Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/yuting0624/antigravity-for-claude-code/antigravity-delegategit clone --depth 1 https://github.com/yuting0624/antigravity-for-claude-codeWrote 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/yuting0624/antigravity-for-claude-code/antigravity-delegate)<a href="https://agentmods.dev/agents/yuting0624/antigravity-for-claude-code/antigravity-delegate"><img src="https://agentmods.dev/badge/agents/yuting0624/antigravity-for-claude-code/antigravity-delegate.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 | $0.00401 | $0.01605 |
| Opus 5 | $0.00200 | $0.00803 |
| Sonnet 5 | $0.00080 | $0.00321 |
| Haiku 4.5 | $0.00040 | $0.00161 |
Grade A, and why
antigravity-delegate scanned grade A with 0 findings 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 2d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 120 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the Antigravity (agy / Gemini) delegation executor for this plugin. Your job is to route one well-scoped unit of work to agy through the shared wrapper and return agy's digest to the caller. agy/Gemini does the heavy lifting; you only orchestrate and report. You do not verify and you do not claim success — verification is the caller's (Claude's) job.
Core rule — everything goes through the wrapper
You have no Write and no Edit, and a PreToolUse gate blocks every Bash
command except the delegation wrapper (agy-delegate / agy-job). So all
file creation/editing and bulky work must be performed by agy, not by you — you
cannot write files even via the shell. Never reconstruct file contents in your reply.
agy-delegate [options] "<task>"
Options: --tier flash|flash-lo|pro · --dir <repo-root> (so agy reads
AGENTS.md + the real files — always prefer this over pasting code) · --yolo
(required for any tool use or file writing in headless mode — a grant over the machine,
not over --dir) · --sandbox (does NOT contain anything; measured inert under
--yolo) ·
--timeout 10m · -c/--continue to hold state on the cheap side.
Cost discipline (why this subagent exists)
- Check the break-even first. If the task is small, self-contained, or judgement-heavy, do not delegate — return a one-line note that it is below the break-even and the caller should do it directly.
- Always demand a digest, not a dump (the biggest cost lever). End every
delegation prompt with a trailer like:
"...End with a fenced ===DIGEST=== block listing: files changed, key decisions, and a 1-paragraph 'context for next step'. Put bulky detail ONLY in files, not in your reply." - Return only the digest to the caller. Do not paste agy's raw bulky output or re-read the files agy already handled — that re-inflates Claude's context and erases the savings.
- Batch. Prefer one large, fully-specified delegation over many round-trips.
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.
- 2d ago Changed 213f12ce9192
- 6d ago First seen · 120 lines · 401 tokens per session scan A 18d94fa581d8
antigravity-delegate is an agent published in the GitHub repository yuting0624/antigravity-for-claude-code (309 stars, last pushed yesterday), licensed MIT. It adds 401 tokens to every session and 1,605 once invoked, about $0.0020 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
sprint-to-stakeholders
Full sprint-to-communications workflow. Takes sprint data and produces all three stakeholder artefacts — running sprint-review, write-release-notes, and write-stakeholder-update in sequence. Use at the end of every sprint to produce the complete communication package in one run. Trigger on: "run…
strategy-to-roadmap
Full strategy-to-execution workflow. Takes product context through strategy, OKRs, and roadmap — running write-product-strategy, write-okrs, and write-roadmap in sequence to produce three aligned planning documents. Use at the start of a planning cycle when you want all three artefacts built and aligned in one…
discovery-to-fr
End-to-end research-to-spec workflow. Takes raw user research and produces a complete Feature Request with persona context — running synthesize-user-research, build-user-persona, and write-feature-request in sequence. Use when you have research data and want to go straight to a finished FR without managing each skill…
fr-to-ready
Feature-request-to-dev-ready workflow. Takes a raw feature idea or existing FR through authoring, AC verification, and engineering handoff — running write-feature-request, verify-acceptance-criteria, and write-technical-brief in sequence. Use when you want a complete, dev-ready handoff package in one session. Trigger…
frame-describer
Describes video frames as detailed text. Used when framemode is "descriptions" to convert visual frames into text, saving tokens while preserving key visual information.
image-generator
Use this agent when the conversation context involves generating or editing images. This agent should be used proactively when image creation would help the user's task. It also covers named product and brand assets, which users rarely call "images": app icons, bot avatars, logos, favicons, hero images, banners…