Record a writing mistake to avoid repeating it
AI agents use mark_mistake to create or update resources in Claude Writer's Aid MCP — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Claude Writer's Aid MCP environment.
This is categorized as Write rather than Read because it creates or modifies state (recording/storing mistake records), but it is low severity since the operation is fully reversible, affects only metadata/annotations rather than the manuscript itself, and poses minimal risk even if misused by an AI agent.
From the tool's definition The tool 'mark_mistake' records writing mistakes, which is a reversible data creation/modification action that persists metadata about identified errors without altering the core manuscript content.
Documented attack patterns abuse exactly the kind of access mark_mistake gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Claude Writer's Aid MCP, and nothing reaches the server without passing your rules. This is the rule we recommend for mark_mistake:
{
"version": "1",
"default": "deny",
"tools": {
"mark_mistake": {
"limits": [
{
"counter": "mark_mistake_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}mark_mistake stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Record a writing mistake to avoid repeating it. It is categorised as a Write tool in the Claude Writer's Aid MCP MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Claude Writer's Aid MCP server in PolicyLayer and add a rule for mark_mistake: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Claude Writer's Aid MCP. Nothing to install.
mark_mistake is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the mark_mistake rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for mark_mistake. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
mark_mistake is provided by the Claude Writer's Aid MCP server (xiaolai/claude-writers-aid-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Claude Writer's Aid MCP, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
34 Claude Writer's Aid MCP tools catalogued and risk-classified — across an index of 43,000+ MCP servers.