 499 skills, 28 packs, 11 profiles, and 100 dry-run workflows
• It has a local HTML library UI and an AI-readable index
• It supports multi-host install with clear fidelity boundaries
• It has 98 specialist agent roles and 135 command shims

However, there is no evidence of revenue, customers, user adoption, or traction beyond the hackathon submission.

Not evidenced

The description states that SkillsForge is not like existing tools such as Claude Code or Superpowers-style plugins, which it claims "just dump more skills into one host."

It positions itself as a system that:

• Routes skills and workflows
• Shrinks context
• Provides safety gates
• Enables multi-host compatibility

No specific competitors are named, but the author implies that current tools lack structure, trust, or cross-platform support.

Inference: The competitive landscape is not clearly defined. It appears to be a niche tool for developers working with AI agents, and there is no indication of direct competition beyond general agent tooling.

• No evidence of traction or adoption: The project was submitted to a hackathon and has no known users or customers.
• Unverified claims: All features and capabilities are self-reported without independent validation.
• Single developer team: Only one member (Tlkh201313) is listed, which may limit scalability or long-term maintenance.
• No commercial model: No indication of how the tool will be monetized or distributed beyond its current form.
• Unclear user value proposition: While described as improving agent performance, no real-world data or use cases are provided.

1. What is the actual utility of SkillsForge in real AI coding sessions? Is there any evidence of improved agent performance?
2. How does it handle trust and validation in practice? Are there any real-world examples of skills being validated or rejected?
3. What are the limitations of multi-host support, and how do they impact usability?
4. Has there been any feedback from users beyond the author’s own testing?
5. Is there a plan to scale beyond the current hackathon prototype, and what would that look like?