shell_read

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

Annotations already declare readOnly/idempotent/non-destructive, and the description adds genuinely new behavioral facts: cursor-based incremental read semantics, bounded output (max_bytes), optional long-polling (wait_ms), and the critical side effect that the VM remains locked by this session, requiring an explicit vm_unlock call. This lock warning is exactly the kind of context that prevents the agent from accidentally holding or leaking locks. No contradiction with 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?

Two sentences with zero filler. The core action, the cursor/bound/poll constraints, and the lock warning are each conveyed in a short clause. The lock caveat is correctly placed second, after the primary action, and every sentence earns its place.

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-param read tool with no output schema, the description covers the operation, the polling behavior, and the lock lifecycle, while annotations carry the safety profile. Remaining gaps: return format is uncharacterized (no output schema exists to fill this in) and there is no guidance on how to obtain a valid session_id (presumably from shell_start). These are notable but minor for a straightforward read operation.

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 coverage is only 20% (just vm_name is described), so the description must compensate—and it does, embedding the semantics of three undocumented parameters in prose: 'bounded' maps to max_bytes, 'after a cursor' to after_cursor, and 'long-polling' to wait_ms. session_id is only weakly tied via 'this session,' and no explicit per-parameter explanations or defaults are given, but the implicit mapping is serviceable.

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 ('Read') and resource ('output after a cursor from an interactive shell'), with the cursor qualifier distinguishing it from one-shot execution or full-output reads. The long-polling option and lock warning add operational specificity. Among the shell siblings (shell_exec, shell_write, shell_start, shell_terminate, shell_close), this is clearly the incremental-output reader.

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 description implies the usage scenario—when an interactive shell session is producing output and you want bounded, cursor-relative reads, optionally waiting via long-polling. It does not name alternatives or exclusions (e.g., when to prefer shell_exec's synchronous output over shell_read), leaving the agent to infer the decision from the interactive-shell wording. The vm_unlock pointer is a post-condition rather than a when-to-use guide.

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