cursor_agent

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does so well: it discloses that in headless mode file changes are only PROPOSED unless CURSOR_ALLOW_YOLO=true, that --force then applies them, that CURSOR_SANDBOX=enabled confines them, and that the result lists changed files plus the model used. It even documents that progress tokens yield a notification per tool call, which is behavior an agent cannot infer from schema or annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

Four information-dense sentences with no filler: capability and model scope first, then mode semantics, then the critical headless-safety caveat, then result/progress behavior. Every sentence carries load and the most consequential fact (propose-only vs --force) is placed where it will be read.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 5-parameter, prompt-driven agent-execution tool with an output schema present, this is complete: execution modes, model sourcing, the propose-vs-apply safety model with its env-var controls, and progress-notification behavior are all covered, and return contents need not be detailed further given the output schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so mode, model, prompt, workspace, and timeout are all already documented in the schema, including the same mode definitions. The description's model-family enumeration and 'run cursor_models for ids' note restate schema content rather than adding syntax or constraints, so the baseline 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Execute a prompt using Cursor's AI agent') and immediately scopes it by noting that any model id the installed cursor-agent accepts works, pointing to cursor_models for ids. An agent can distinguish this from cursor_reply, cursor_sessions, and cursor_health without opening any schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

The three modes are spelled out with their operational meaning ('agent' = tools/terminal/search, 'plan' = design-focused, 'ask' = read-only), which gives clear when-to-use guidance for mode selection, and cursor_models is named as the source of model ids. It stops short of explicit routing rules against sibling tools like cursor_reply or cursor_sessions, so it is clear context without exclusions.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.