claude-switch-models-setup
An agent skill by daymade, from daymade/claude-code-skills. Tags: cli, cloud, configuration, multi-agent, setup.
What it does
Set up and maintain multiple isolated Claude Code CLI profiles so students and power users can run different LLM providers (Kimi K3, Kimi K2.7 highspeed, MiniMax M3, MiniMax M2.7, GLM, DeepSeek, StepFun, Anthropic) in separate terminal windows at the same time. Use this skill whenever the user asks about multi-provider Claude setup, multiple Claude Code windows, switching models or the csk/csd/csg aliases, CLAUDE_CONFIG_DIR, the ~/.claude-profiles directory, or running Kimi/MiniMax/GLM/DeepSeek/StepFun alongside Anthropic. It also owns profile-drift troubleshooting — a third-party profile missing skills/hooks/plugins the default profile has, claude-profiles-doctor reporting a real directory where a symlink belongs, or settings not converging — and per-provider context-window configuration (the [1m] marker or explicit CLAUDE_CODE_MAX_CONTEXT_TOKENS).
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add daymade/claude-code-skills --skill claude-switch-models-setupOr copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/daymade-claude-code/claude-switch-models-setup ~/.claude/skills/claude-switch-models-setupSafety box score
Grade F, high risk: 89 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill sets up a multi-profile Claude Code environment, including shell integration, LaunchAgent automation, provider routing files, and backup pruning. The main risks are broad local filesystem changes, stored API credentials, automatic background syncing, and a few core implementation files that were not readable in this review.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 2 (moderate) | It runs shipped shell/Python code and subprocesses (`bash`, `python3`, `git`, `launchctl`, `uv python install`), but it is mostly bounded to setup/maintenance rather than arbitrary code execution. |
| Network access | 2 (moderate) | It is designed to send prompts to remote provider endpoints via `ANTHROPIC_BASE_URL`/`ANTHROPIC_AUTH_TOKEN`, and `scripts/setup.sh` also installs Python with `uv python install 3.12`, which may fetch over the network. |
| Filesystem reach | 3 (high) | It writes and rewrites home-directory config and state (`~/.claude/settings/<provider>.json`, `~/.claude-profiles`, `~/Library/LaunchAgents`), and `scripts/prune-source-sync-backups.py` can recursively delete buckets with `shutil.rmtree(entry["bucket"])`. |
| Secrets & credentials | 3 (high) | The skill stores provider tokens in plaintext in `~/.claude/settings/<provider>.json` and explicitly discusses sharing the macOS Keychain/MCP OAuth store via `CLAUDE_SECURESTORAGE_CONFIG_DIR=""`. |
| Privilege & persistence | 3 (high) | It edits shell startup files, installs and enables a per-user LaunchAgent with `launchctl bootstrap/enable`, and provisions its own Python runtime under `~/Library`/`~/.config`. |
| Instruction hijack surface | 3 (high) | The skill explicitly recommends aliases that invoke `claude-profile ... --dangerously-skip-permissions`, which is a direct permission-check bypass. |
| Irreversible actions | 3 (high) | It registers a SessionStart hook that runs on every session and installs a LaunchAgent that continuously repairs/syncs state, so it can change profile configuration automatically without per-run confirmation. |
| Transparency | 2 (moderate) | Several core files were not readable in this review (`scripts/claude-plugins-sync.py`, `scripts/sync-profile-settings.py`, `scripts/sync-local-skill-sources.py`, `references/local-source-sync-architecture.md`, `references/troubleshooting.md`), limiting audit coverage of the main sync logic. |
Before installing
- Read `scripts/setup.sh`, `scripts/sync-local-skill-sources-daemon.sh`, and `scripts/prune-source-sync-backups.py` first; confirm the LaunchAgent, `uv python install`, and `shutil.rmtree(...)` are acceptable.
- Decide whether you want plaintext provider keys in `~/.claude/settings/<provider>.json` and whether `CLAUDE_SECURESTORAGE_CONFIG_DIR=""` should ever be allowed on this machine.
- If you plan to use the pruning tool, run it in dry-run mode first and inspect the buckets it would delete before using `--apply`.
Rated 2026-10-19 by an automated review of 30 files from the skill's repository.
Source
- Repository
- daymade/claude-code-skills (all skills from this repository)
- Path
- daymade-claude-code/claude-switch-models-setup/SKILL.md
- Branch
- main
- Updated
- 2026-09-20
Related skills
- signing-in-to-aws — Gets AWS credentials for CLI/SDK access via `aws login`.
- codex-1m-context-window-setup — Configures and verifies an expanded, model-aware context window for OpenAI Codex CLI and Codex Desktop by safely updating the shared base config.
- wizard — Generate an interactive bash wizard that walks a human through steps only they can perform.
- agents-get-started — Use when a developer wants to create a new agent project or get started with AgentCore.
- coordinating-multi-space-devops-agent — Coordinate the AWS DevOps Agent across multiple AgentSpaces from one Claude Code session — route questions to the right space (prod vs staging vs knowledge).
- managed-db-services — Configure DigitalOcean Managed MySQL, MongoDB, Valkey, Kafka, and OpenSearch for App Platform.