# GreaTruth AI Data Index by Alvance

> Alvance GreaTruth 后训练数据目录，覆盖可执行 RL 环境（RLVR）、真实与合成 SFT 轨迹、评测集与 Rubric，提供可运行样例、独立 Verifier、Schema 和下载归档。

## 产品定位

- 数据样例同时保留上下文、执行过程与验证结果，以支持复核、评估与训练。
- Agent 执行类数据包含工具调用、环境状态、错误恢复和结果验证等过程信息。
- 公开样例、下载件、Schema 与机器可读目录按数据集成组发布。

## Product lines

- PRODUCT 01 垂域数据 (Vertical Domain Data): https://display.greatruth.cloud/lines/vertical-domain/
  面向编程、咨询、产品经理、物流、电商、网安和通用办公场景的垂域任务样例，保留完整题面、输入材料、参考交付物、Rubric、评测记录和配套说明书。
- PRODUCT 02 RL强化训练数据 (RL Training Data): https://display.greatruth.cloud/lines/rl-sft/
  面向终端执行、软件工程、客户支持、专业办公与桌面操作的强化学习任务数据。样例按来源保留任务题面、环境或工作区、工具接口、Verifier 和评测记录，可用于核验训练任务及反馈结构。
- PRODUCT 03 STEM 数据 (STEM Data): https://display.greatruth.cloud/lines/stem/
  HLE / GPQA / FrontierMath 难度带的博士级数理评测与长文本研究题，交付题目、完整解答、Rubric，以及适用的模型作答和逐项评分证据。
- PRODUCT 04 真实轨迹 (Real Trajectories): https://display.greatruth.cloud/lines/real-trajectory/
  Claude Code、OpenClaw 与 Hermes 的脱敏生产会话，按 SFT 轨迹交付任务上下文、CoT、工具调用、子代理过程、环境响应和最终输出。
- PRODUCT 05 高质量轨迹数据 (High-quality Trajectory Data): https://display.greatruth.cloud/lines/high-quality-trajectory/
  面向 Agent 后训练的合成 SFT 轨迹，覆盖 MCP 多工具环境任务与 WorkBuddy 多智能体任务树；保留完整 System Prompt、CoT、工具结果、运行状态和最终工作区。

## Selection policy

- RL training data: TerminalBench 4, DeepSWE, AutomationBench, JobBench, Computer Use, ClawBench, and specialist task samples; evaluation records are interpreted within each task and revision.
- Real trajectories: publish the highest-data-quality candidate in each source collection.
- High-quality trajectories: publish synthetic task deliveries with system prompts, environment or workspace state, tool records, collaboration metrics, outputs, and explicit completion status.
- STEM data: publish diagnostic evaluation cases or complete verified question records, according to each source contract.
- Every dataset has its own source adapter, viewer, schema, and selection report.

## Discovery

- Catalog JSON: https://display.greatruth.cloud/data/catalog.json
- Agent manifest: https://display.greatruth.cloud/ai-data.json
- Selection schema: https://display.greatruth.cloud/data/schemas/sample-selection.schema.json
- Harbor samples: https://display.greatruth.cloud/data/harbor/package-meta.json
- Authentication: none for listed evaluation samples

## Datasets

### Claude Code Agent Trajectory

- ID: gr-agent-cc-001
- Category: real_trajectory
- Selection: lowest_tool_error_rate
- Selected sample: 62732ab4-0bc4-447a-8e1c-c99e58abbbd6
- User query: 先看下 src/api.mjs 里 set 函数的 for...in 循环，`if (!attributes[attributeName])` 那块会直接跳过值为 0、false 或空字符串的属性。再去 test/tests.js 里查一下有没有覆盖这些情况的用例。
- Records: 103 api_turns
- Viewer steps: 124
- Source events: 103
- Tool calls: 83
- Estimated token length: 123.8K
- Main trajectory length: 123.8K
- Token estimation: last-turn usage total
- Viewer: https://display.greatruth.cloud/claude-code/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-claude-code-high-quality.zip
- Schema: https://display.greatruth.cloud/data/schemas/claude-api-turn.schema.json
- SHA-256: dd1605c53ee4608a57663eb13a7b2c140c638137948651988c66d87855c0790d
- Selected raw trajectory sample: https://display.greatruth.cloud/downloads/greatruth-claude-code-high-quality.jsonl
- Selected raw trajectory sample SHA-256: 34f0594afd71391ddc48cf5184f3d826bba118b673c9d0a39d343dd822219d0a
- Normalized Claude Code API-turn trace: https://display.greatruth.cloud/downloads/greatruth-claude-code-normalized.json
- Normalized Claude Code API-turn trace SHA-256: 88f5363a7a723284d3a55de90ddebd5dd6ec87923380fd2ebf0fd907b685163d
- Sample selection report: https://display.greatruth.cloud/data/selections/claude-code-selection.json
- Sample selection report SHA-256: dc6af2bd26ca81c83bbb4a917a7daf945e1b8be8a57ddcfe3b0d061f1d0f1b36

