Borrowing it
Nothing to install: this file belongs to cloudrift-ai/emmy. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/cloudrift-ai/emmy/main/.agents/skills/refresh-golden/SKILL.mdgit clone --depth 1 https://github.com/cloudrift-ai/emmyWrote 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/skills/cloudrift-ai/emmy/refresh-golden)<a href="https://agentmods.dev/skills/cloudrift-ai/emmy/refresh-golden"><img src="https://agentmods.dev/badge/skills/cloudrift-ai/emmy/refresh-golden/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/cloudrift-ai/emmy/refresh-golden"><img src="https://agentmods.dev/badge/skills/cloudrift-ai/emmy/refresh-golden.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00115 | $0.03178 |
| Opus 5.5 | $0.00046 | $0.01271 |
| Sonnet 5.5 | $0.00023 | $0.00636 |
| Haiku 4.5 | $0.00012 | $0.00318 |
Grade A, and why
refresh-golden 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 yesterday.
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 — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Refresh goldens
The unit of work is one golden file. The input is a golden path, a list of them, or nothing, which means every
repository golden — the same argument shape emmy golden check and emmy golden restamp take. A repository golden is
a hardware golden under emmy/compiler/pipeline/search/golden/records/ or a recipe's
recipes/<model>/golden/<card>.json. It is the tune DB's shape — kernels, routing and rows, beside the traced
programs the kernels came from — and make test asks one question of it, per traced program and without a card:
Is the file the fresh lowering of its own programs? A restamp of it must change nothing: every stored kernel is the kernel the current compiler lowers the program to (the stored body and the fresh one compute the same exact identity), and every kernel-set decision is one the fresh parent takes the same way. A golden stores no identity and no stamps — both are computed from the stored body — so a change to how identity is computed makes no file stale.
A file failing that is stale: its rows are rows of kernels a deploy no longer builds, so a strict boot finds no evidence. The check has no list of expected failures: a red node stays red until the file is refreshed. Never re-record a row to make it green.
Inputs to confirm
- Which files. A path, a recipe (all its cards), or all repository goldens. With no argument, start from
emmy golden checkwith no argument: it names every stale file, and a file it passes needs nothing. - Which cards are reachable. Step 3 runs on the golden's exact card (
gpu_nameandcompute_capin the file head). Without it, the file stops at step 2 with proposals, and the report says so. - Whether deletion is on the table. Step 4 proposes it; the author decides.
Per file
Run the steps below on one file at a time, and finish each file — committed, its test nodes run, its lines in the
two lists removed — before starting the next. Refreshing every golden is this loop over the list emmy golden check
prints, cheapest first: hardware goldens, then recipe goldens by file size; the Qwen3.8 and DeepSeek V100 files lower
whole-layer traces and take minutes each. Group step 3 by card so one rental measures every file that needs it.
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.
- yesterday Changed · +3 lines d08696877d28
- 6d ago Changed · +3 lines c0392729d16b
- 7d ago Changed · +2 lines 3323af853b2c
- 8d ago Changed · +17 lines e35d8ac3f272
- 14d ago Changed · +1 lines 27119a71ad86
- 16d ago Changed · +5 lines 60f35adb17be
- 17d ago Changed · +3 lines f11d7bcbf4dd
- 18d ago Changed · +3 lines b8beb152bd6c
- 19d ago Changed ba56dfa42bb4
- 21d ago First seen · 156 lines · 115 tokens per session scan A db4873e10510
refresh-golden is a skill published in the GitHub repository cloudrift-ai/emmy (116 stars, last pushed today), licensed Apache-2.0. It adds 115 tokens to every session and 3,178 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-09-26.
Other skills, from other repositories
promptfoo-redteam-setup
Create or refine a Promptfoo redteam config and generate probes from target behavior, code, or OpenAPI evidence. Use for purpose, trust boundaries, plugins, strategies, and grading guidance. Use promptfoo-provider-setup for connection work and promptfoo-redteam-run for an existing scan.
promptfoo-provider-setup
Connect Promptfoo to a model, live HTTP API, local Python/JavaScript provider, or app code. Use for request/auth mapping, response parsing, OpenAPI setup, and connection smoke tests. Use promptfoo-evals for broader eval coverage and promptfoo-redteam-setup for attack selection.
promptfoo-redteam-run
Execute, inspect, and rerun an existing Promptfoo redteam scan. Use for generated YAML, result exports, attack success rates, grader/target errors, filtered reruns, and CI gates. Use promptfoo-provider-setup for connections and promptfoo-redteam-setup for new scan plans.
discount-review
Inspect the discount policy fixture with a repeatable review checklist and helper script.
code-review
Reviews code for bugs, security issues, and best practices.
standards-check
Checks that a project follows standard conventions.