@baimingtao/lbs-agent 1.0.0

package/extensions/skills/optional/research/bioinformatics/SKILL.md

@@ -0,0 +1,235 @@
+---
+name: bioinformatics
+description: 连接 bioSkills 和 ClawBio 400 多项生物信息学技能的网关。涵盖 genomics、transcriptomics、单细胞、variant calling、药物基因组学、metagenomics、结构生物学等。支持按需获取特定领域的参考资料。
+version: 1.0.0
+platforms: [linux, macos]
+metadata:
+  hermes:
+    tags: [bioinformatics, genomics, sequencing, biology, research, science]
+    category: research
+---
+
+# 生物信息学技能网关
+
+当被问及 bioinformatics、genomics、sequencing、variant calling、基因表达、单细胞分析、蛋白质结构、药物基因组学、metagenomics、系统发育学或任何计算生物学任务时，请使用此技能。
+
+此技能是通往两个开源生物信息学技能库的网关。它没有将数百个特定领域的技能打包在一起，而是对它们建立索引，并按需获取您所需的内容。
+
+## 来源
+
+◆ **bioSkills** — 385 项参考技能（代码模式、参数指南、决策树）
+  仓库：https://github.com/GPTomics/bioSkills
+  格式：每个主题一个包含代码示例的 SKILL.md。Python/R/CLI。
+
+◆ **ClawBio** — 33 项可运行的流水线技能（可执行脚本、可复现性包）
+  仓库：https://github.com/ClawBio/ClawBio
+  格式：带有演示的 Python 脚本。每项分析都会导出 report.md + commands.sh + environment.yml。
+
+## 如何获取和使用技能
+
+1. 从下方的索引中确定领域和技能名称。
+2. 克隆相关仓库（使用浅克隆以节省时间）：
+   ```bash
+   # bioSkills (reference material)
+   git clone --depth 1 https://github.com/GPTomics/bioSkills.git /tmp/bioSkills
+
+   # ClawBio (runnable pipelines)
+   git clone --depth 1 https://github.com/ClawBio/ClawBio.git /tmp/ClawBio
+   ```
+3. 阅读特定技能：
+   ```bash
+   # bioSkills — each skill is at: <category>/<skill-name>/SKILL.md
+   cat /tmp/bioSkills/variant-calling/gatk-variant-calling/SKILL.md
+
+   # ClawBio — each skill is at: skills/<skill-name>/
+   cat /tmp/ClawBio/skills/pharmgx-reporter/README.md
+   ```
+4. 将获取的技能作为参考资料遵循。这些**不是** Hermes 格式的技能 —— 请将它们视为专业的领域指南。它们包含正确的参数、合适的工具标志和经过验证的流水线。
+
+## 按领域划分的技能索引
+
+### 序列基础
+bioSkills:
+  sequence-io/ — read-sequences, write-sequences, format-conversion, batch-processing, compressed-files, fastq-quality, filter-sequences, paired-end-fastq, sequence-statistics
+  sequence-manipulation/ — seq-objects, reverse-complement, transcription-translation, motif-search, codon-usage, sequence-properties, sequence-slicing
+ClawBio:
+  seq-wrangler — 序列质量控制 (QC)、比对和 BAM 处理（封装了 FastQC、BWA、SAMtools）
+
+### 读取质量控制和比对
+bioSkills:
+  read-qc/ — quality-reports, fastp-workflow, adapter-trimming, quality-filtering, umi-processing, contamination-screening, rnaseq-qc
+  read-alignment/ — bwa-alignment, star-alignment, hisat2-alignment, bowtie2-alignment
+  alignment-files/ — sam-bam-basics, alignment-sorting, alignment-filtering, bam-statistics, duplicate-handling, pileup-generation
+
+### Variant Calling 与注释
+bioSkills:
+  variant-calling/ — gatk-variant-calling, deepvariant, variant-calling (bcftools), joint-calling, structural-variant-calling, filtering-best-practices, variant-annotation, variant-normalization, vcf-basics, vcf-manipulation, vcf-statistics, consensus-sequences, clinical-interpretation
+ClawBio:
+  vcf-annotator — 具备祖源感知上下文的 VEP + ClinVar + gnomAD 注释
+  variant-annotation — Variant 注释流水线
+
+### 差异表达 (Bulk RNA-seq)
+bioSkills:
+  differential-expression/ — deseq2-basics, edger-basics, batch-correction, de-results, de-visualization, timeseries-de
+  rna-quantification/ — alignment-free-quant (Salmon/kallisto), featurecounts-counting, tximport-workflow, count-matrix-qc
