- Home
- Skills
- Mobile Development
- Charles Proxy QA: inspect and mock mobile app traffic
More screenshots
Works with the AI Agents you already use
Charles Proxy QA: inspect and mock mobile app traffic
Let your AI agent see what your mobile app sends and make it receive any server response you need: an empty list, a zero balance, a 500, a slow reply. No backend changes.
$9
Charles Proxy QA: inspect and mock mobile app traffic
Example session with this skill installed
My wallet app talks to its API through Charles. Show me which requests the wallet screen makes, then make the transactions list empty and make the balance request fail with HTTP 500. Prove it and clean up after.
- Read your context and instructions
- Compiled the charles proxy qa
The wallet screen makes three requests: GET /json (config), POST /anything with action=get_balance, POST /anything with action=get_transactions. Both POSTs share one URL, so I used dispatcher rules matched on the action field.
- Empty transactions: patch rule on get_transactions sets /json/items to []. The real server still answers; only that field changes. The app shows 0 items.
- Balance error: fixture rule on get_balance with status 500. The app shows HTTP 500.
Proof: after reopening the screen, both responses carry the X-Charles-MCP-Rule header with the rule id.
Cleanup: both rules removed, the app gets real responses again (balance 125.40, 2 transactions). Your Charles session was not cleared. Map Remote for the domain is still on, as it was before I started.
Connects securely to your tools. The creator never sees your data.
About this skill
Who it's for: QA engineers and mobile developers who use Charles Proxy.
The skill drives charles-mcp, an open-source MCP server for Charles Proxy. It tells your agent when to use it, which rules not to break, and the order of calls for each task.
What you can ask for
- "What did the app send when I tapped Pay?" The agent captures the traffic, groups it by endpoint and opens only the request you care about.
- "Show the wallet with a zero balance." The agent takes the real response and changes only that field. The rest of the response keeps its shape, key order and number types, so the app parses it exactly as before.
- "Return 500 on login" or "make the balance call take 5 seconds." You get error screens, loaders and timeouts without asking the backend team.
- "Did the analytics event fire?" The agent searches request bodies by event name.
- "Mocks work on some hosts but not others." The agent reads how Charles is configured (without its license or passwords) and names the mapping that is broken.
- Each mock is checked against fresh traffic (a header shows which rule answered). Each one is removed when the scenario ends.
- Saved
.chlsjsessions, protobuf decoding and request replay are supported. Replay always asks for your consent first.
How it differs from scripting Charles by hand
- One Map Remote rule per domain covers every path. Mocks are matched by request body fields such as
action, so APIs that use a single URL for many operations still work. - Real responses are patched by field rather than retyped.
- The agent never clears your Charles session, quits Charles or sends replayed requests without asking.
Requirements
- Charles Proxy with the Web Interface enabled.
- The free charles-mcp MCP server. The skill gives your agent the one-line install:
uvx --from git+https://github.com/ExQA/charles-mcp charles-mcp - Works with Claude Code, Codex, Cursor and other agents that load skills and MCP servers.
Source and docs: https://github.com/ExQA/charles-mcp (MIT)
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Mobile Development.
The skill does nothing without Charles Proxy and the charles-mcp MCP server. It does not capture traffic itself. No WebSocket mocking, and no simulation of dropped connections (only delays and status codes). Adding a Charles rule for a new domain needs manual steps in Charles or a Charles restart.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago
- Software
- Charles Proxy 4 or 5
uv (runs the charles-mcp server)Agent with MCP server supportCharles MCP
Git
The charles-mcp MCP server is free and open source (MIT). The skill tells the agent the install command. In Charles, enable Proxy Web Interface Settings once.
Permissions
Allowed Hosts
File Scopes
The skill has no scripts, only SKILL.md. Captured traffic can hold tokens and personal data; the skill tells the agent not to repeat secrets and never to write captured values into code or commits. The agent is told not to open the Charles config file, which holds the license and passwords.