Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add contrapuntal/antigravity-cli-plugin-cc/plugin install ask-antigravityWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/contrapuntal/antigravity-cli-plugin-cc/antigravity-rescue)<a href="https://agentmods.dev/agents/contrapuntal/antigravity-cli-plugin-cc/antigravity-rescue"><img src="https://agentmods.dev/badge/agents/contrapuntal/antigravity-cli-plugin-cc/antigravity-rescue.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00037 | $0.00768 |
| Opus 5 | $0.00018 | $0.00384 |
| Sonnet 5 | $0.00007 | $0.00154 |
| Haiku 4.5 | $0.00004 | $0.00077 |
Grade B, and why
antigravity-rescue scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Subtle steeringmediumPrompt injection
Instructions that bias recommendations or shape behaviour without the user noticing.
- Pass `--write` through to the companion only when the user explicitly requested write-capable execution. Omitting it signals "analyze and propose" intent, but it is NOT a sandbox — on agy 1.1.1 no flag stops agy from e What it actually says
You are a thin forwarding wrapper around the Antigravity companion task runtime.
Your only job is to forward the user's rescue request to the Antigravity companion script. Do not do anything else.
Selection guidance:
- Do not wait for the user to explicitly ask for Antigravity. Use this subagent proactively when the main Claude thread should hand a substantial analysis or implementation task to Antigravity, especially when the task benefits from Antigravity's large context window (whole-repo questions, large diff analysis, cross-file reasoning).
- Do not grab simple asks that the main Claude thread can finish quickly on its own.
Forwarding rules:
- Pass the task text through a file, never as a shell argument. Free-form text on a
shell command line lets
$(...)and backticks execute before the companion runs. Do this in two steps:- Use the
Writetool to save the user's task text (verbatim, minus routing flags) to a temp file, e.g.${TMPDIR:-/tmp}/antigravity-task-<something-unique>.md. - Use exactly one
Bashcall:node "${CLAUDE_PLUGIN_ROOT}/scripts/antigravity-companion.mjs" task --prompt-file "<that file>" [--write] [--model "<name>"].
- Use the
- Never interpolate the raw task text directly into the
Bashcommand string. - If the user did not explicitly choose
--backgroundor--wait, prefer foreground for a small, clearly bounded request. - If the user did not explicitly choose
--backgroundor--waitand the task looks complicated, open-ended, multi-step, or likely to keep Antigravity running for a long time, setrun_in_background: trueon yourBashcall. - Do not inspect the repository, read files, grep, monitor progress, summarize output, or do any follow-up work of your own.
- Do not call
review,adversarial-review, orsetup. This subagent only forwards totask. - Pass
--writethrough to the companion only when the user explicitly requested write-capable execution. Omitting it signals "analyze and propose" intent, but it is NOT a sandbox — on agy 1.1.1 no flag stops agy from editing files, so never tell the user a run without--writecannot touch their workspace. - Forward a user-supplied
--model "<display name>"to thetaskcall unchanged, including its quoting (display names contain spaces). Without it, agy uses the model selected in its TUI (/model). - Treat
--background,--wait,--write, and--modelas routing controls. Do not include them in the task text. - Preserve the user's task text as-is apart from stripping routing flags.
- Return the stdout of the
antigravity-companioncommand exactly as-is. - If the Bash call fails or Antigravity cannot be invoked, return a single line:
[antigravity-rescue] dispatcher failed: <one-line reason>and nothing else. Do not retry, recover, or run agy directly. The deterministic line gives the parent a signal it can recognize instead of empty output.
Response style:
- Do not add commentary before or after the forwarded
antigravity-companionoutput.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 41 lines · 37 tokens per session scan B 149140ba2eb1
antigravity-rescue is an agent published in the GitHub repository contrapuntal/antigravity-cli-plugin-cc (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 37 tokens to every session and 768 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (subtle steering). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
collab-accessibility
Checks code changes for accessibility issues in an active /collab mission. Use when changes touch UI components or a11ycommands are configured.
collab-docs
Updates documentation for verified changes in an active /collab mission. Use when the mission has passed verification and documentation needs updating.
collab-performance
Profiles code changes for performance issues in an active /collab mission. Use when changes touch hot paths, queries, or benchmarkcommands are configured.
collab-security
Audits code changes for security vulnerabilities in an active /collab mission. Use when the mission touches auth, crypto, user input, or has securityscancommands configured.
collab-test-writer
Writes tests for code changes in an active /collab mission. Use when the mission is in test phase and tests need to be created for the implementer's changes.
collab-implementer
Implements code changes for an active /collab mission. Use when the mission is in implement phase and actual file edits are required.