Skip to main content
Glama
vmware-skills

io.github.zw008/vmware-debug

plan_case

Read-onlyIdempotent

Plan the next diagnostic evidence steps for a VMware case, naming reachable sources and gaps so each submitted round sharpens the following plan.

Instructions

[READ] What to fetch next for this case — step 02, recomputed each call.

WHEN: right after open_case, and again after each round of evidence. It is not a checklist: submit something and the next plan is shorter, lose a source and it routes around it.

RETURNS: {category, category_signals, steps, already_covered, held_back, unavailable, ceiling, note}. Steps are interleaved across evidence classes, so you get breadth before depth — corroboration is counted in distinct sources, which is what actually moves the grade. Each step is {evidence_class, skill, tool, purpose, objects, window, degraded} — call that skill's tool, then submit the result with submit_case_evidence.

GOTCHAS: unavailable is the important half. A source this install cannot reach is listed there with how_to_supply rather than left out, so the gap is visible now instead of when the conclusion refuses to firm up. An empty steps is never silent — note says whether everything reachable is already in, or whether nothing here can be reached. category_signals names the words that chose the category, and any category that also matched — check it first, since the rest runs off that one word.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
case_idYesThe case to plan for (from open_case/list_cases).
categoryNoForce the symptom class instead of inferring it. Exactly one of: storage, network, compute, ha_drs, configuration, accelerator, kubernetes, hardware, host_lifecycle, power_lifecycle, auth, platform, licensing, data_collection. Omit to infer, then read `category_signals` for the word that decided it.
max_stepsNoHow many steps to return, at least 1. The default of 6 is two rounds across three evidence classes — enough to reach the two independent sources a Probable grade costs, without handing a model a fourteen-item menu it will work through. Raise it when `held_back` says there is more for this category; a storage case has fourteen reachable tools. Below 1 is refused rather than obeyed, because an empty plan already means "nothing left to fetch".
available_skillsNoNarrow to the skills actually installed, in either spelling ("monitor" or "vmware-monitor"). Omit to assume all of them, which can produce steps this install cannot run.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv2.3.0

TDQS

A4.8/5.0
Behavior5/5

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

Annotations already establish read-only/idempotent/closed-world, and the description still adds substantial behavior: outputs are recomputed per call, `unavailable` sources carry how_to_supply rather than being dropped, empty `steps` is never silent (a `note` explains why), and `category_signals` should be inspected first. These are real operational traits not derivable from the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The WHEN/RETURNS/GOTCHAS structure front-loads the essentials and the return-shape block is dense but information-bearing. It runs long and leans on prose-heavy gotchas, but almost every sentence carries distinct guidance rather than padding.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

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 must describe the return contract itself, and it does: the top-level keys, the per-step shape {evidence_class, skill, tool, purpose, objects, window, degraded}, and edge-case semantics. For a 4-parameter planning tool, an agent has everything needed to call and interpret it correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is already 100%, so 3 is the floor; the description goes beyond it by explaining the rationale for max_steps' default (6 = two rounds across three classes, and two independent sources to reach Probable), the below-1 refusal behavior, and how category inference ties to category_signals. It clarifies the consequence of parameters rather than restating them.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Opens with a specific verb+resource: '[READ] What to fetch next for this case — step 02, recomputed each call.' The scope (next-fetch planning for a case) and its read-only nature are unambiguous, and it is clearly separable from siblings like submit_case_evidence, grade_case, and open_case.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The WHEN block gives explicit timing: 'right after open_case, and again after each round of evidence.' It also states the interactive contract with a named sibling ('call that skill's tool, then submit the result with submit_case_evidence') and clarifies it is not a checklist, so the agent knows both when and how to re-invoke it.

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