Everything a developer needs. Nothing locked.
These features are free forever. They encode how your team builds and prove the value locally, before you need a shared organizational layer.
Encode your doctrine
Turn what your team already has (SOPs, runbooks, style guides, review standards) into versioned doctrine every AI agent follows. Point Spec Kitty at an existing runbook and it scaffolds a matching directive or tactic for your team to fill in; once synthesized, it travels locally with the repository and loads automatically, scoped to the action underway.
Full spec-to-delivery workflow
Move from intent to implementation using a structured local workflow: specify → plan → tasks → implement → review → merge. Works with Claude Code, Cursor, Codex CLI, Gemini, Copilot, and OpenCode.
Mission types
Choose a workflow shape that fits the work: software development, research, planning, or documentation. Each type produces different artifacts and a different decision trail.
Work packages & Kanban discipline
Break large AI-assisted work into smaller, trackable units with lanes, statuses, and reviewable boundaries. Large tasks become structured delivery, not a single sprawling prompt.
Git worktree isolation
Local worktree patterns keep parallel work scoped and safe to review without touching the main branch. Agents work in isolation without trampling each other's context.
Durable artifacts, not chat transcripts
Every mission leaves a persistent local record: specs, plans, task files, Architecture Decision Records (ADRs), research notes, work-package prompts, review notes, and mission history.
Works with any AI agent
Supports Claude Code, Codex CLI, Cursor, Gemini, Copilot, and OpenCode. Developers keep the AI tool they already prefer — Spec Kitty adds structure above the agent layer, not on top of it.
User-funded local inference
Inference runs through your own AI tools and accounts. Spec Kitty structures the delivery record above the agent layer. It does not require hosted inference or a Team Kitty subscription to run the CLI.
The shared layer that makes AI-assisted work governable.
Team Kitty receives the durable outputs of local-first CLI work and makes them visible, reviewable, and governable to everyone who needs to see them, without interrupting the developer. Doctrine layers the same way: org-wide standards, team packs, and project overrides merge into one enforced rulebook, so enterprise policy and team autonomy coexist.
New — Cross-repository missions
One mission. Multiple repositories. One shared narrative.
Product owner work often touches several codebases at once. Cross-repository missions give each a scoped work stream under a single mission_id with one Team Kitty view.
Shared dashboards
Browser access to projects, repositories, builds, missions, and work packages — no terminal required. Product, engineering, leadership, and reviewers get the same operating picture.
Private sync & repo admission
Each user has a private intake area. Repositories only become visible in the shared team layer when explicitly admitted — not automatically on sync. The organization controls which codebases are in scope.
Artifact & dossier visibility
Stakeholders can review the spec, plan, and rationale of any mission without digging through local files or Git history. The why, what, and how — all in one shared place.
Audit trail & mission history
A complete record of how work moved from idea to implementation — status transitions, decision events, artifact records, and activity history. Teams understand what happened, when, and who changed state.
Centralised connector layer
One governed OAuth layer for GitHub, GitLab, Jira, and Linear — instead of every developer managing their own local MCP credentials. Better security posture, less setup, lower token spend, and real-time PM/TPM delivery alignment.
Mission scorecards & health views
Scorecards and pulse views derived from stored mission history, status transitions, and review summaries. Teams see mission flow and blockers without reading every artifact. Based on synced data — not hosted AI inference.
Decision & collaboration workflows
Capture human decisions, warnings, review comments, and notes in the context of the mission — not scattered across Slack and terminals. A durable record of what was decided, by whom, and why.
Team membership & billing
The organizational container for membership, seat entitlements, and billing. The paid model maps to shared Team Kitty access, not to enabling CLI features or gating local workflow.
Not in the current product — coming later
- Hosted AI spec drafting or planning
- Collaborative real-time spec editing
- AI-generated artifact quality scoring
- SSO / SAML / SCIM enterprise identity
- Compliance exports, legal hold, data residency
- Dedicated SLA and enterprise support packages
- Shipped VS Code extension
- Self-hosting or air-gapped deployment
Benefits by role
The CLI is built for developers. Team Kitty extends the value to the whole organization, without requiring anyone to change how they work today.
🧑💻 Developers
- Keep the full local workflow — nothing locked behind a plan
- Keep using the AI agent you already prefer
- Private-by-default sync before sharing anything with the team
- Reduced status-reporting overhead: Team Kitty shows progress
📋 Product Managers
- See specs, plans, tasks, and decisions without opening a terminal
- Track whether implementation is aligned with agreed work packages
- Run cross-repository missions when work spans multiple codebases
- Gain a shared language with engineering around mission and work package status
🏗 Engineering Leads
- Turn existing runbooks and style guides into doctrine every agent follows, not tribal knowledge
- Monitor active missions and work packages across repositories
- See blockers, decision history, and review status at a glance
- Govern tool diversity — standardise the workflow without standardising the agent
🏢 Executives & Operations
- Know what AI-assisted work is happening, under whose authority
- See governance and audit evidence without blocking developer velocity
- Buy Team Kitty as a coordination product, not a tax on local CLI usage
- Answer the board question: "You funded the AI tools — what are they building?"