v1.0-beta.4 · now with mcp bridge

The workspace for shipping
with AI agents

Work items, memories, docs and rules — accessible to every agent you use. Native CLI, TUI and MCP bridge.

works with claude code, cursor, copilot, windsurf, codex, kiro, gemini, aider
~/acme · fenixmcp · connected
fenix mcp start --agent=claude-code
→ spinning up mcp bridge on :8421
✓ claude-code connected · 56 actions registered
claude "what's blocking sprint 14?"
assistant → reading sprint:S-14, memory:decisions
2 blockers · PROJ-248 auth redirect (rafael), PROJ-251 cli autocomplete (marina)
both tied to decision mem:a7f2 "keep SSO on same domain"
→ answered from real project data · no hallucination
works with
any agent
Claude Code
Cursor
Copilot
Windsurf
Codex
Kiro
Gemini CLI
Aider

Your AI doesn't know anything about your project.

Every morning you open Claude Code and explain it all again. Which project, which patterns, which tasks. Your decisions die at the end of the chat. Your docs live on Notion, disconnected from the code.

Context that dies
Every session with your AI starts from zero. You re-explain the project, the patterns, the tasks. Decisions vanish.
Tribal knowledge
Standards, architecture, past trade-offs — all stuck in the tech lead's head. Nothing for the next dev or the next agent.
Tool sprawl
Jira for tasks, Notion for docs, Slack for decisions, Confluence for standards. Five tabs, one brain, zero context.

Everything the lifecycle needs. In one workspace.

Not another project manager. A knowledge-first workspace where every artifact — tasks, memories, docs, rules — is readable by your team and by your agents.

webdesktopcli · tui
6 hierarchies
Work Items & Boards
Epic → Feature → Story → Task → Subtask → Bug. Kanban and table views, drag-and-drop, inline editing, story points, filters.
01
semantic
Memory Graph
Semantic search with embeddings. 80%+ similarity auto-dedupes. Notes linked to work items, docs and sprints.
02
markdown
Docs & Runbooks
Knowledge base with Markdown, Mermaid, versioning. Page, guide, api_doc, tutorial, runbook, playbook, checklist.
03
exportable
Skills Marketplace
Shareable AI instructions. Create, share and install skills. Export to Cursor, Claude, Copilot, Windsurf.
04
$
Native CLI + TUI
`fenix work list`, `fenix todo`, `fenix board --ui`. Bash/zsh/fish autocomplete. Vim motions inside the TUI.
05
openapi
API Catalog
Registry of your team's API specs. Import OpenAPI / Swagger. Lifecycle tracking: alpha, beta, stable, deprecated.
06

Your AI, actually connected
to your project.

Fenix MCP is a server that connects any AI agent to real project data — work items, memories, rules, docs, sprints, TODOs, API catalog. Ask once, read forever.

56+
mcp actions
8
supported clis
CLAUDE CODE · your editor
> user
implement the cli autocomplete from sprint 14. follow team patterns.
> claude
→ fenix.work.get("PROJ-251")
→ fenix.rules.list(category="cli")
→ fenix.memory.search("autocomplete design")
→ fenix.docs.read("cli/commands.md")
writing src/cli/autocomplete.ts
MCP BRIDGE
:8421
FENIX · your project
< work:PROJ-251
"Add CLI autocomplete for projects"
priority: med · points: 3
< rules:cli (3 match)
cli-001 · commander.js only
cli-002 · zsh + bash + fish parity
cli-014 · never block on network
< memory:a7f2 (0.94)
"autocomplete: cache refreshes every 5min in ~/.fenix/completions.json"
✓ context delivered · 4 artifacts
// actions available to any agent
work.*
├─ list
├─ create
├─ update
├─ move_status
├─ assign
├─ link
memory.*
├─ save
├─ search
├─ dedupe
├─ link
├─ favorite
docs.*
├─ read
├─ write
├─ version
├─ publish
├─ list
rules.*
├─ list
├─ export
├─ apply
├─ fork
+ sprints, boards, todos, api_catalog · 56+ total
read the mcp docs →

Terminal-first, always.

Every work item, doc and memory is reachable from the CLI or the TUI. Vim motions inside the board. Bash/zsh/fish autocomplete. Never leave the terminal.

FENIX CLI
npm i -g @fenix/cli
installed fenix@1.0.0-beta.4
fenix auth login
→ opening browser…
logged in as rafael@acme.corp
fenix work list --mine
PROJ-248 auth redirect high in-progress
PROJ-251 cli autocomplete med todo
PROJ-237 tui drag & drop med in-progress
→ 3 items · synced
FENIX BOARD — SPRINT 1412 items
TODO · 4
PROJ-251 autocomplete
PROJ-252 doc export
PROJ-258 invite flow
PROJ-260 audit log
WIP · 2
▸ PROJ-248 auth
PROJ-237 tui d&d
DONE · 6
PROJ-240 cli help
PROJ-231 logs
PROJ-229 fts
PROJ-228 pat
j/k move · h/l column · open · m moveq quit

Wherever you work, Fenix is there.

Same backend, same data, three surfaces. Your team chooses the one that fits each moment.

fenix.app/board
Web App
React + Vite
stable
Full workspace in the browser. No install.
linuxmacoswindows
L
W
Desktop App
Tauri 2.0
stable
Native for Linux, macOS, Windows.
Mobile App
iOS + Android
in dev
Review items on the go.

Same category. Different species.

Management tools were built for humans. Fenix was built for humans AND AI agents.

capability
Fenix
Jira
Linear
Notion
Native MCP bridge
Semantic memory graph
Plans persist cross-session
Rules marketplace
Multi-CLI support (8+)
Native TUI
API catalog (semantic)
Kanban + sprints
Desktop native app
Markdown docs

Engineers who chose Fenix.

"
"We moved the whole workspace to Fenix. The MCP bridge alone saves my team an hour a day of context-setting with Claude."
RN
Rafael Nunes
Staff Engineer · Acme
"
"TUI with vim motions on a Kanban is something I've wanted for years. It's weird that it took this long."
MP
Marina Prado
Tech Lead · Orbit
"
"Semantic memory with auto-dedupe is the missing piece. Finally a workspace that actually understands context."
JT
João Teixeira
Principal Eng · Northwind

Simple pricing. Priced per dev.

Free while your team is small. Pay for what you use after that. No seat games.

Free
For solo devs and tiny teams.
$0
/ forever
Up to 5 team members
1 workspace
Web + desktop app
MCP bridge (core actions)
Community rules marketplace
Memory with embeddings
MOST POPULAR
Pro
For teams shipping daily.
$12
/ dev / month
Unlimited members
Unlimited workspaces
Full MCP (56+ actions)
Private rules & skills
Sprint analytics
Priority support
Audit log (90 days)
SSO (Google, GitHub)
Enterprise
For orgs with compliance needs.
Custom
/ annual
Everything in Pro
Self-hosted option
SAML + SCIM
Audit log (unlimited)
Custom data residency
Dedicated support
Security review
no credit card on free plan · cancel anytime on pro

Questions engineers ask.

MCP (Model Context Protocol) is an open spec that lets AI agents call structured tools. Fenix exposes 56+ actions over MCP — your agent can read, create and update work items, memories, docs and rules directly, with no prompt engineering.

The operating system
for dev with AI.

Free for teams up to 5. No credit card. Two-minute setup.

$npx fenix@latest initcopy