+  expression-matrix/ — counts-ingest, gene-id-mapping, metadata-joins, sparse-handling
+ClawBio:
+  rnaseq-de — 包含 QC、标准化和可视化的完整差异表达 (DE) 流水线
+  diff-visualizer — 针对 DE 结果的丰富可视化与报告生成
+
+### 单细胞 RNA-seq
+bioSkills:
+  single-cell/ — preprocessing, clustering, batch-integration, cell-annotation, cell-communication, doublet-detection, markers-annotation, trajectory-inference, multimodal-integration, perturb-seq, scatac-analysis, lineage-tracing, metabolite-communication, data-io
+ClawBio:
+  scrna-orchestrator — 完整的 Scanpy 流水线（QC、聚类、标记物、注释）
+  scrna-embedding — 基于 scVI 的潜在嵌入与批次整合
+
+### 空间转录组学
+bioSkills:
+  spatial-transcriptomics/ — spatial-data-io, spatial-preprocessing, spatial-domains, spatial-deconvolution, spatial-communication, spatial-neighbors, spatial-statistics, spatial-visualization, spatial-multiomics, spatial-proteomics, image-analysis
+
+### 表观基因组学
+bioSkills:
+  chip-seq/ — peak-calling, differential-binding, motif-analysis, peak-annotation, chipseq-qc, chipseq-visualization, super-enhancers
+  atac-seq/ — atac-peak-calling, atac-qc, differential-accessibility, footprinting, motif-deviation, nucleosome-positioning
+  methylation-analysis/ — bismark-alignment, methylation-calling, dmr-detection, methylkit-analysis
+  hi-c-analysis/ — hic-data-io, tad-detection, loop-calling, compartment-analysis, contact-pairs, matrix-operations, hic-visualization, hic-differential
+ClawBio:
+  methylation-clock — 表观遗传学年龄估计
+
+### 药物基因组学与临床
+bioSkills:
+  clinical-databases/ — clinvar-lookup, gnomad-frequencies, dbsnp-queries, pharmacogenomics, polygenic-risk, hla-typing, variant-prioritization, somatic-signatures, tumor-mutational-burden, myvariant-queries
+ClawBio:
+  pharmgx-reporter — 基于 23andMe/AncestryDNA 的药物基因组学 (PGx) 报告（包含 12 个基因、31 个 SNP、51 种药物）
+  drug-photo — 药物照片 → 个性化 PGx 剂量卡（通过视觉识别）
+  clinpgx — 用于基因-药物数据和 CPIC 指南的 ClinPGx API
+  gwas-lookup — 跨 9 个基因组数据库的联邦变体查找
+  gwas-prs — 基于消费者基因数据的多基因风险评分
+  nutrigx_advisor — 基于消费者基因数据的个性化营养建议
+
+### 群体遗传学与 GWAS
+bioSkills:
+  population-genetics/ — association-testing (PLINK GWAS), plink-basics, population-structure, linkage-disequilibrium, scikit-allel-analysis, selection-statistics
+  causal-genomics/ — mendelian-randomization, fine-mapping, colocalization-analysis, mediation-analysis, pleiotropy-detection
+  phasing-imputation/ — haplotype-phasing, genotype-imputation, imputation-qc, reference-panels
+ClawBio:
+  claw-ancestry-pca — 针对 SGDP 参考面板的祖源 PCA 分析
+
+### 宏基因组学与微生物组
+bioSkills:
+  metagenomics/ — kraken-classification, metaphlan-profiling, abundance-estimation, functional-profiling, amr-detection, strain-tracking, metagenome-visualization
+  microbiome/ — amplicon-processing, diversity-analysis, differential-abundance, taxonomy-assignment, functional-prediction, qiime2-workflow
+ClawBio:
+  claw-metagenomics — 鸟枪法宏基因组分析（物种分类、耐药基因、功能通路）
+
+### 基因组组装与注释
+bioSkills:
+  genome-assembly/ — hifi-assembly, long-read-assembly, short-read-assembly, metagenome-assembly, assembly-polishing, assembly-qc, scaffolding, contamination-detection
