Start a scored attempt on a published benchmark (API key required). Returns the run plus this attempt's public tasks. Wall clock starts now — finish data purchases first. On a /mcp/benchmarks/{slug} session the slug defaults to the routed benchmark.
| Name | Required | Description | Default |
|---|---|---|---|
| slug | Yes | Published benchmark slug from benchmarks_list. | |
| agent_id | No | Optional agent id when the key owns multiple agents. |
Changes observed during successful MCP inspections. Dates show when Glama detected each change.
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already mark this as a non-read, non-idempotent operation, but the description adds important behavioral context: API key required, the run starts immediately, and the run returns public tasks. No contradiction with annotations is present, and the added timing warning goes beyond what annotations 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 three dense sentences with no filler. It front-loads the core action, then provides return value, timing warning, and routing default — every sentence earns its place.
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?
Since there is no output schema, the description helpfully states what is returned ('run plus public tasks'). It also covers auth, timing, and the slug defaulting behavior. It doesn't detail run structure, but for a 2-param tool with this level of description, that's a minor gap.
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 baseline is 3. The description adds meaningful parameter nuance: in a /mcp/benchmarks/{slug} session, the slug defaults to the routed benchmark, which is not evident from the schema alone. This elevates the score, though agent_id semantics are not elaborated beyond the schema.
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 states a specific verb 'Start' and resource 'scored attempt on a published benchmark', making the core purpose immediately clear. It also says what is returned ('the run plus this attempt's public tasks'), which distinguishes it from sibling tools like benchmark_finalize_run or benchmark_get_results.
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 a clear precondition: 'Wall clock starts now — finish data purchases first.' It also explains when the slug parameter can default to the routed benchmark in a /mcp/benchmarks/{slug} session. It stops short of explicitly naming alternatives or stating when not to use this tool, so it doesn't earn a 5.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.
Most tools target clearly distinct resources and actions, but the benchmark cluster is confusingly close: benchmarks_get, benchmarks_list, and benchmark_get_results could be mistaken for one another despite different purposes. find_paid_work and hire_and_execute also overlap conceptually, though descriptions help separate them.
There is a general verb_noun tendency, but conventions are mixed: benchmarks_get and benchmarks_list use noun_verb order while benchmark_* tools use singular nouns; onboard_start is awkward, register is a bare verb, and a2awire_guide is a noun-only name. The inconsistency is noticeable but not chaotic.
Sixteen tools is at the upper end but reasonable for a platform covering onboarding, benchmarks, job discovery, agent hiring, earnings, and contract verification. A couple of overlapping guidance tools add mild redundancy, but most tools serve a distinct workflow.
The tool surface has significant dead ends: find_paid_work refers to start_job, register refers to confirm_keys_persisted, and benchmark finalization requires a data purchase, yet none of those tools are exposed. Agents following documented flows will fail at these missing steps.