Loops: causal loop diagrams

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

Annotations only cover openWorld/idempotent/destructive; the description adds the crucial partial-failure semantics ('applied in order, valid ones kept, rejected ones returned with reasons so you can fix them'), which an agent cannot infer from structured fields. It also discloses the full return payload (diagram, computed loops, what changed, open-and-edit link) and a non-obvious constraint ('you never position anything').

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?

Front-loaded with purpose, then behavior, then the command reference - a sensible ordering for a batch-command tool. Density is high but nearly every clause carries load; the inline command enumeration is long and would read better as a list, but nothing is clearly padding.

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 tool with nine union command variants, no schema descriptions and no output schema, the description supplies command grammar, identity rules, failure handling, design constraints and the return shape. An agent could construct a valid, well-formed call from this text alone.

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 0%, so the description carries the burden, and it largely does: it enumerates the per-op payloads for addVariable and addLink, explains that loops are computed rather than supplied, and disambiguates the tempId convention (t1, t2) from existing ids (v3, l4). It leaves updateVariable, removeVariable, updateLink, removeLink, setTitle and setContext field shapes and the 200-command cap undocumented, though those op names are self-evident.

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 and resource ('Create a new causal loop diagram') and immediately scopes the inputs (title, one-sentence context, command batch) and destination ('store it on a shared link'). 'New' plus the shared-link storage distinguishes it from the edit_diagram and get_diagram siblings without needing the 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?

Gives concrete authoring guidance (1-3 loops at the heart of the story, at most about 12 variables, neutral measurable names) and a mandatory workflow step ('always give that link to the person'). It stops short of explicitly naming edit_diagram as the tool for subsequent changes, so the create-vs-edit boundary is implied rather than stated.

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

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

Goes well beyond the annotations: the batch applies to the latest stored version including edits made on the site, the response shows the current diagram and what changed outside the chat, commands are applied in order with valid ones kept and rejected ones returned with reasons, and removing variables/links is called out as destructive (consistent with destructiveHint=true). It even discloses a capability boundary — "You never position anything."

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?

Front-loaded with purpose, then identifier format, then concurrency/response behavior, then the command catalog — nothing is wasted and every clause carries information. The command catalog is nevertheless a dense run-on paragraph that would scan better as discrete entries.

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 nested batch-mutation tool with no output schema, the description covers the input contract, identifier schemes, partial-failure semantics, concurrency against on-site edits, destructive operations, and the return content. An agent has enough to construct and interpret a call without further inference.

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?

With 0% schema coverage on a deeply nested oneOf, the description compensates substantially by enumerating every op name with its field sets (addVariable {tempId, name, description}, addLink {from, to, polarity, delay?, explanation}, updateLoop {loop…, name, description}, etc.) and by explaining the tempId convention (t1, t2) versus existing ids (v3, l4). It leaves some semantics unstated, notably what `delay` means, the temporal polarity wording, and the 200-command cap.

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 ("Change an existing diagram") and scopes it to existing diagrams, which distinguishes it from the siblings create_diagram and get_diagram without needing their schemas. The sentence about what a batch does reinforces that this is a mutation of an existing artifact.

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?

Explains that `diagram` accepts an id or a site link, that commands reuse existing variables, and that new loops must be named — practical guidance for constructing a call. It never states when not to use it (e.g. for a brand-new diagram, call create_diagram), so it stops short of a full routing statement.

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

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

Annotations already declare read-only, idempotent, and closed-world behavior. The description adds meaningful context beyond them: it returns the diagram, its loops, changes since the last call, an edit link, and a prompt-injection warning that diagram text is data, not instructions.

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 description is compact and front-loaded: purpose first, then when to use it, then what it returns, then the safety note. Every sentence adds useful information without repetition or filler.

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 simple one-parameter read tool with no output schema, the description covers input formats, usage contexts, returned content, change tracking, and the key prompt-injection safety concern. Nothing essential for calling or interpreting the tool is missing.

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 for the single `diagram` parameter is 0%, so the description must compensate. It does so by explaining that the parameter accepts either an id or a pasted site link in the form `.../d/{id}`, though it does not mention the 200-character limit or finer id syntax.

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?

The description states a specific verb and resource: read the latest version of a diagram by id or pasted site link. It is clearly distinct from create_diagram and edit_diagram in practical use, though it does not explicitly name or contrast those siblings.

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?

It gives explicit usage contexts: bringing a diagram into a conversation, or when the person says they changed it on the site. It does not state when not to use it or explicitly route to create_diagram or edit_diagram for other scenarios.

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