AI agents invoke worktree_merge to trigger actions in Lockstep. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
Merging a worktree is an irreversible or hard-to-reverse git operation that modifies the repository's branch/commit history and working files. While not purely destructive (it doesn't delete), it executes an external operation with significant side effects.
From the tool's definition Tool name 'worktree_merge' and description 'Merge an implementer' — triggers a git worktree merge operation, which is an external VCS operation that can cause conflicts, overwrite changes, or alter repository state.
Attacks that exploit this kind of access
Merge an implementer. It is categorised as a Execute tool in the Lockstep MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Lockstep MCP server in PolicyLayer and add a rule for worktree_merge: 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 Lockstep. Nothing to install.
worktree_merge is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the worktree_merge 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 worktree_merge. 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.
worktree_merge is provided by the Lockstep MCP server (tmmoore286/lockstep-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →