AI-native observability and security for engineering teams and coding agents. Search production logs in plain language, investigate issues and root causes, and create or edit live dashboards in your Sazabi workspace. Requires a Sazabi account (OAuth, or a secret key via the X-Sazabi-Secret-Key header).
Scored across 2 tools
The two tools split cleanly along discovery vs invocation: `search` finds connectors/tools and loads input schemas, `execute` runs a named tool. There is no overlap in the resource+action each targets, and the descriptions explicitly cross-reference the intended search→execute workflow.
Both tools use the same convention: a single lowercase verb (`search`, `execute`). There are no mixed casing or verb-style deviations, and the pattern is exactly what a two-operation gateway needs.
Two tools is intentionally minimal for a gateway/proxy: discovery plus invocation is the smallest complete surface, and extra operations (schema loading, run polling, connector listing) are folded into these two. It is well-scoped, though the thin surface means agents get no static argument schema help up front.
Search covers connector listing, per-connection tool listing, schema loading, and API-only scoping; execute covers invocation, deferred-run polling via `sazabi/runs.get`, and structured error retry. Minor gaps exist around run cancellation/management and batch invocation, but the core discover-invoke-poll lifecycle is covered.
Invoke a tool by pasting name from search, plus arguments.
name is sazabi/<operationId> (Sazabi API) or <connectionKey>/<toolName> (connected MCP). Example: { "name": "sazabi/logs.ask", "arguments": { ... } }.
Optional projectId targets one project's connectors for this call without reconnecting. Org-scoped credentials (org secret key, OAuth) must pass it to reach connectors; project-scoped keys are locked to their own project. Sazabi sazabi/<op> calls carry projectId inside their own arguments instead.
Deferred Sazabi operations wait up to ~25s; if unfinished, poll with name "sazabi/runs.get" and arguments { runId }. Pass wait: false in their arguments to get the run handle back at once instead.
A rejected call returns the API's error code, message, and offending fields: fix the arguments and retry rather than repeating the same call.
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes | Copy-paste name from search: sazabi/<operationId> or <connectionKey>/<toolName>. | |
| arguments | No | Arguments matching the hit's inputSchema. | |
| projectId | No | Target project's connectors for this call. Overrides the credential's default project; org-scoped credentials must pass it to reach connectors. Sazabi sazabi/<op> calls carry projectId in their own arguments instead. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare this tool as non-read-only and destructive, and the description adds valuable behavioral context beyond that: deferred calls can wait ~25s, polling is available, and rejected calls return error code, message, and offending fields. It also explains credential-scoping behavior for `projectId`, though it does not detail what kinds of side effects invoked tools may have.
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 front-loaded with the core action and name format, then adds scoping, deferred behavior, and error handling. It is appropriately detailed for a dispatcher tool, though some `projectId` details repeat the schema and could be tightened.
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?
For a complex invocation tool with no output schema, the description covers the essential call mechanics, name formats, project credential scoping, timeout/polling behavior, and error handling. It does not describe the successful return shape, but the deferred-run and error notes give enough operational context.
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, but the description adds meaningful semantics: `name` must be `sazabi/<operationId>` or `<connectionKey>/<toolName>`, `arguments` must match the hit's inputSchema, and `projectId` has credential-scope-dependent behavior. The example further clarifies the expected call shape.
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 action: invoke a tool by pasting a `name` from search plus `arguments`, and it clarifies the two valid name forms with an example. It implicitly distinguishes this execution tool from the sibling `search` by sourcing the name from search results, though it does not explicitly say 'use search first to discover tools.'
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?
It gives clear usage context: copy the name from search, pass matching arguments, and use `projectId` according to credential scope. It also provides outcome-specific guidance for deferred operations (poll with `sazabi/runs.get` or pass `wait: false`) and rejected calls (fix arguments and retry).
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search this project's connected MCP tools and the Sazabi API. This project: vendor connectors are loaded after a project-scoped secret key authenticates. Empty query lists connectors (not the SDK catalog). If a query matches a provider with multiple connections, search returns those connections — pass namespace= to list that connection's tools. Hits are compact; repeat with the hit's namespace, copy its name as query, and set includeInputSchema=true to load exactly that argument schema. Schema detail never fuzzy-matches. Each hit includes a copy-paste name for execute({ name, arguments }). namespace=sazabi searches only the Sazabi API.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | Maximum connector or tool hits. Default 10; compact discovery returns at most 20 per page. | |
| query | No | Free-text query. Omit or empty to list connectors. A provider name with multiple connections returns those connections instead of blended tools. | |
| offset | No | Hits to skip. Default 0. | |
| namespace | No | connectionKey to list/search one connector, or `sazabi` for the Sazabi API catalog only. | |
| projectId | No | Project whose connectors to search. Overrides the credential's default project; org-scoped credentials must pass it to list connectors. | |
| includeInputSchema | No | Load the exact argument schema for one selected hit. Requires namespace and a non-empty query copied from the hit's name; schema-detail mode never fuzzy-matches and always returns at most one hit. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Beyond readOnlyHint=true, the description discloses important behaviors: connectors load after project-scoped secret-key authentication, empty queries list connectors, compact hits may need a follow-up exact schema lookup, schema-detail mode never fuzzy-matches and returns at most one hit, and compact discovery paginates at at most 20 per page.
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 front-loaded with the core purpose and packs only operationally relevant details. It is dense and could be more scannable, but no sentence is clearly wasted.
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?
For a six-parameter discovery tool with no output schema, the description covers the missing context an agent needs: authentication prerequisite, empty-query behavior, connection-specific namespace behavior, follow-up schema loading, pagination limits, and the handoff to execute.
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 defines each parameter. The description still adds meaningful relationships: namespace selects a connectionKey or sazabi, includeInputSchema requires namespace plus a non-empty query copied from the hit's name, and projectId matters for org-scoped credentials.
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 and resource: search connected MCP tools and the Sazabi API. It clearly positions the tool as discovery for execute, including how hits can be supplied to execute({ name, arguments }).
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?
It gives explicit usage rules: empty query lists connectors rather than the SDK catalog, a provider with multiple connections returns those connections, namespace scopes to one connection or the Sazabi API, and includeInputSchema is used only after selecting a hit by namespace and copied name.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
executesearch
Add one secure layer between your agents and this server.