Forgeplane
Forgeplane describes itself as making deterministic, composable tools for the agent era, intended to be trusted by the people accountable for software that increasingly writes and runs itself. Its current highlighted project, Pinrail, gives agents a structured human-review step before consequential actions.
At a Glance
- Individual developers using coding agents
- Development teams using Claude Code, Codex, Cursor, OpenCode, Gemini CLI, Kimi, or other command-running agents
- Teams that need human approval before actions under a person's name or against production systems
- Automation and CI workflow users
- +1 more
AI Tools by Forgeplane
(1)Pinrail
Human Review App for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about Forgeplane
Latest News
Pinrail 0.1.3 released with Linux notification cleanup and dark-GNOME menu-bar icon fixes.
Pinrail 0.1.2 released with plugin SDK installation changes and macOS 26/sidebar fixes.
Pinrail 0.1.1 released with review-header and narrow-window layout changes.
Pinrail creator introduced the desktop human-in-the-loop review inbox in a Show HN launch post.
Products & Services
A free, open-source desktop review inbox for agent work. An agent submits a structured review and waits; a person decides in a view suited to the content, and the decision returns to the agent as Markdown or JSON. It runs locally, with a loopback API and SQLite storage, on macOS and Linux; Windows is planned.
Plugins shipped with Pinrail for code review, lists, feedback/questions, Markdown documents, and image review. Each plugin defines the agent payload, the decision schema, and the review UI.
An official Forgeplane repository of installable sample plugins for artifact/page review, audio, calendar, canvas, email, logo, 3D model, motion, palette, trade, video, and visual-diff reviews.
SDK and development harness for creating, serving, and testing custom Pinrail plugin views; plugins can be plain HTML or built with frameworks such as React, Vue, or Svelte.
Market Position
Forgeplane positions Pinrail as a local-first, structured human-in-the-loop review layer for agentic workflows: instead of approving agent output in terminal scrollback or chat, users review it in a purpose-built view and return structured decisions. A Hacker News commenter compared it favorably with Plannotator; Pinrail differentiates through multiple content-specific plugins, revision rounds, a CLI/CI workflow, local-only storage, and an extensible plugin SDK.
Leadership
Founders
pnezis
Pinrail's creator and the author who introduced it on Hacker News. The author says they built Pinrail after repeatedly having to approve agent work in terminal scrollback; the sources identify the creator by the handle pnezis rather than a personal name or prior employer.
Founding Story
The creator says Pinrail began after an unsatisfactory experience approving twelve AI-generated pull-request comments in terminal scrollback or chat. They first had the agent write comments into a custom HTML page over the diff so individual findings could be accepted, rejected, or edited, then generalized that review pattern to emails, documents, images, and other agent outputs.
Business Model
Revenue Model
Pinrail is presented as free and open source under Apache 2.0, with local desktop downloads and no account or telemetry. No paid subscription, usage charge, or commercial licensing model is stated in the reviewed sources.
Target Markets
- Individual developers using coding agents
- Development teams using Claude Code, Codex, Cursor, OpenCode, Gemini CLI, Kimi, or other command-running agents
- Teams that need human approval before actions under a person's name or against production systems
- Automation and CI workflow users
- Plugin authors and agent-tool builders
- Reviewing and selectively posting AI-generated pull-request comments
- Approving or editing agent-drafted emails before sending
- Reviewing generated pages, designs, images, logos, animations, audio, video, and 3D models
- Answering agent questions and approving proposed lists or actions
- Human approval gates in scripts and CI jobs
- Iterative review of agent revisions with a record of prior decisions