Metis — Verify Text Citations

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

With no annotations, the description carries the full load and does substantial work: it discloses the extraction heuristic (source + page, plus bare DOIs), the deliberate false-negative bias and its rationale, and the ledger write side effect controlled by `record` (default True). It omits whether verification makes external network calls, what happens on lookup failure, and what the verdict categories are, which matters for a tool that clearly reaches beyond local data.

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?

Purpose is front-loaded and the rationale sentence for conservative extraction genuinely earns its place by telling the agent what to expect from results. Minor waste: the 'Returns:' block is redundant given an output schema exists, and the 'Tier A —' prefix carries no information for an agent.

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 3-parameter tool with one required arg, the description covers intent, usage context, extraction behavior, side effects, and (via schema) return shape. The remaining gap is environmental: no mention of whether it needs network access, credentials, or how long a large text block might take.

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 description coverage is 0%, so the description must compensate entirely — and it does. `text` is defined as the prose to scan, `artifact_path` is explained by its downstream purpose (recorded in the ledger to enable a specific query), and `record` is given both its effect and its default. Nothing about the three parameters is left ambiguous.

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: 'find every citation in a block of prose and check each one.' That is concrete and actionable. It does not, however, differentiate itself from the several nearby siblings that also verify things (verify_claim, verify_doi, citation_debt), and the 'Tier A' prefix is unexplained jargon.

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 clear positive guidance on when to reach for it: 'on a draft, a lesson, a report section, or anything about to be written to disk.' That covers the trigger condition well. It stops short of naming alternatives or exclusions, so an agent must infer why it would pick this over verify_doi for a single reference.

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