+  genome-annotation/ — eukaryotic-gene-prediction, prokaryotic-annotation, functional-annotation, ncrna-annotation, repeat-annotation, annotation-transfer
+  long-read-sequencing/ — basecalling, long-read-alignment, long-read-qc, clair3-variants, structural-variants, medaka-polishing, nanopore-methylation, isoseq-analysis
+
+### 结构生物学与化学信息学
+bioSkills:
+  structural-biology/ — alphafold-predictions, modern-structure-prediction, structure-io, structure-navigation, structure-modification, geometric-analysis
+  chemoinformatics/ — molecular-io, molecular-descriptors, similarity-searching, substructure-search, virtual-screening, admet-prediction, reaction-enumeration
+ClawBio:
+  struct-predictor — 带有比较功能的本地 AlphaFold/Boltz/Chai 结构预测
+
+### 蛋白质组学
+bioSkills:
+  proteomics/ — data-import, peptide-identification, protein-inference, quantification, differential-abundance, dia-analysis, ptm-analysis, proteomics-qc, spectral-libraries
+ClawBio:
+  proteomics-de — 蛋白质组学差异表达
+
+### 通路分析与基因网络
+bioSkills:
+  pathway-analysis/ — go-enrichment, gsea, kegg-pathways, reactome-pathways, wikipathways, enrichment-visualization
+  gene-regulatory-networks/ — scenic-regulons, coexpression-networks, differential-networks, multiomics-grn, perturbation-simulation
+
+### 免疫信息学
+bioSkills:
+  immunoinformatics/ — mhc-binding-prediction, epitope-prediction, neoantigen-prediction, immunogenicity-scoring, tcr-epitope-binding
+  tcr-bcr-analysis/ — mixcr-analysis, scirpy-analysis, immcantation-analysis, repertoire-visualization, vdjtools-analysis
+
+### CRISPR 与基因组工程
+bioSkills:
+  crispr-screens/ — mageck-analysis, jacks-analysis, hit-calling, screen-qc, library-design, crispresso-editing, base-editing-analysis, batch-correction
+  genome-engineering/ — grna-design, off-target-prediction, hdr-template-design, base-editing-design, prime-editing-design
+
+### 工作流管理
+bioSkills:
+  workflow-management/ — snakemake-workflows, nextflow-pipelines, cwl-workflows, wdl-workflows
+ClawBio:
+  repro-enforcer — 将任何分析导出为可复现包（包含 Conda 环境 + Singularity + 校验和）
+  galaxy-bridge — 从 usegalaxy.org 访问 8,000 多种 Galaxy 工具
+
+### 专业领域
+bioSkills:
+  alternative-splicing/ — splicing-quantification, differential-splicing, isoform-switching, sashimi-plots, single-cell-splicing, splicing-qc
+  ecological-genomics/ — edna-metabarcoding, landscape-genomics, conservation-genetics, biodiversity-metrics, community-ecology, species-delimitation
+  epidemiological-genomics/ — pathogen-typing, variant-surveillance, phylodynamics, transmission-inference, amr-surveillance
+  liquid-biopsy/ — cfdna-preprocessing, ctdna-mutation-detection, fragment-analysis, tumor-fraction-estimation, methylation-based-detection, longitudinal-monitoring
+  epitranscriptomics/ — m6a-peak-calling, m6a-differential, m6anet-analysis, merip-preprocessing, modification-visualization
+  metabolomics/ — xcms-preprocessing, metabolite-annotation, normalization-qc, statistical-analysis, pathway-mapping, lipidomics, targeted-analysis, msdial-preprocessing
+  flow-cytometry/ — fcs-handling, gating-analysis, compensation-transformation, clustering-phenotyping, differential-analysis, cytometry-qc, doublet-detection, bead-normalization
