Metis — Show Learned Preferences

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

With no annotations, the description must carry the behavioral burden. It implies this is a passive inspection/reporting tool ('to make the learning auditable'), which is valuable context beyond the name, but it says nothing about what happens when no feedback exists, whether it is read-only, or how volatile the reported model is. Useful framing, incomplete operational disclosure.

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 first sentence is well front-loaded and earns its place, but the following two sentences are changelog and design-rationale ('Added 2026-09-10, when the feedback loop was finally wired...') that do not help an agent decide or call correctly. They add flavor at the cost of signal density.

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 zero-parameter reporting tool with an output schema, the essentials are covered: what is reported and how it is partitioned. Minor gaps remain around empty-state behavior and the correction path implied by 'a preference model you cannot inspect is one you cannot correct', which no sibling is named for.

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 tool takes zero parameters, so there is nothing for the description to disambiguate. Baseline 4 applies; the description only adds that results are split by news versus papers, which is a return-shape hint rather than parameter semantics.

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?

It names the specific resource — the preference model learned from keep/decline verdicts — and its scope ('separately for news and papers'). The phrasing is a bit rhetorical ('what your keeps and declines have taught Metis'), but an agent can tell it apart from generic profile tools like get_user_profile or get_user_topics. It does not explicitly name a sibling it replaces, so no 5.

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 use case is only implied through the reflection that the model must be auditable and correctable; there is no explicit 'use this when...' or a named alternative such as show_agent_decisions. An agent can infer the tool is for inspecting/correcting learned preferences, but must reason about it.

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