claude-agy-mcp

Overview Schema Source Related Servers Discussions Admin

follow_up

Ask follow-up questions about a prior delegation by reusing its session_id, so the full context stays on the Antigravity side; enable write:true to permit file edits.

Instructions

Continue a previous Antigravity session by session_id (returned by every other tool). USE THIS for follow-up questions about a prior delegation — the full prior context is already on agy's side, so you don't resend anything. Pass model to get a second opinion on the same history from a different model without re-sending it. Read-only by default; pass write: true to rework a delegation that edited files.

Input Schema

TableJSON Schema

Name	Required	Description
`cwd`	No	Absolute path to the working directory / project root. Defaults to the server's cwd.
`dirs`	No	Extra directories to add to agy's workspace, for cross-repo or worktree-vs-base analysis.
`model`	No	Override the model. Accepts a display name ("Gemini 3.8 Flash (High)"), an id ("gemini-3.8-flash-high"), or a family selector ("gemini-pro@latest-high"). Normally omit — the tool routes automatically.
`write`	No	Allow file edits and command execution. Off by default.
`effort`	No	Reasoning tier. Gemini models carry their tier in the name, so this switches to the sibling model at that tier (Flash (High) -> Flash (Medium)); agy's own --effort flag only reaches models without a tier. Defaults to the user's set_model choice, else the tool's own tier.
`schema`	No	JSON Schema (as a string) constraining the answer. Returns structuredContent as well as text. Costs noticeably more latency and tokens, so use it only when you will parse the result.
`question`	Yes	The follow-up question.
`session_id`	Yes	The session id returned by a previous claude-agy-mcp call.
`slash_commands`	No	Let agy expand your slash commands and skills into the prompt. Off by default: a hostile file in the workspace can otherwise steer the delegated model.

Schema Changelog

Changes observed during successful MCP inspections.

Changed11 schema fields changedv3.1.0Sep 13, 2026
- addedInput schema / properties / cwd / minLength
```
Added value: +1
```
- addedInput schema / properties / dirs / items / minLength
```
Added value: +1
```
- addedInput schema / properties / dirs / maxItems
```
Added value: +256
```
- addedInput schema / properties / model / maxLength
```
Added value: +512
```
- addedInput schema / properties / model / minLength
```
Added value: +1
```
- addedInput schema / properties / model / pattern
```
Added value: +"^[^-]"
```
- addedInput schema / properties / question / pattern
```
Added value: +"\\S"
```
- addedInput schema / properties / session_id / maxLength
```
Added value: +512
```
- addedInput schema / properties / session_id / minLength
```
Added value: +1
```
- addedInput schema / properties / session_id / pattern
```
Added value: +"^[^-]"
```
- addedInput schema / properties / write
```
Added value: +{
+  "description": "Allow file edits and command execution. Off by default.",
+  "type": "boolean"
+}
```

Changed5 schema fields changedv2.1.0Sep 12, 2026

addedInput schema / properties / dirs

Added value: +{
+  "description": "Extra directories to add to agy's workspace, for cross-repo or worktree-vs-base analysis.",
+  "items": {
+    "type": "string"
+  },
+  "type": "array"
+}

addedInput schema / properties / effort

Added value: +{
+  "description": "Reasoning tier. Gemini models carry their tier in the name, so this switches to the sibling model at that tier (Flash (High) -> Flash (Medium)); agy's own --effort flag only reaches models without a tier. Defaults to the user's set_model choice, else the tool's own tier.",
+  "enum": [
+    "low",
+    "medium",
+    "high"
+  ],
+  "type": "string"
+}

changedInput schema / properties / model / description

Previous value: -"Override the model (exact name from `agy models`, e.g. \"Gemini 3.1 Pro (High)\"). Normally omit — the tool routes automatically."New value: +"Override the model. Accepts a display name (\"Gemini 3.8 Flash (High)\"), an id (\"gemini-3.8-flash-high\"), or a family selector (\"gemini-pro@latest-high\"). Normally omit — the tool routes automatically."

addedInput schema / properties / schema

Added value: +{
+  "description": "JSON Schema (as a string) constraining the answer. Returns structuredContent as well as text. Costs noticeably more latency and tokens, so use it only when you will parse the result.",
+  "type": "string"
+}

addedInput schema / properties / slash_commands

Added value: +{
+  "description": "Let agy expand your slash commands and skills into the prompt. Off by default: a hostile file in the workspace can otherwise steer the delegated model.",
+  "type": "boolean"
+}

Changed1 schema field changedv2.0.0Aug 19, 2026
- removedInput schema / additionalProperties
```
Removed value: -false
```
First observedv1.0.0Aug 14, 2026

TDQS

A4.3/5.0

Behavior4/5

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

With no annotations, the description carries real weight: it discloses that prior context lives on agy's side so nothing is resent, that the tool is read-only by default with `write: true` enabling file edits and command execution, and that slash_commands is off by default due to a prompt-injection risk. That security rationale is exactly the kind of context annotations would normally provide.

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?

Front-loaded with the core purpose, then layered behavior and mode flags in short clauses; there is little filler. It runs slightly dense at four sentences with a mid-sentence caveat about agy's --effort handling, but every sentence adds callable information.

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

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 9-parameter, no-output-schema tool with zero annotations, the description covers the non-obvious risks (mutation default, injection exposure, latency cost of `schema`) that an agent most needs. Less obvious parameters (cwd, dirs, effort) are documented in the schema, so the overall picture is adequate.

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 100%, so a 3 is the baseline, but the description adds genuine meaning beyond the schema for the two most consequential flags — `write` toggles mutating behavior and `model` enables cross-model second opinions on the same history. The remaining parameters are left to the schema.

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: 'Continue a previous Antigravity session by session_id', and immediately scopes it to follow-up questions about a prior delegation. That phrasing, plus the note that session_id is 'returned by every other tool', distinguishes it cleanly from delegate/delegate_many.

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

Usage Guidelines4/5

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

Explicit context for use ('USE THIS for follow-up questions about a prior delegation') and a concrete alternate mode (pass `model` for a second opinion on the same history). It implies but never states the inverse exclusion — i.e. use `delegate` for brand-new work — so the routing guidance is strong but not fully closed.

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

Deploy Server

follow_up

Instructions

Input Schema

Schema Changelog

TDQS

Other Tools