+  systems-biology/ — flux-balance-analysis, metabolic-reconstruction, gene-essentiality, context-specific-models, model-curation
+  rna-structure/ — secondary-structure-prediction, ncrna-search, structure-probing
+
+### 数据可视化与报告
+bioSkills:
+  data-visualization/ — ggplot2-fundamentals, heatmaps-clustering, volcano-customization, circos-plots, genome-browser-tracks, interactive-visualization, multipanel-figures, network-visualization, upset-plots, color-palettes, specialized-omics-plots, genome-tracks
+  reporting/ — rmarkdown-reports, quarto-reports, jupyter-reports, automated-qc-reports, figure-export
+ClawBio:
+  profile-report — 分析概况报告
+  data-extractor — 从科学图像中提取数值数据（通过视觉识别）
+  lit-synthesizer — 搜索 PubMed/bioRxiv，生成摘要及引文图谱
+  pubmed-summariser — 基于基因/疾病的 PubMed 检索及结构化简报
+
+### 数据库访问
+bioSkills:
+  database-access/ — entrez-search, entrez-fetch, entrez-link, blast-searches, local-blast, sra-data, geo-data, uniprot-access, batch-downloads, interaction-databases, sequence-similarity
+ClawBio:
+  ukb-navigator — 跨越 12,000 多个 UK Biobank 字段的语义搜索
+  clinical-trial-finder — 临床试验发现工具
+
+### 实验设计
+bioSkills:
+  experimental-design/ — power-analysis, sample-size, batch-design, multiple-testing
+
+### 组学机器学习
+bioSkills:
+  machine-learning/ — omics-classifiers, biomarker-discovery, survival-analysis, model-validation, prediction-explanation, atlas-mapping
+ClawBio:
+  claw-semantic-sim — 疾病文献的语义相似度索引 (PubMedBERT)
+  omics-target-evidence-mapper — 聚合跨组学来源的靶点级证据
+
+## 环境设置
+
+这些技能假设您在使用一台生物信息学工作站。常见依赖项包括：
+
+```bash
+# Python
+pip install biopython pysam cyvcf2 pybedtools pyBigWig scikit-allel anndata scanpy mygene
+
+# R/Bioconductor
+Rscript -e 'BiocManager::install(c("DESeq2","edgeR","Seurat","clusterProfiler","methylKit"))'
+
+# CLI tools (Ubuntu/Debian)
+sudo apt install samtools bcftools ncbi-blast+ minimap2 bedtools
+
+# CLI tools (macOS)
+brew install samtools bcftools blast minimap2 bedtools
+
+# Or via Conda (recommended for reproducibility)
+conda install -c bioconda samtools bcftools blast minimap2 bedtools fastp kraken2
+```
+
+## 注意事项
+
+- 获取的技能**不是** Hermes 的 SKILL.md 格式。它们使用自己的结构（bioSkills：代码模式指南；ClawBio：README + Python 脚本）。请将它们作为专家参考资料阅读。
+- bioSkills 是参考指南 —— 它们展示了正确的参数和代码模式，但本身并不是可执行的流水线。
+- ClawBio 技能是可执行的 —— 许多带有 `--demo` 标志，可以直接运行。
+- 这两个仓库都假定您的系统已安装了相关的生物信息学工具。在运行流水线之前，请检查先决条件。
+- 对于 ClawBio，请首先在克隆的仓库中运行 `pip install -r requirements.txt`。
+- 基因组数据文件可能非常大。在下载参考基因组、SRA 数据集或构建索引时，请注意磁盘空间。

package/extensions/skills/optional/research/darwinian-evolver/SKILL.md

@@ -0,0 +1,161 @@
+---
+name: darwinian-evolver
+description: 使用 Imbue 的 evolution loop 进化 prompt/regex/SQL/code。
+version: 0.1.0
+author: Bihruze (Asahi0x), Hermes Agent
+license: MIT
+platforms: [linux, macos]
+metadata:
+  hermes:
+    tags: [evolution, optimization, prompt-engineering, research]
+    related_skills: [arxiv, jupyter-live-kernel]
+---
+
+# Darwinian Evolver
+
+运行 Imbue 的 [darwinian_evolver](https://github.com/imbue-ai/darwinian_evolver) —— 一个由 LLM 驱动的进化搜索循环 —— 来针对特定的 fitness function 优化 **prompt、regex、SQL 查询或小段代码**。
+
+状态：这是对上游工具的轻量级封装。该 skill 会自动安装它，引导 agent 编写 `Problem` 定义（organism + evaluator + mutator），并通过上游 CLI 或一个小型的自定义 Python 驱动程序来运行循环。
