Skip to main content

Tool Catalog (mcp_catalog)

mcp_catalog is DocJacket's self-discovery tool. One call returns the live inventory of every tool the connected assistant can use right now — with descriptions, scope tiers, non-obvious gotchas, suggested pairings, and example argument shapes. Help docs that hard-code tool names will drift; mcp_catalog always reflects what's actually deployed.

It's the right answer to "what can my AI assistant actually do with DocJacket?" because:

  • It's scoped to your token. If you granted read-only access at consent, the catalog returns only read tools — no surprises about what writes might fire.
  • It carries metadata that tools/list doesn't. The MCP protocol's standard tools/list returns name + description + input schema. The catalog adds gotchas, pairing hints, example call shapes, and deprecation status — the stuff that helps an assistant plan a workflow, not just find a tool.
  • It's identical regardless of arguments. No parameters, no state. Safe to cache for a conversation; safe to call at the start of any planning step.

How to invoke​

Ask your AI assistant in plain English:

"Use DocJacket's mcp_catalog and group the tools by category." "Show me every DocJacket tool I have access to with this token." "Run mcp_catalog and summarize the tools available to my token."

The assistant calls the tool, gets back the JSON below, and formats it for you. You don't write the call yourself.

A worked example​

A single mcp_catalog response, truncated to representative read-tool entries:

{
  "tools": [
    {
      "name": "list_active_transactions",
      "description": "List every active transaction in the current organization (statuses tagged 'open' — Active, Pending, Under Contract). Returns address, status, closing date, completion %. Pass include_parties=true to also get every party on each deal with their email and role...",
      "requiredScope": "read",
      "hasOutputSchema": false,
      "gotchas": [
        "Capped at 200 transactions; orgs with more should chain search_transactions for the long tail.",
        "statusType='open' covers Active/Pending/Under Contract but excludes Closed — historical context needs search_transactions.",
        "include_parties + include_key_dates each add a per-transaction round-trip; default both off when you only need the address/status list."
      ],
      "pairsWith": [
        "get_next_required_actions",
        "get_transaction",
        "get_missing_documents"
      ],
      "exampleCall": "{\"include_parties\": true, \"limit\": 30}"
    },
    {
      "name": "get_transaction",
      "description": "Full detail on one transaction — parties, key dates, contingencies, tasks, timeline status, and financial terms. Pass the transactionId returned by search_transactions or list_active_transactions.",
      "requiredScope": "read",
      "hasOutputSchema": false,
      "gotchas": [],
      "pairsWith": []
    },
    {
      "name": "get_key_dates",
      "description": "Return all key dates for one transaction, including date, status, source, and linked timeline context.",
      "requiredScope": "read",
      "hasOutputSchema": false,
      "gotchas": [],
      "pairsWith": [
        "get_transaction",
        "get_next_required_actions"
      ]
    }
  ],
  "totalCount": 8
}

What each field means​

FieldWhat it is
nameTool name as you'd pass to tools/call. snake_case, verb-first.
descriptionHuman-readable purpose + when to use it. Same text MCP's standard tools/list returns.
requiredScopeTier scope the caller needs: read for lookups; draft or actions for rollout-gated create, update, send, or apply tools. The catalog only returns tools your token can actually call.
requiredGranularScopeOptional fine-grained scope on top of the tier (e.g. messaging:send, transactions:create). Empty when the tier alone is enough.
hasOutputSchemaWhether the tool declares a structured output schema. true means the response shape is documented and validated; false is text/JSON without a contract.
gotchas[]Non-obvious behaviors worth knowing before calling — caps, scope nuances, side-effects. Empty when nothing surprising.
pairsWith[]Tools that naturally chain with this one. Empty when the tool stands alone.
exampleCallA realistic argument shape, as a JSON string. Useful when figuring out an unfamiliar tool. Absent when the tool has no required arguments.
deprecatedByIf set, the tool is in soft-kill mode; this names the replacement tool to use instead. Absent on live tools.
removalEtaIf set (ISO 8601 date), the date after which the deprecated tool stops working. Absent on live tools.

Why this matters​

Two reasons it's load-bearing for getting good results out of an AI assistant connected to DocJacket:

  1. The assistant plans better when it knows the surface. With gotchas, pairsWith, and exampleCall, the assistant can chain tools intelligently — "I'll call list_active_transactions first with include_parties: true, then get_missing_documents for the deals closing this week" — without you having to spell out the recipe.
  2. You stay in control without memorizing names. You don't need to remember whether the tool is search_transactions or transaction_search. You ask "show me deals at 1234 Main" — the assistant calls mcp_catalog to figure out the right tool, then calls it.

Every customer-facing doc page on this site references tool names; mcp_catalog is the source of truth all of them point at. If anything ever drifts, the catalog wins.

Read more​