# Schonarth Pivot Setting Up Serena

> Setting Up Serena

- Skill: `tomevault-io/schonarth-pivot-setting-up-serena` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add tomevault-io/schonarth-pivot-setting-up-serena`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/schonarth-pivot-setting-up-serena/raw
- Safety review: pending (external: skill-scanner PASS, skillspector CAUTION)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: tomevault-io (https://skillmd.com/u/tomevault-io)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/tomevault-io/schonarth-pivot-setting-up-serena

---


# Setting Up Serena

Serena is an MCP server that gives coding agents semantic,
symbol-level code intelligence. It replaces grep-based
search with precise go-to-definition, find-references, and
atomic rename across the codebase.

## When to Use

- User asks to install or configure Serena
- Serena MCP tools are unavailable in a session
- A new project needs Serena onboarding
- The TypeScript LSP is missing (Serena symbolic tools
  silently fail on `.ts`/`.tsx` without it)

## Instructions

### Step 1 — Install Serena

Requires `uv`. If `uv` is not installed, tell the user to
install it first: https://docs.astral.sh/uv/getting-started/installation/

```bash
uv tool install -p 3.13 serena-agent@latest --prerelease=allow
```

Then register Serena as a user-scoped MCP server in the
clients you use:

Claude Code:

```bash
claude mcp add --scope user serena -- \
  serena start-mcp-server --context claude-code \
  --project-from-cwd
```

Codex:

```bash
codex mcp add serena -- \
  serena start-mcp-server --context=codex \
  --project-from-cwd
```

Verify registration:

```bash
claude mcp list
codex mcp list
```

`serena` must appear in the output for each client you plan
to use.

### Step 2 — Install Claude Code Hooks

Add the following to `.claude/settings.json` (create the
file if it does not exist). Merge with any existing content
— do not overwrite other hooks.

See `references/hooks-config.md` for the full hooks block.

The hooks do four things:

| Hook | Command | Purpose |
|------|---------|---------|
| `PreToolUse` (all) | `serena-hooks remind` | Reminds Claude to prefer Serena tools |
| `PreToolUse` (`mcp__serena__*`) | `serena-hooks auto-approve` | Auto-approves Serena tool calls |
| `SessionStart` | `serena-hooks activate` | Activates the project at session start |
| `Stop` | `serena-hooks cleanup` | Cleans up when the session ends |

Codex does not use Claude's hooks file. Register Serena in
Codex separately via `codex mcp add ...` above.

### Step 3 — Install TypeScript LSP

Serena's symbolic tools for `.ts`/`.tsx` files require the
TypeScript language server binary and the Claude Code plugin.

**Install the binary:**

```bash
npm install -g typescript-language-server typescript
which typescript-language-server   # must return a path
```

**Install the plugin** for Claude Code:

```
/plugin install typescript-lsp@claude-plugins-official
/reload-plugins
```

If the marketplace is not available, add it first:

```
/plugin marketplace add anthropics/claude-plugins-official
/plugin install typescript-lsp@claude-plugins-official
/reload-plugins
```

Verify: run `/plugin` → **Installed** tab → `typescript-lsp`
must be listed and active.

Codex uses Serena's `codex` MCP context and does not need
the Claude plugin install step.

### Step 4 — Configure the Project Language

Serena needs to know the project language to activate the
LSP. Check `.serena/project.yml`:

```bash
cat .serena/project.yml 2>/dev/null || echo "NOT FOUND"
```

If the file does not exist or `typescript` is not listed
under `languages:`, add it:

```yaml
languages:
  - typescript
```

### Step 5 — Run Project Onboarding

Start a **fresh conversation** after completing the steps
above (the MCP server must be active). Then call:

```
mcp__serena__onboarding
```

Serena will read the project structure and write memory
files under `.serena/memories/`. After onboarding
completes, start another fresh conversation — onboarding
fills context and the next session benefits from the
generated memories.

Review the generated memories and adjust any that are
inaccurate.

## Common Edge Cases

- **`serena start-mcp-server` not found**: The `uv tool
  install` step was skipped or failed. Re-run it.
- **MCP tools still missing after install**: Restart the
  client (Claude Code or Codex) to reload the MCP server
  list.
- **TypeScript LSP errors tab shows "Executable not found"**:
  `typescript-language-server` is not on `$PATH`. Re-run
  `npm install -g typescript-language-server typescript`.
- **Hooks not firing**: The `serena-hooks` binary must be on
  `$PATH` — it is installed alongside Serena via `uv tool
  install`. Verify with `which serena-hooks`.
- **Onboarding fills context**: Always start a fresh
  conversation before and after running onboarding.

## Related Skills

- `.claude/skills/managing-design-system/` — requires Serena
  and the TypeScript LSP to function
- `.claude/skills/building-skills/` — for creating new skills

---
> Source: [schonarth/pivot](https://github.com/schonarth/pivot) — distributed by [TomeVault](https://tomevault.io).
<!-- tomevault:4.0:skill_md:2026-05-23 -->