+
+**许可：** 上游工具基于 **AGPL-3.0** 协议。该 skill 仅通过上游 CLI 或 `subprocess`/`uv run` 调用来执行它（单纯的聚合）。**切勿**将上游的类直接导入到 Hermes 本身中。
+
+## 何时使用
+
+- 用户说“优化这个 prompt”、“进化一个用于 X 的 regex”、“自动改进这段代码/SQL”、“搜索更好的指令”。
+- 你有一个打分器（精确匹配、regex 通过率、单元测试、LLM 聊天机器人裁判、运行时指标），并且有一个初始候选对象。如果你没有打分器，请停下来先定义一个——这是最困难的部分。
+- 成本可接受：一次典型的运行需要进行 50–500 次 LLM 调用。在 gpt-4o-mini 上只需几分之一美元；在 Claude Sonnet 上可能需要几美元。
+
+在以下情况**不要**使用：
+- 优化目标是可微分的（使用 gradient descent / DSPy）。
+- 你只需要尝试 2–3 个变体——直接手写即可。
+- fitness 信号纯粹是主观的，没有任何可衡量的标准。
+
+## 前置条件
+
+- Python ≥3.11
+- `git`, `uv` (或 `pip`)
+- 以下任意一个：`OPENROUTER_API_KEY`, `ANTHROPIC_API_KEY`, 或 `OPENAI_API_KEY`
+
+该 skill 附带了一个小型的 `parrot_openrouter.py` 驱动程序，它通过 OpenAI SDK 使用 `OPENROUTER_API_KEY`，因此 OpenRouter 上的任何模型都可以使用。上游 CLI 本身硬编码了 Anthropic，并需要 `ANTHROPIC_API_KEY`。
+
+## 安装 (One-Time)
+
+通过 `terminal` 工具运行：
+
+```bash
+mkdir -p ~/.hermes/cache/darwinian-evolver && cd ~/.hermes/cache/darwinian-evolver
+[ -d darwinian_evolver ] || git clone --depth 1 https://github.com/imbue-ai/darwinian_evolver.git
+cd darwinian_evolver && uv sync
+```
+
+验证：
+
+```bash
+cd ~/.hermes/cache/darwinian-evolver/darwinian_evolver \
+  && uv run darwinian_evolver --help | head -5
+```
+
+## 快速开始 — 内置 Parrot 示例
+
+小型冒烟测试（需要 `ANTHROPIC_API_KEY`）：
+
+```bash
+cd ~/.hermes/cache/darwinian-evolver/darwinian_evolver
+uv run darwinian_evolver parrot \
+  --num_iterations 2 \
+  --num_parents_per_iteration 2 \
+  --mutator_concurrency 2 --evaluator_concurrency 2 \
+  --output_dir /tmp/parrot_demo
+```
+
+输出内容：
+- `/tmp/parrot_demo/snapshots/iteration_N.pkl` —— 每次迭代 pickle 后的种群
+- `/tmp/parrot_demo/<jsonl>` —— 每次迭代的 JSON 日志（路径在结束时打印）
+
+在浏览器中打开 `~/.hermes/cache/darwinian-evolver/darwinian_evolver/darwinian_evolver/lineage_visualizer.html` 并加载 JSON 日志以查看进化树。
+
+## 快速开始 — OpenRouter Driver (无 Anthropic Key)
+
+该 skill 附带了 `scripts/parrot_openrouter.py` —— 相同的 parrot 问题，但 LLM 调用是通过 OpenRouter 进行的，因此任何 provider 都可以工作。
+
+```bash
+# 从 skill 安装的任何位置开始：
+SKILL_DIR=~/.hermes/skills/research/darwinian-evolver
+DE_DIR=~/.hermes/cache/darwinian-evolver/darwinian_evolver
+
+cd "$DE_DIR" && \
+  EVOLVER_MODEL='openai/gpt-4o-mini' \
+  uv run --with openai python "$SKILL_DIR/scripts/parrot_openrouter.py" \
+    --num_iterations 3 --num_parents_per_iteration 2 \
+    --output_dir /tmp/parrot_or
+```
+
+使用 `scripts/show_snapshot.py` 检查结果：
+
+```bash
+uv run --with openai python "$SKILL_DIR/scripts/show_snapshot.py" \
+  /tmp/parrot_or/snapshots/iteration_3.pkl
+```
+
+预期输出：按分数排名的 7 个进化后的 prompt 模板，其中最好的得分在 0.6–0.8 左右（初始的种子 `Say {{ phrase }}` 得分为 0.000）。
+
+## 定义自定义 Problem
+
+该 skill 附带了 `templates/custom_problem_template.py` —— 复制、编辑、运行。
+你必须定义三样东西：
+
+1. **`Organism`** —— 一个 Pydantic `BaseModel` 的子类，包含正在进化的 artifact（`prompt_template: str`, `regex_pattern: str`, `sql_query: str`, `code_block: str` 等）。添加一个 `run(*args)` 方法来运行它。
+
+2. **`Evaluator`** —— `.evaluate(organism) -> EvaluationResult(score=..., trainable_failure_cases=[...], holdout_failure_cases=[...], is_viable=True)`。
+   - **`score`** 在 `[0, 1]` 之间。越高越好。
+   - **`trainable_failure_cases`** —— mutator 所看到的内容。包含足够的上下文（输入、预期、实际），以便 LLM 进行诊断。
