Skip to content
/
Phase D Batch 4

Phase D Batch 04 — OpenSpec Ecosystem (Dashboards, IDE Bridges, Packaging)

Phase D Batch 04 — OpenSpec Ecosystem (Dashboards, IDE Bridges, Packaging)

Roster (10)

slug stars distribution cli_binary local_ui orchestration multi_model tier
specboard 4 npm-package yes (specboard) web-dashboard (port 3456) none no A
spek 21 vscode-extension no web + vscode + jetbrains none no A
openspec-badge-action 4 bash-script-bundle no none (SVG output) sequential no A
vitepress-plugin-openspec 0 npm-package no none (VitePress output) sequential no A
openspec-ui-jixoai 72 npm-package yes (openspecui, port 3100) web-dashboard sequential yes (implied) A
opcode-plugin-openspec 109 npm-package no none (OpenCode UI) none no A
intellij-openspec 2 ide-extension (JetBrains) no jetbrains-plugin sequential yes (Anthropic/OpenAI/Gemini) A
clawspec 43 npm-package no none (OpenClaw chat) hierarchical no A
dotforge 6 bash-script-bundle no (Claude Code commands) none event-driven no A
plangate 2 standalone-repo yes (plangate, 30+ cmds) none hierarchical yes (Claude+Codex+Gemini) A

Intra-batch patterns

All 10 frameworks are consumers of the OpenSpec contract — none of them define a novel spec format. They divide into three functional layers: (1) visualization/viewing tools (specboard, spek, vitepress-plugin-openspec, openspec-badge-action), (2) workflow execution companions that extend OpenSpec into a specific IDE/platform (openspec-ui-jixoai, opcode-plugin-openspec, intellij-openspec, clawspec), and (3) governance tools that operate at the meta-level of AI agent behavior rather than spec content (dotforge, plangate). A striking pattern is the UI surface diversity: this batch contains a Bun HTTP server, a React+Express web app, a VitePress plugin, a CI GitHub Action badge generator, a VS Code extension, a JetBrains plugin, and an OpenClaw chat plugin — more delivery formats than any other batch. Most are single-tool (targeting either Claude Code, OpenCode, OpenClaw, or IntelliJ), not multi-tool.

The OpenSpec CLI (@fission-ai/openspec) is the shared dependency across 7 of 10 frameworks — it functions as the "kernel" of this ecosystem while these tools are its "drivers." DotForge and PlanGate are outliers: they operate on Claude Code itself rather than on OpenSpec workflows.

Most interesting finds

  1. ClawSpec — The ACP-based durable background execution with gateway-restart survivability (re-attaching to live worker sessions rather than starting new ones) is the most sophisticated durability architecture in this batch and rivals claude-flow in engineering complexity for its scope. The planning journal (accumulated requirement discussions injected into subsequent planning turns) is a unique memory mechanism.

  2. DotForge — The compiled behavior governance approach (YAML behavior declarations → bash hooks with 5-level escalation → permanent override audit trail) and the meticulous tracking of 30+ Claude Code API changes across 35 versions represents a unique contribution: it turns Claude Code's configuration surface into a governed, auditable system. The 5-month silent bug in session-report.sh (corrupting 54 JSON files) documented in the CHANGELOG is an unusually honest transparency about known issues.

Items written as Tier C

None — all 10 had coherent READMEs and sufficient public material for full 11-file reports.

Cross-references discovered

  • openspec-ui-jixoai is the official companion to the openspec seed (same Fission AI maintainer lineage); it extends the seed's 11 skills with 10 additional skills and adds a PTY terminal + OPSX compose panel.
  • spek ships the frontend-design skill from the superpowers plugin ecosystem (symlinked in .claude/skills/) — direct dependency on a seed framework's skill.
  • intellij-openspec and spek both provide IntelliJ IDEA integration but at entirely different levels: spek's IntelliJ plugin is a read-only JCEF viewer while intellij-openspec is a full workflow orchestration IDE plugin.
  • opcode-plugin-openspec (OpenCode) and clawspec (OpenClaw) serve the same "OpenSpec planning/implementation in a non-Claude-Code tool" niche but for different platforms and with drastically different complexity.
  • plangate and dotforge are both governance tools targeting Claude Code but at different levels: plangate governs the development workflow (spec → plan → implement → PR), while dotforge governs Claude Code's runtime behavior (hook enforcement + config sync).
  • spek also generates OpenSpec badges via scripts/generate-badges.ts (same functional goal as openspec-badge-action) — parallel solutions for badge generation in the ecosystem.