Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By surebeli
SpecTeam: AI-native spec review and decision alignment for product and engineering teams.
npx claudepluginhub surebeli/specteam --plugin spec-teamResolve divergence between collaborators via a consensus workflow. Reads open/proposed divergences from DIVERGENCES.md. A proposer submits a resolution, other parties approve or reject, and the proposer finalizes after consensus. THESIS.md is only updated at finalization time.
Archive a design proposal by moving it from .spec/design/ to .spec/archive/{timestamp}/. Checks DIVERGENCES.md for references — warns if the file is part of an open/proposed divergence. Updates THESIS.md decision log and commits with diff.
View structured git diff for .spec/ documents grouped by collaborator. Supports --last, --commit=abc123, --against=origin/main for flexible diff ranges. Highlights DIVERGENCES.md state transitions (open→proposed→resolved). Shows workflow impact analysis.
Import design documents from an external URL or MCP resource into the local .spec/design/ directory. Relies on whatever fetching ability the host AI tool already has (e.g. Claude Code web fetch, an MCP resource the user has configured) — SpecTeam does not ship connectors. Use this to sync external context into the Git-native workflow.
Initialize SpecTeam: set up collaborator identity, normalize source design docs into .spec/, create core files, and establish Git diff baseline. Use this when starting the SpecTeam workflow for the first time or onboarding a new collaborator.
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Design doc driven task management for human-agent collaboration
Documentation agents — technical writer, documentation architect
Multi-LLM documentation review for catching inconsistencies, codebase mismatches, and gaps. Supports Opus, GPT, and Gemini in parallel with synthesis and guided resolution.
Dynamically assemble expert agent teams for complex tasks using Claude Code's agent teams feature
Claude Copilot — a framework of specialized agents, skills, commands, and hooks that give Claude persistent memory, expert methodology, and task tracking. Agents cover architecture (ta), engineering (me), QA (qa), design (sd/design), DevOps (do), documentation (doc), and knowledge onboarding (kc).
Multi-agent team orchestration for Claude Code. Set up parallel AI agent teams with file-based planning, progress tracking, and role-based collaboration.
A self-evolving knowledge system for AI-paired builders. Built on Karpathy's LLM-Wiki principle, the CORE is a self-closing ingest/synthesis loop + auto-dreaming that resurfaces frozen pages when their relevance returns. Phase 1 reach (current): AI-paired engineering — compile project business semantics so agents read project conventions before they write code. Phase 2 (designed): team spec authoring + dispute resolution. Builders inherit a kata, adapt it to their project, transcend the form. 13 skills (init, import, ingest, search, graph, tier, digest, query, lint, config, dream, watch, sync). Multi-CLI session ingest in flight (v1.11).
Karpathy-style persistent markdown knowledge base — custom frontmatter dimensions, three-tier memory aging (active/archived/frozen), external plugin fallback (deepwiki-cli, web search). 10 skills: init, import, ingest, search, graph, tier, digest, query, lint.
Thin plugin layer over the llm-hopper file-based protocol. Dispatches task-typed work to vendor CLI subprocesses (codex, kimi, opencode, copilot, agy). No harness reaction core; vendor CLIs bring their own runtime. State lives in plain markdown under .hopper/. See https://github.com/surebeli/hopper-plugin.
Stage: prompt-first workflow (stable) · platform foundation (W1 in progress) — see Roadmap.
SpecTeam keeps specs, decisions, and AI agents aligned.
中文文档: README.zh-CN.md
SpecTeam is a Git-native workflow for AI-native spec review and decision alignment.
SpecTeam is the current market-facing wedge: AI-native spec review and decision alignment for product and engineering teams using multiple AI tools. It helps teams detect divergence across PRDs, architecture docs, and AI-generated proposals, make decisions quickly, and sync the outcome back into a shared source of truth.
SpecTeam now uses one consistent name across the repository, prompt skills, commands, and product story.
.claude/commands/ (recommended)git clone https://github.com/surebeli/SpecTeam.git /tmp/spec-team
# Install to current project
mkdir -p .claude/commands
for skill in /tmp/spec-team/plugin/skills/*/SKILL.md; do
cp "$skill" ".claude/commands/$(basename $(dirname $skill)).md"
done
# Or install globally (applies to all projects)
mkdir -p ~/.claude/commands
for skill in /tmp/spec-team/plugin/skills/*/SKILL.md; do
cp "$skill" ~/.claude/commands/$(basename $(dirname $skill)).md
done
/plugin marketplace/plugin marketplace add surebeli/SpecTeam
/plugin install spec-team@SpecTeam
git clone https://github.com/surebeli/SpecTeam.git ~/.codex/skills/spec-team
Copy SPECTEAM.md to your project root, then tell your AI tool:
You are now the SpecTeam Workflow. Follow all rules in ./SPECTEAM.md strictly.
Skill: init
We provide a simulated scenario to let you experience SpecTeam's conflict detection and resolution in 1 minute.
# 1. Clone and install skills
git clone https://github.com/surebeli/SpecTeam.git
cd SpecTeam
# 2. Init and point to mock data
# When asked for document directories, enter: ./tests/mock-scenarios/demo-1-conflict/alice, ./tests/mock-scenarios/demo-1-conflict/bob
/spec-init
# 3. Detect conflicts between alice (REST) and bob (GraphQL)
/spec-review
# 4. Resolve a conflict (e.g., D-001)
/spec-align D-001