+   - **`holdout_failure_cases`** —— 对 mutator 隐藏的内容。使用这些来检测过拟合。
+   - **`is_viable=True`** 除非 organism 完全损坏（抛出异常、返回 None 等）。得分为 0 的可用 organism 是可以的——它只会在父代选择中被降低权重。
+
+3. **`Mutator`** —— `.mutate(organism, failure_cases, learning_log_entries) -> list[Organism]`。
+   通常：构建一个包含当前 organism + 失败案例 + 提出修复请求的 LLM prompt；解析 LLM 的响应；返回一个新的 `Organism`。如果解析失败则返回 `[]` —— 循环会处理它。
+
+然后编写一个驱动脚本，将 `Problem(initial_organism, evaluator, [mutators])` 接入到 `EvolveProblemLoop` 中，并遍历 `loop.run(num_iterations=N)` —— 附带的 `scripts/parrot_openrouter.py` 是参考示例。
+
+## 真正重要的 Hyperparameters
+
+| flag | default | when to change |
+| --- | --- | --- |
+| `--num_iterations` | 5 | 当你信任 evaluator 时增加到 10–20 |
+| `--num_parents_per_iteration` | 4 | 降至 2 以进行低成本的探索 |
+| `--mutator_concurrency` | 10 | 降至 2–4 以避免触发 rate limits |
+| `--evaluator_concurrency` | 10 | 同上；evaluator 也会调用 LLM |
+| `--batch_size` | 1 | 当你的 mutator 能够处理多个失败案例时增加到 3–5 |
+| `--verify_mutations` | off | 当 mutator 产生浪费时开启（根据 Imbue，后续运行可节省 >10 倍成本） |
+| `--midpoint_score` | `p75` | 除非分数聚集，否则不要动 |
+| `--sharpness` | 10 | 不要动 |
+
+## Pitfalls
+
+1. **`Initial organism must be viable`** —— 即使是得分为 0 的种子，也要在你的 `EvaluationResult` 中设置 `is_viable=True`。循环拒绝不可用的 organism，因为这意味着循环没有进化的基础。
+2. **Provider 的内容过滤器会导致运行失败。** Azure 支持的 OpenRouter 模型会拒绝类似 "ignore previous instructions" 的短语并返回 HTTP 400 错误。将 LLM 调用包装在 `try/except` 中，并返回 `f"<LLM_ERROR: {e}>"` —— evolver 只会将该 organism 的分数评为 0 并继续运行。
+3. **`loop.run()` 是一个 generator** —— 在你遍历它之前，调用它不会运行任何东西。使用 `for snap in loop.run(num_iterations=N):`。
+4. **Snapshots 是嵌套的 pickle。** `iteration_N.pkl` 包含一个带有 `population_snapshot`（更多 pickle 字节）的字典。要 unpickle 它，你必须能够在与它被 pickle 时相同的点分路径下导入 `Organism` 类。
+5. **并发默认设置非常激进。** 在大多数 provider 上，10/10 会触发 rate limits。请从 2/2 开始。
+6. **CLI 硬编码为 Anthropic。** `uv run darwinian_evolver <problem>` 会尝试获取 `ANTHROPIC_API_KEY` 并使用 Claude Sonnet。要使用任何其他 provider，请编写一个类似 `parrot_openrouter.py` 的驱动程序。
+7. **AGPL。** 切勿在 Hermes 核心中使用 `from darwinian_evolver import ...`。位于 `~/.hermes/skills/...` 下的自定义驱动脚本是属于用户端的，没有问题。
+8. **无 PyPI 包。** `pip install darwinian-evolver` 会拉取到错误的东西。请始终从 GitHub repo 安装。
+
+## 验证
+
+安装并运行 parrot 之后，以下命令的退出码为 0 即可证明成功：
+
+```bash
+DE_DIR=~/.hermes/cache/darwinian-evolver/darwinian_evolver
+ls "$DE_DIR/darwinian_evolver/lineage_visualizer.html" >/dev/null && \
+cd "$DE_DIR" && uv run darwinian_evolver --help >/dev/null && \
+echo "darwinian-evolver: OK"
+```
+
+## References
+
+- [Imbue research post](https://imbue.com/research/2026-02-27-darwinian-evolver/)
+- [ARC-AGI-2 results](https://imbue.com/research/2026-02-27-arc-agi-2-evolution/)
+- [imbue-ai/darwinian_evolver](https://github.com/imbue-ai/darwinian_evolver) (AGPL-3.0)
+- [Darwin Gödel Machines](https://arxiv.org/abs/2505.22954)
+- [PromptBreeder](https://arxiv.org/abs/2309.16797)

package/extensions/skills/optional/research/darwinian-evolver/scripts/parrot_openrouter.py

@@ -0,0 +1,218 @@
+"""
+parrot_openrouter: same as the upstream `parrot` example but the LLM call goes
+through OpenRouter (OpenAI SDK) instead of Anthropic native. Lets us run an
+end-to-end evolution with whatever model the user already has paid access to.
