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

With no annotations, the description carries the full burden and does well: it declares read-only semantics, that no validation is performed, the prerequisite that the manifest must exist inside the server working directory, and the hard failure mode otherwise. It stops short of describing idempotency or performance/IO characteristics, but covers the safety-relevant behavior.

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?

Front-loaded with purpose, then usage routing, then prerequisites and returns. The return-field enumeration is long but earns its place because no output schema exists. Slightly dense, but no filler sentences.

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?

No output schema means the description must enumerate return values, and it does (manifest_path, version_ok, loaded counts, loading_errors, warnings). Combined with the stated prerequisites and failure conditions, an agent has everything needed to call it correctly, which is unusual for a tool with no annotations.

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?

One parameter at 100% schema coverage, so the schema already documents the absolute-path requirement and the working-directory rejection. The description restates the same constraint rather than adding format, default, or resolution detail, so the baseline 3 applies.

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 ('Load and inspect a Precis project configuration file') plus a scope qualifier ('without validating data') that immediately separates it from the sibling validate_data. An agent can distinguish it from describe_constraints and infer_schema without opening any 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?

Gives an explicit trigger ('when validate_data fails') and a diagnostic purpose (unsupported version, missing files, dangling references), naming the alternative sibling outright. This is exactly the when/when-not/alternative routing an agent needs.

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

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

With no annotations provided, the description carries the full burden and does well: it declares 'takes no arguments, read-only', discloses the return shape, and notes the content is derived from the actions registry as the single source of truth matching the validation engine. This is real behavioral context beyond a bare description, though it omits things like caching or versioning.

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?

Front-loaded with the core purpose, then usage, then return shape. The parenthetical enumeration of all 10 type names is slightly verbose but genuinely informative since no output schema exists. No wasted filler.

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?

With no output schema and no annotations, the description compensates fully: it explains what is returned (types array with type/refs/params), the provenance of the data, and the read-only no-arg nature. An agent has everything needed to call and interpret this tool correctly.

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 the baseline is 4. The description reinforces this with 'takes no arguments' and instead documents the semantics of the returned refs/params fields, which is useful but not parameter guidance per se.

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: list the 10 constraint types Precis supports, and even enumerates them (NotNull/Unique/.../Composite). An agent immediately knows this is a reference-lookup tool rather than a validator or schema inferrer.

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?

Explicitly says to use it 'when writing or editing *.constraint.yaml files, or when unsure which parameters a constraint type accepts', which is a clear triggering context. It does not name or contrast with the sibling tools (validate_data, check_config, infer_schema), but those have distinct purposes so the risk of confusion is low.

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

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

With no annotations, the description carries the full burden and does so well: it declares read-only behavior, that nothing is written to disk and the caller decides whether to save, the approximate sample size driving inference, the failure mode for paths outside the working directory, and the caveat that out-of-sample extreme values may change the real type.

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?

Front-loaded with the core action, followed by usage, return shape, caveats and constraints in a logical order. It is on the long side and repeats the sample-size default that the schema already states, but every sentence carries usable information.

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?

There is no output schema, so the description justifiably describes the returned schema dictionary and its V2 fields. Combined with the usage trigger, the sampling caveat and the path restriction, an agent has everything needed to call this correctly.

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 100%, so parameters like table_id, sample_rows and source_path are already documented in the schema; the description largely restates the default 1000-row sample and the working-directory path constraint. It adds the semantic link between the draft's column types and the sample, but little beyond structured data.

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 ('infer column types by sampling the head of a data file') and names the artifact produced ('generate a draft V2 schema structure'). It is clearly distinguishable from validate_data, which the description positions as a separate verification step.

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?

Explicitly says when to use it ('when creating a schema for a new data source or rebuilding an existing schema') and points to validate_data as the downstream verification step. It stops short of stating when *not* to use it or naming a true alternative tool for the same job.

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

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

With no annotations, the description carries the full burden and does so: it declares read-only behavior ('modifies no files'), spells out the exact return contract (is_valid, errors shape, summary), and names the failure mode (path outside working dir or missing file returns isError). Preconditions are stated up front. This is unusually complete disclosure for an unannotated tool.

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?

Front-loaded with purpose, then usage, return contract, and preconditions in a logical order; every sentence carries information. It is dense and runs long for a single paragraph, which costs it a point on readability.

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?

No output schema exists, but the description documents the return shape inline and points to docs/contracts/validate-json-v1.md for full field definitions. Preconditions and error behavior are covered, so an agent has everything needed to invoke and interpret the result.

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 100%, so the schema already documents all three parameters, giving a baseline of 3. The description reinforces the manifest constraint (must point to an existing project.precis.yaml inside the working directory) but adds nothing about the table or data_directory parameters beyond what the schema already says.

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+resource ('Validate Precis project data against all constraint rules') and even enumerates the 10 constraint types, which is far more precise than the sibling names check_config, describe_constraints, or infer_schema imply. An agent can tell this is the data-validation tool, not a schema/constraint introspection tool.

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?

Explicitly says when to reach for it: 'check data quality or to re-validate data after editing constraint configuration'. That covers the two main use contexts, but it never contrasts itself against the sibling tools (e.g. describe_constraints) so the routing is inferred rather than stated.

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