[Feature]: Add --output-dir CLI flag to playwright-cli

[178inaba]

🚀 Feature Request

It would be great to have an --output-dir CLI option for playwright-cli commands, allowing users to specify where snapshot files are written without relying on the PLAYWRIGHT_MCP_OUTPUT_DIR environment variable or a config file.

The --output-dir option already exists in MCP server mode (playwright run-mcp-server --output-dir), but is not yet available for the CLI client (playwright-cli).

Example

# Current: requires environment variable
PLAYWRIGHT_MCP_OUTPUT_DIR=/tmp/pw-output playwright-cli open --headed "https://example.com"

# Proposed: CLI flag
playwright-cli open --headed --output-dir /tmp/pw-output "https://example.com"

Motivation

When using playwright-cli as a tool for AI coding agents (such as Claude Code), the CLI writes snapshot files to .playwright-cli/ in the current working directory by default. This can cause some friction:

• The current working directory is typically a git repository, so .playwright-cli/ shows up as untracked files in git status.
• Since the agent is used across many different repositories, adding .playwright-cli/ to each project's .gitignore is not practical.
• The existing workarounds — the PLAYWRIGHT_MCP_OUTPUT_DIR environment variable, a config file (.playwright/cli.config.json with outputDir), or a global gitignore — all work, but a CLI flag would be more discoverable and self-documenting.

Since the outputDir config option is already supported in the config file schema and MCP server mode, exposing it as a CLI flag for playwright-cli would follow the same pattern as tools like pytest --basetemp and tsc --outDir.

I would be happy to submit a pull request for this if the team is open to the idea. Thank you for considering!

[yury-s]

You can already set custom outputDir in the config and start with --config option pointing to it.

[178inaba]

@yury-s
Thank you for the suggestion! I'm aware of the config file approach, but my use case is running playwright-cli as a generic Claude Code skill across many different repositories. In that context:

• Creating a config file per repository is exactly the kind of extra file I'm trying to avoid (it would also show up in git status)
• Creating a temporary config file just to set outputDir adds unnecessary complexity for scripted/automated usage
• A CLI flag like --output-dir is more ergonomic for this kind of tool integration, similar to how the MCP server already supports it

[dgozman]

@178inaba Thank you for the issue. We've discussed this and decided that we don't want to introduce options like this one for playwright-cli. Existing workarounds with PLAYWRIGHT_MCP_OUTPUT_DIR environment variable and --config are sufficient. Thank you for filing!

[178inaba]

@dgozman
Thanks for the clarification and for discussing this.
I understand the decision and will use the existing workarounds.
Appreciate your time!