record_learning

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

With annotations present (readOnlyHint=false, destructiveHint=false), the bar is lower, and the description adds real behavioral context: persistence affects future sessions, a 'detect' block enables auto-recognition next session, and the tool 'never invents a URL, so this is the only way one appears' — a genuine policy disclosure. It also instructs URL verification before recording. Nothing contradicts the annotations: readOnlyHint=false is consistent with a persist operation.

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 somewhat long, but every segment earns space: the purpose is front-loaded, the updates shape needs its examples, and the NEW-family and sources-list paragraphs cover mandatory edge cases. It is densely structured with little fluff, though the many inline JSON examples make it verbose by necessity.

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?

All three required parameters are explained, edge cases (new family detection, sources-only URL introduction) are covered, and an output schema exists, so return values need no explanation. The remaining gaps are minor: it presupposes familiarity with 'the guidance shape', and it does not describe what happens on failure or how recorded settings interact with get_model_guidance. For a high-complexity tool the description is still complete enough for a first-call agent.

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 full burden and fully compensates: it assigns meaning to 'family' (any family name, with the NEW-family detect/loader variant), 'source' (URL/model page) and 'updates', with concrete shape examples such as {"sampling": {"cfg": {"default": 3.5}}}. It also explains the separate 'sources' list block and how the pattern matches checkpoint filenames. This is exemplary parameter documentation for a bare schema.

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 opens with a precise verb+resource: 'Persist researched settings so FUTURE sessions start smarter.' This clearly expresses a write/update operation over a learning/guidance resource, which is readily distinguishable from the retrieval-oriented sibling get_model_guidance and from organize_workflow. The intent is unambiguous even without reading 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?

The description is rich with intra-tool guidance: when updating provide the guidance shape, for a NEW family include a detect block, and include a sources list only to introduce URLs. However, it does not explicitly say when to use this tool vs. the alternatives (e.g., get_model_guidance for the read side of the same data), nor does it state any exclusions or conditions that would route an agent elsewhere.

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