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

Annotations already declare readOnly, openWorld, idempotent, and non-destructive behavior, but the description adds substantial non-obvious behavior: skipped symbols are listed with reasons rather than silently omitted, missing inputs are never zero-filled, the routine is arithmetic over supplied rows rather than a trading view, and it composes rather than fetches. This goes well beyond what annotations and schema can convey.

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 long but deliberately structured with distinct bullet points and warnings. It front-loads the core purpose and then uses each section to add operational context. Some lines are emphatic to the point of verbosity, but every section contributes behaviorally relevant information rather than 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?

Given the six parameters, existing output schema, and rich annotations, the description covers the important contextual ground: what the tool runs, where rows come from, how skipped symbols are reported, how as_of is inherited, and the zero-substitution hazard. The main residual gap is full semantic detail for a few parameters, but the description is strong enough for an agent to invoke the tool correctly in most cases.

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 carries the full parameter-semantics burden. It names all six arguments and gives real meaning to recipe, rows, and as_of, plus a warning about zero substitution. However, today, top_n, and min_yield are left mostly to inference; there are no per-parameter definitions beyond routine names and defaults in the schema. This is helpful but not complete.

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 specific verb and resource: 'Replay a saved multi-step routine over rows you fetched.' It also names the three concrete routines and explicitly contrasts itself with fetch-style tools: 'IT COMPOSES, IT DOES NOT FETCH. Rows come from query_dataset.' This makes the tool's role unmistakable and distinguishes it from siblings like query_dataset and run_research.

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 gives clear context on when to use the tool: after fetching rows via query_dataset, and it warns that a saved routine is not a trading view. It also explains as_of inheritance, which sets expectations about the call chain. It does not explicitly enumerate alternatives such as run_backtest or run_research, so the exclusion is mostly implied rather than fully explicit.

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