rationale_search

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

With no annotations available, the description carries the full burden. The conditional `record` field disclosure ('A hit whose thread has a distilled decision record carries it as `record`') is genuinely useful and reveals result-shape behavior beyond a naive reading. However, the 'cached' aspect isn't qualified (staleness, when to trust vs refresh), and the `worktree` edge case ('silently ignored') only appears in schema, not description. Core search behavior covered, supporting details absent.

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?

Two sentences, roughly 50 words, and both earn their place. The purpose is front-loaded in sentence one, and the second sentence delivers a valuable behavioral nuance. Slight deduction for the second sentence's convolution ('carries it as `record`' has an awkward antecedent) and the nested parenthetical, but overall this is tight, effective prose.

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 7 params, no annotations, and no output schema, the description covers only the core search contract and the `record` field. Missing: semantics of the graph output params (which appear to control some graph-return mode), what 'cached' means for result freshness, behavior on empty results, and how this output's graph mode relates to `compare_graph_*` siblings. The description would be sufficient for a simpler tool, but not this one.

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 only 29%, putting the burden on the description to compensate. It doesn't: the description's only param-relevant contribution is 'by keyword' (qualifying the query param) and mentioning `record`, which is an output field, not a parameter insight. The opaque `graph_limit` and `include_graph` params (with only the cryptic McpGraphMode enum 'compact'/'full'/'none') remain completely unexplained by the description.

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?

Specific verb+resource: 'Search cached tracker rationale snippets... by keyword' clearly identifies the search target and scope. Parenthetical '(review comments, issue / change-request discussion)' and the keyword-vs-semantic implication distinguish it from many siblings. However, it doesn't explicitly differentiate from closely related search tools (`semantic_search`, `memory_search`, `papertrail_issue_search`), so what exactly makes this corpus distinct from those is left implied.

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 parenthetical source specification gives clear context about what content is covered, which implicitly signals when this tool applies. However, there are no explicit when/when-not-to-use instructions, no named alternatives, and no distinction from sibling search tools. For an agent choosing among `rationale_search`, `semantic_search`, `memory_search`, or `commit_search`, no decision support is offered.

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