agy-review
Skill2 starsChanged 5 months agoMITView on GitHub
Adversarially review anything - a plan, design, code change, piece of writing, decision, argument, config, or idea - with the `agy` (Antigravity) CLI. A fast second opinion from an independent model that pokes holes in whatever you give it - questionable assumptions, flaws, risks, gaps, failure modes, and a SHIP/REVISE/RETHINK verdict. Use when you want to red-team / stress-test / sanity-check / critique something, "poke holes in this", "what am I missing", "what's wrong with this", "get a second opinion", "review this with agy", or harden it before committing. Pairs with brainstorming and writing-plans.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- ilya-kozyrev/claude-agent-hubtheir own version, first seen Nov 1, 2026
- ilya-kozyrev/delamaintheir own version, first seen Nov 1, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- ilya-kozyrev/claude-agent-hub★ 5same name
- ilya-kozyrev/delamain★ 5same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 50
Always reconcile the critique with your own reasoning and the actual context.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
agy-review
Run anything past the agy CLI for a quick adversarial critique — questionable assumptions, flaws, risks, gaps, failure modes, and a SHIP/REVISE/RETHINK verdict — from a second, independent model. Use it to harden a thing before you commit to it, not to replace your own judgment.
It reviews whatever you give it — a plan or design is just one case. Also: a code change, an essay or doc, a product/architecture decision, an argument or pitch, a config, a prompt, a name, an idea. Raw code, numbered specs, prose — all fine.
What to review (arguments)
Invoked as /agy-review [text or file to review]. Interpret the argument — $ARGUMENTS — as what to critique:
- a file path → review its contents (pass it with
-f); - pasted text or an inline description → review that;
- nothing → review the most relevant artifact in the current context (e.g. the plan, draft, or decision you just produced), and state what you picked.
How to run it
One script does everything: scripts/agy-review.sh. Give it the content via stdin, a file, or an argument; add -c to say what the thing is so the critique is on-target.
# pipe content in (most common when reviewing your own draft)
printf '%s\n' "$THING" | scripts/agy-review.sh -
# from a file, with context about what it is / the goal
scripts/agy-review.sh -f design.md -c "B2B SaaS onboarding flow; must ship this sprint"
scripts/agy-review.sh -f rate-limiter.md -c "Node/Express service behind a load balancer"It prints the critique to stdout and exits 0. Read the bullets, then weigh them — adopt the real ones, dismiss the off-base ones, and tell the user which is which.
Model
Defaults to Gemini 3.5 Flash (High) (Flash in its highest thinking mode), per preference. Override with AGY_REVIEW_MODEL="<name>" (see agy models for exact strings) or -m.
Reading the result
The last line is VERDICT: SHIP / REVISE / RETHINK - <biggest issue>. Treat it as input, not a verdict you must obey:
- SHIP — no blocking issues found; still skim the bullets.
- REVISE — fix the flagged issues, then proceed.
- RETHINK — it may be fundamentally off; reconsider before committing.
Always reconcile the critique with your own reasoning and the actual context. A second model is good at catching blind spots and bad at knowing your context — keep what's true, drop what isn't, and tell the user which is which.
Notes
- Needs the
agyCLI onPATH(agy --version). Missing → exit2. - The script runs
agyfrom a throwaway empty directory, so it never reads, edits, or crawls your repo — it just asksagya question and prints the answer. - Reviews typically run in ~5–20s.
- Env knobs:
AGY_REVIEW_MODEL·AGY_PRINT_TIMEOUT(2m) ·AGY_HARD_TIMEOUT(150s). - Exit codes:
0critique printed ·1usage error ·2agynot found or returned nothing.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

