Antigravity CLI

# Fix list: Antigravity CLI

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/antigravity-cli, the October 2026 research run, assessed 10 October 2026. Grade C, 57.7 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Antigravity CLI: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Reliability, 43 out of 100, up to 11.4 more on the total

Why it scored 43: Local-package reading. An official installer for macOS, Linux and Windows, with an Android build for Termux, that checks a SHA-512 checksum against Google's release manifest before it installs (20). No public CI or test suite, since the repository holds no source and no workflows, only issue templates (0). Open issues and pull requests could not be read from this session, so the crash and regression line is unchecked (0). Version 1.3.2 is the latest, and version 1.2.10 changed headless exit codes for streamed model errors from 0 to 3 under a Fixed heading, with no breaking-change heading (8). Version 1.x, so stable under the rule (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 2. Payments & pricing, 30 out of 100, up to 8.8 more on the total

Why it scored 30: Harness reading of the published rubric. No payment protocol in the docs, the pricing page or the plans page (0). Plan prices are public without a login, at $20 a month for Google AI Pro, $100 and $200 for Ultra and $30 a seat for Enterprise Standard. Consumption is priced at Gemini Enterprise rates, and those rates were not read, so this is plan-level pricing (10). The pricing page lists a $0 individual plan that needs no subscription. None of the pages read asks for a payment method, but the account flow was not tested, so the point is taken from the page (20). Sign-in is a browser flow with a Google account, or a Gemini API key set by hand, and the CLI won't start unauthenticated (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 3. Security & auth, 53 out of 100, up to 8.2 more on the total

Why it scored 53: Framework reading (credentials and telemetry defaults, approvals and sandboxing, guardrails, audit, security programme), five lines. Credentials are a Google account sign-in kept in the operating system keyring, or a Gemini API key in settings.json. Interactions are collected until the user opts out, and the settings page does not state the default (10). The Default preset sandboxes terminal commands on macOS and Linux, hides ~/.ssh and .env, limits network access to approved domains and asks before MCP tools run, with allow rules by server and tool. The sandbox is labelled preview, the Turbo preset turns it off, and Windows uses the older model (18). Hooks run before and after tool calls and can block one, and the CLI shows a Reason line when a hook flags an action. No documented defence against prompt injection in fetched content was found (7). The CLI log records model resolution, and approvals show in the conversation, but no exportable audit log was found (10). google.com has a valid security.txt with reports to g.co/vulnz, antigravity.google serves none, and no CLI advisory appears in the changelog. The advisory page was not reachable, so that part is unchecked (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 4. Transparency & trust, 54 out of 100, up to 4 more on the total

Made of editorial 52, provenance 56.

Why it scored 54: Closed source, with clear terms. The binary is under the Google Antigravity Additional Terms, the Universal Terms and the Privacy Policy (15). The terms say Interactions are used to evaluate, develop and improve Google and Alphabet products and machine learning, that Google employees and contractors may review them, and that the user can ask for deletion by email. The README and settings page agree with that, but no retention period appears in the pages read (15). Dated notices exist for the Gemini CLI transition and for the removal of Claude Sonnet 4.6, Claude Opus 4.6 and GPT-OSS 120B on 2 November 2026, but there is no written deprecation policy for the CLI (10). Enable Telemetry in the Account settings controls sharing of Interactions and can be turned off. The settings page describes the same key as anonymous usage statistics and crash reports, and its default is not stated (12).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: antigravity.google, no registry record we could read (0 of 15)
- Privacy policy: read, states 7 of the 8 things a reader expects, and has 1 clause that costs points (7.3 of 10)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 5. Schema & documentation, 78 out of 100, up to 3.6 more on the total

Why it scored 78: Framework reading. No published JSON Schema for settings.json, mcp_config.json or hooks.json was found, although the docs type each key (10). llms.txt at antigravity.google/llms.txt, and every docs page has a Markdown copy (10). Pages say when to use each permission preset, what each allows and when the sandbox applies (15). Presets, output formats, modes and effort levels are enumerated (13). The headless page documents the output envelope, the status values and exit codes, with a jq example (15). A dated changelog on the docs site and versioned release notes in the repository (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 6. Agent ergonomics, 78 out of 100, up to 3.6 more on the total

Why it scored 78: Framework reading, adapted to a harness driven by a pipeline. MCP tools can be allowed or refused by server and tool, and a 20,000-token budget for user rules stops large rule sets from crowding out skills and MCP tools (16). No turn or spend cap appears in the flag reference, and `--print-timeout` sets a wall-clock limit only (10). The JSON envelope carries a status field and token usage, and the stream-json result event carries token counts (16). Exit codes 0, 1, 2 and 3 are documented, but a tool soft-denied in headless mode still exits 0 (14). `--continue` and `--conversation` resume earlier runs (14). A Python SDK on PyPI, which is a separate package, and no second SDK language for the CLI (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 7. Maintenance & community, 68 out of 100, up to 2.8 more on the total

Why it scored 68: 1.3.2 on 8 October 2026 (30). 47 numbered releases in the changelog from 1.1.2 (13 July) to 1.3.2 (20). Responsiveness rests on a public changelog, a community forum linked from the terms and the GitHub repository, but no replies could be read (8). The google-antigravity Python SDK, version 0.1.21 released on 7 October 2026 under Apache-2.0, is a separate package and no SDK for the CLI itself was found (10). No public CI or dependency list for the CLI, since the source is not published (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: GitHub issues, pull requests and security advisories for google-antigravity/antigravity-cli (github.com answered 403 for the issues page and the GitHub API is closed to this session)
- unchecked: whether the $0 individual plan asks for a payment method, since the account flow was not tested
- unchecked: the default of Enable Telemetry, which the settings page does not state, and the retention period for Interactions
- unchecked: an incident history for the CLI or its quota service, since no status page was found
- unchecked: GitHub stars, so popularity has no count
- The docs say --print-timeout defaults to five minutes, while agy --help says 0s
- The settings page describes Enable Telemetry twice, as sharing Interactions and as anonymous usage statistics and crash reports
- The docs changelog marks v1.3.1 as the latest release on 10 October 2026, while the repository changelog and the release manifest show 1.3.2 from 8 October
- The Windows sandbox default is not stated in the pages read

## Weaknesses

- Closed source. The repository holds the README, changelog and examples, with no source and no licence file
- Interactions are used to improve Google and Alphabet products and models, and Google staff and contractors may review them, unless the user opts out
- The sandbox is labelled preview in the permissions docs, and Windows still uses the older permission model that the docs describe separately
- A headless run exits 0 when a tool that needs approval is refused, and the only sign is a notice on stderr
- The plans page gives no request counts for the weekly quota, which Google sets by available capacity

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Read the `status` field from `--output-format json` and read stderr too. A tool refused for lack of approval still exits 0
- Pre-approve each shell command and tool a headless run needs with `action(target)` rules under `permissions.allow` in ~/.gemini/antigravity-cli/settings.json, because headless runs cannot ask
- Export `GEMINI_API_KEY` after setting modelProvider to gemini. The CLI won't start when the key is missing, and setting the variable alone has no effect
- Turn off Enable Telemetry in the Account settings before running on private repositories if Interactions should not be used for product improvement
- Set `--print-timeout` explicitly. The docs give a five-minute default, while `agy --help` shows 0s

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.