+
+Run with:
+    uv --project darwinian_evolver run python parrot_openrouter.py \
+        --num_iterations 3 --output_dir /tmp/parrot_out
+
+Reads `OPENROUTER_API_KEY` from the environment.
+"""
+from __future__ import annotations
+
+import argparse
+import os
+import sys
+from pathlib import Path
+
+import jinja2
+from openai import OpenAI
+
+# 来自上游的内置问题类型 (AGPL — 在生产环境中仅通过子进程运行)
+from darwinian_evolver.cli_common import build_hyperparameter_config_from_args
+from darwinian_evolver.cli_common import register_hyperparameter_args
+from darwinian_evolver.cli_common import parse_learning_log_view_type
+from darwinian_evolver.evolve_problem_loop import EvolveProblemLoop
+from darwinian_evolver.learning_log import LearningLogEntry
+from darwinian_evolver.problem import EvaluationFailureCase
+from darwinian_evolver.problem import EvaluationResult
+from darwinian_evolver.problem import Evaluator
+from darwinian_evolver.problem import Mutator
+from darwinian_evolver.problem import Organism
+from darwinian_evolver.problem import Problem
+
+DEFAULT_MODEL = os.environ.get("EVOLVER_MODEL", "openai/gpt-4o-mini")
+
+
+def _client() -> OpenAI:
+    key = os.environ.get("OPENROUTER_API_KEY")
+    if not key:
+        sys.exit("OPENROUTER_API_KEY is not set")
+    return OpenAI(api_key=key, base_url="https://openrouter.ai/api/v1")
+
+
+def _prompt_llm(prompt: str) -> str:
+    try:
+        r = _client().chat.completions.create(
+            model=DEFAULT_MODEL,
+            max_tokens=1024,
+            messages=[{"role": "user", "content": prompt}],
+        )
+        return r.choices[0].message.content or ""
+    except Exception as e:
+        # 将任何提供者错误（速率限制、内容过滤、架构拒绝）
+        # 视为失败的响应。进化器只需将其视为该生物体的低分
+        # 并继续执行 —— 这比直接终止运行要友好得多。
+        return f"<LLM_ERROR: {type(e).__name__}: {e}>"
+
+
+class ParrotOrganism(Organism):
+    prompt_template: str
+
+    def run(self, phrase: str) -> str:
+        try:
+            prompt = jinja2.Template(self.prompt_template).render(phrase=phrase)
+        except jinja2.exceptions.TemplateError as e:
+            return f"Error rendering prompt: {e}"
+        if not prompt:
+            return ""
+        return _prompt_llm(prompt)
+
+
+class ParrotEvaluationFailureCase(EvaluationFailureCase):
+    phrase: str
+    response: str
+
+
+class ImproveParrotMutator(Mutator[ParrotOrganism, ParrotEvaluationFailureCase]):
+    IMPROVEMENT_PROMPT_TEMPLATE = """
+We want to build a prompt that causes an LLM to repeat back a given phrase verbatim.
+
+The current prompt template is:
+```
+{{ organism.prompt_template }}
+```
+
+Unfortunately, on this phrase:
+```
+{{ failure_case.phrase }}
+```
+the LLM responded with:
+```
+{{ failure_case.response }}
+```
+
+Diagnose what went wrong, then propose an improved prompt template. Put the new
+template in the LAST triple-backtick block of your response.
