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 skills add OpenDCAI/Dataflow-LoopAI --skill judgergit clone --depth 1 https://github.com/OpenDCAI/Dataflow-LoopAIWrote 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/opendcai/dataflow-loopai/judger)<a href="https://agentmods.dev/skills/opendcai/dataflow-loopai/judger"><img src="https://agentmods.dev/badge/skills/opendcai/dataflow-loopai/judger/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/opendcai/dataflow-loopai/judger"><img src="https://agentmods.dev/badge/skills/opendcai/dataflow-loopai/judger.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00000 | $0.06568 |
| Opus 5.5 | $0.00000 | $0.02627 |
| Sonnet 5.5 | $0.00000 | $0.01314 |
| Haiku 4.5 | $0.00000 | $0.00657 |
Grade A, and why
Judger 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 16d 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 — 398 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Judger Skill
Purpose
无 LangGraph 的独立评测流水线。支持四种任务类型:
- code — 代码生成评测(evalplus 的 HumanEval+ / MBPP+,或 LiveCodeBench),计算 pass@k
- text2sql — SQL 生成评测,SQLite 执行校验
- general_text — 通用文本评测(One-Eval DataFlowEvalTool)
- math — 数学/AIME 评测(生成、答案提取和判分在 Docker 镜像内完成)
How to Invoke
唯一入口:loopai.skills.Judger.run()
DB_PATH 和 TASK_ID 从环境变量自动获取:
DB_PATH=api/db/db.sqlite3 TASK_ID=<task_id> \
python -c "from loopai.skills.Judger import run; run()"
或通过 CLI:
DB_PATH=api/db/db.sqlite3 TASK_ID=<task_id> loopai-judger
首次修改 setup.py 后需重新安装项目才会生成 loopai-configer /
loopai-judger 命令;开发环境也可直接用
python -m loopai.skills.Configer.cli 和 python -m loopai.skills.Judger.cli。
Configuration
配置通过 Configer skill 写入 TaskModel.state,分两部分:
全局字段(state["judger"] 顶层,所有 bench 共享)
| 字段 | 默认值 | 说明 |
|---|---|---|
eval_model_path |
无 | 模型路径(必填) |
eval_temperature |
0 |
采样温度,bench 可覆盖 |
eval_top_p |
0.95 |
Top-P 采样,bench 可覆盖 |
eval_max_tokens |
16384 |
最大输出 token 数(含思考推理),bench 可覆盖。vLLM 以 --generation-config vllm 启动,模型目录 generation_config.json 的 max_new_tokens 不会再把上限压低(如 Qwen3-8B-Base 的 2048) |
eval_enable_thinking |
不设置 | 思考模式开关(None 跟随模型默认 / True 开 / False 关),bench 可覆盖 |
eval_batch_size |
10 |
宿主 generate 阶段每批并发多少条 prompt;只对 code 的 evalplus 分支和 text2sql 生效(LiveCodeBench 在容器里自己生成,不吃这个值),bench 可覆盖 |
eval_case_num |
10 |
每问题样本数,bench 可覆盖 |
eval_vllm_tensor_parallel_size |
1 |
vLLM 张量并行数 |
eval_vllm_gpu_memory_utilization |
0.9 |
vLLM GPU 显存利用率 |
cuda_visible_devices |
"0" |
指定 GPU |
output_dir |
"./outputs" |
输出根目录 |
eval_model_name |
空 | /v1/models 暴露的模型名;留空使用 eval_model_path |
eval_top_k / eval_min_p |
-1 / 0 |
math 请求采样参数 |
Bench 配置(state["judger"])
所有评测集通过 benchlist 和 extra_benchlist 列表配置。格式必须是 JSON 数组([{...},{...}]),不是 JSONL(每行一个对象):
{
"benchlist": [
{
"name": "gsm8k",
"task_type": "general_text",
"problem_path": "/data/gsm8k/test.jsonl",
"eval_type": "key2_qa",
"key_mapping": {}
},
{
"name": "human_eval",
"task_type": "code",
"problem_path": "data/evalplus/humaneval_plus.jsonl",
"case_num": 10,
"batch_size": 10,
"format_type": "humaneval+"
},
{
"name": "livecodebench_codegen",
"task_type": "code",
"problem_path": "data/livecodebench/test.jsonl",
"format_type": "livecodebench",
"lcb_scenario": "codegeneration",
"case_num": 1
},
{
"name": "bird_dev",
"task_type": "text2sql",
"problem_path": "/data/bird/dev.jsonl",
"text2sql_dir": "/data/bird/dev_databases",
"case_num": 10,
"batch_size": 10
},
{
"name": "aime26",
"task_type": "math",
"problem_path": "/data/aime26_test.jsonl",
"case_num": 2
}
],
"extra_benchlist": []
}
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.
- 16d ago Changed · +212 lines 27d5dbf45b09
- 1mo ago First seen · 186 lines · 0 tokens per session scan A b5c4d624ecae
Judger is a skill published in the GitHub repository OpenDCAI/Dataflow-LoopAI (29 stars, last pushed 16d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 6,568 tokens. 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 skills, from other repositories
b3os-verification
Skill "b3os-verification" from b3rys/b3rys-team-os, covering b3rys verification — prove it, and keep it proven, 1. three principles, 2. five ratchets — verification that only gets stronger, 3. layers and rubric and 4. proving the checks — mutants.
load-testing
Use when designing and executing load testing, stress testing, endurance testing, spike testing, and performance testing. Includes test planning, test script creation, result analysis, and performance optimization. Based on CPE certification and performance testing best practices.
test-automation-architecture
Use when designing test automation frameworks, implementing page object patterns, creating test reporting systems, and managing parallel test execution. Includes test strategy, test data management, and test infrastructure setup. Based on CTFL-TA and ISTQB Advanced Level certifications.
testing-strategy
Use when designing comprehensive testing strategies, implementing test coverage, setting up CI/CD quality gates, and defining unit, integration, and E2E test approaches. Includes testing pyramid, coverage thresholds, and test automation best practices. Based on software testing certification standards and industry…
tdd-workflow
Test-Driven Development workflow skill with patterns for unit, integration, and e2e testing. Use when implementing features to ensure all code is born tested with comprehensive coverage.
fix-failing-tests
Iteratively fix all failing tests until the test suite is green.