Fusion (write)

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

Annotations already declare destructive/non-idempotent, but the description adds genuinely new behavior: every change is read back and receipt.verified=false signals the change did not take, disconnect is one undo step with undo restoring the link, load_tool_settings replaces existing parameters, set_inputs runs in one undo step because the comp is not locked, and rename_tool preserves expressions referencing the old name. These are non-obvious side effects and reversibility facts an agent could not get from the structured fields.

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?

The highest-value content — the overwrite warning and the read-back verification rule — is front-loaded in the first sentence, followed by a scannable one-line-per-action catalog and a separate shortcuts section. The length is justified because the schema does not enumerate action parameters, so every line carries information.

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?

An output schema exists and the description still flags the key return signal (receipt.verified=false), so return-value explanation is not required. Between the action catalog, the risk warning, the undo semantics, and the shortcut/batch routing, an agent has everything needed to call this destructive multi-action tool correctly.

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?

The schema's op object is only {action, additionalProperties:true}, so the description carries nearly all parameter knowledge and does so per action, marking required params with '*' and adding meaning where types alone would not suffice (empty expression string clears, keys are (time, value) pairs, sets batches many inputs). The lone weakness is create_text_title's '+2 more', which hides two params, but describe_action is named as the escape hatch.

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?

Opens with a specific verb+resource+risk statement ('Change fusion state; may overwrite current values') and then enumerates all 17 supported actions with their scope, so an agent knows exactly what class of mutation this tool performs. It is unambiguously distinct from fusion_read, fusion_delete, and workflow_write.

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?

Explicitly routes whole-task work to the workflow_write.title_overlay shortcut ('One call instead of create_text_title, add_tool, connect, set_inputs, set_keyframes') and multi-step/repeated work to batch_run ('ONE batch_run'). It also points at describe_action as a fallback, so the agent has a decision path for both simple and compound cases.

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