+""".strip()
+
+    def mutate(
+        self,
+        organism: ParrotOrganism,
+        failure_cases: list[ParrotEvaluationFailureCase],
+        learning_log_entries: list[LearningLogEntry],
+    ) -> list[ParrotOrganism]:
+        fc = failure_cases[0]
+        prompt = jinja2.Template(self.IMPROVEMENT_PROMPT_TEMPLATE).render(
+            organism=organism, failure_case=fc
+        )
+        try:
+            resp = _prompt_llm(prompt)
+            parts = resp.split("```")
+            if len(parts) < 3:
+                return []
+            new_tpl = parts[-2].strip()
+            return [ParrotOrganism(prompt_template=new_tpl)]
+        except Exception as e:
+            print(f"mutate error: {e}", file=sys.stderr)
+            return []
+
+
+class ParrotEvaluator(Evaluator[ParrotOrganism, EvaluationResult, ParrotEvaluationFailureCase]):
+    TRAINABLE_PHRASES = [
+        "Hello world.",
+        "bla",
+        "Bla",
+        "bla.",
+        '"bla bla".',
+        "Just say 'foo' once with no extra words.",
+    ]
+    HOLDOUT_PHRASES = [
+        "bla, but only once.",
+        "'bla'",
+    ]
+
+    def evaluate(self, organism: ParrotOrganism) -> EvaluationResult:
+        train_fails: list[ParrotEvaluationFailureCase] = []
+        hold_fails: list[ParrotEvaluationFailureCase] = []
+        for i, p in enumerate(self.TRAINABLE_PHRASES):
+            r = organism.run(p)
+            if r != p:
+                train_fails.append(ParrotEvaluationFailureCase(
+                    phrase=p, response=r, data_point_id=f"trainable_{i}"))
+        for i, p in enumerate(self.HOLDOUT_PHRASES):
+            r = organism.run(p)
+            if r != p:
+                hold_fails.append(ParrotEvaluationFailureCase(
+                    phrase=p, response=r, data_point_id=f"holdout_{i}"))
+        n_total = len(self.TRAINABLE_PHRASES) + len(self.HOLDOUT_PHRASES)
+        n_ok = n_total - len(train_fails) - len(hold_fails)
+        return EvaluationResult(
+            score=n_ok / n_total,
+            trainable_failure_cases=train_fails,
+            holdout_failure_cases=hold_fails,
+            # 始终可行的。即使是 0 分的种子也是有效的起点；
+            # 变异器仍应有机会去修复它。
+            is_viable=True,
+        )
+
+
+def make_problem() -> Problem:
+    return Problem[ParrotOrganism, EvaluationResult, ParrotEvaluationFailureCase](
+        evaluator=ParrotEvaluator(),
+        mutators=[ImproveParrotMutator()],
+        initial_organism=ParrotOrganism(prompt_template="Say {{ phrase }}"),
+    )
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser()
+    register_hyperparameter_args(ap.add_argument_group("hyperparameters"))
+    ap.add_argument("--num_iterations", type=int, default=3)
+    ap.add_argument("--mutator_concurrency", type=int, default=4)
+    ap.add_argument("--evaluator_concurrency", type=int, default=4)
+    ap.add_argument("--output_dir", type=str, required=True)
+    args = ap.parse_args()
+
+    out = Path(args.output_dir)
+    out.mkdir(parents=True, exist_ok=True)
+
+    hp = build_hyperparameter_config_from_args(args)
+    loop = EvolveProblemLoop(
+        problem=make_problem(),
+        learning_log_view_type=parse_learning_log_view_type(hp.learning_log_view_type),
+        num_parents_per_iteration=hp.num_parents_per_iteration,
+        mutator_concurrency=args.mutator_concurrency,
+        evaluator_concurrency=args.evaluator_concurrency,
+        fixed_midpoint_score=hp.fixed_midpoint_score,
+        midpoint_score_percentile=hp.midpoint_score_percentile,
+        sharpness=hp.sharpness,
+        novelty_weight=hp.novelty_weight,
+        batch_size=hp.batch_size,
+        should_verify_mutations=hp.verify_mutations,
+    )
+
+    import json
+    log_path = out / "results.jsonl"
+    snap_dir = out / "snapshots"
+    snap_dir.mkdir(exist_ok=True)
+    print("Evaluating initial organism...")
+    for snap in loop.run(num_iterations=args.num_iterations):
+        (snap_dir / f"iteration_{snap.iteration}.pkl").write_bytes(snap.snapshot)
+        _, best_eval = snap.best_organism_result
+        print(f"iter={snap.iteration} pop={snap.population_size} "
+              f"best_score={best_eval.score:.3f}")
+        with log_path.open("a") as f:
+            f.write(json.dumps({
+                "iteration": snap.iteration,
+                "best_score": best_eval.score,
+                "pop_size": snap.population_size,
+                "score_percentiles": {str(k): v for k, v in snap.score_percentiles.items()},
+            }) + "\n")
+    print(f"\nDone. Results in: {out}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())