### Claude Code 40-turn Agent Trajectory

- ID: gr-agent-cc-002
- Category: real_trajectory
- Selection: lowest_tool_error_rate
- Selected sample: 64e58ff6-14fa-44bd-a0ff-02bce1e07c11
- User query: 先用 nox 跑一下现有测试，针对 Python 3.13 的 tests session，确认全部通过。如果跑不起来就告诉我。
- Records: 131 api_turns
- Viewer steps: 174
- Source events: 131
- Tool calls: 108
- Estimated token length: 218.2K
- Main trajectory length: 218.2K
- Token estimation: last-turn usage total
- Viewer: https://display.greatruth.cloud/claude-code-40turn/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-claude-code-40turn-high-quality.zip
- Schema: https://display.greatruth.cloud/data/schemas/claude-api-turn.schema.json
- SHA-256: 1d7f23f4e91fcc8a14618fdbf7f2093f418fa8d593cf55bb42521d5bc6538a51
- Selected raw trajectory sample: https://display.greatruth.cloud/downloads/greatruth-claude-code-40turn-high-quality.jsonl
- Selected raw trajectory sample SHA-256: f1c51b44d868963081f796827ea51a34efd4ed603ff8a511c53edecb37c6a4e1
- Normalized Claude Code API-turn trace: https://display.greatruth.cloud/downloads/greatruth-claude-code-40turn-normalized.json
- Normalized Claude Code API-turn trace SHA-256: 48f9b4e63f110522a2f3b0bd5fe1c769b3bc3928c9a200b3415c76788516ba6c
- Sample selection report: https://display.greatruth.cloud/data/selections/claude-code-40turn-selection.json
- Sample selection report SHA-256: 17bd51994661d8f85cd267081aadc3e476310b3a44c64d91aad60d710c2fc6c0

### OpenClaw Multi-agent Trajectory

- ID: gr-agent-openclaw-001
- Category: real_trajectory
- Selection: highest_data_quality
- Selected sample: 000125
- User query: 我刚收到上游团队的反馈草案，指南主稿、快速开始和审核清单边界确实没拉开，快速开始里连最低 Go 版本都没写。你先完整读一遍 input_materials/origin/ 下的所有材料（material_context.json、material_manifest.json 以及列出的 origin 文件），再基于材料里的 known_gaps 和 key_entities 做以下事：用表格加注释的方式，清晰界定三部分各自覆盖什么、重叠什么、缺口在哪；针对 quickstart 缺的 Go 版本要求（Go 1.22+？），如果材料里没明确写，可以用 bocha_web_search 快速搜一下当前主流团队建议的最低版本并记录来源 URL。先把分析写到 deliverables/boundary_analysis.md，要求包含 key_entities（go test -count=1 ./ 等命令要不要补到 quickstart 里）和每个缺口的处置建议。先在内部走一遍推理、对照材料约束再落盘，输出可复查。
- Records: 128 events
- Viewer steps: 60
- Source events: 128
- Tool calls: 52
- Estimated token length: 86.5K
- Main trajectory length: 74.0K
- Subtrajectory length: 12.4K
- Token estimation: last-turn usage total partitioned by normalized main/subtrajectory content length
- Viewer: https://display.greatruth.cloud/openclaw/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-openclaw-high-quality.zip
- Schema: https://display.greatruth.cloud/data/schemas/openclaw-event.schema.json
- SHA-256: 705be286e93c9e6255bec419c24be2a6995d693df71fee593c274d626ffa64a8
- Selected raw trajectory sample: https://display.greatruth.cloud/downloads/greatruth-openclaw-high-quality.jsonl
- Selected raw trajectory sample SHA-256: a5d5b7f42586af43a200f9e8e3e9eb6b1218066b41ecfa53a41c240eac6e97a4
- Normalized OpenClaw multi-agent event trace: https://display.greatruth.cloud/downloads/greatruth-openclaw-normalized.json
- Normalized OpenClaw multi-agent event trace SHA-256: 612a2162e835d7d3478637593188913c77d6e550ffc4be913aef08bfc4761eea
- Sample selection report: https://display.greatruth.cloud/data/selections/openclaw-selection.json
- Sample selection report SHA-256: 85fb7c1dfb53347e679cfacf49d92b16e781687ad79d2af96d02417ae1638d32

### Hermes Function Calling Trajectories

- ID: gr-agent-hermes-001
- Category: real_trajectory
- Selection: highest_data_quality
- Selected sample: 00002
- User query: For same-day handoff, I need you to close the remaining source-validation risk rather than adding more caveat files: in `/data/tasks/00002/output`, make one conservative attempt to replace or clearly quarantine the illustrative series by using saved public evidence from a mix of official, industry, and open-data domains, prioritizing the official FRED/ALFRED path via bocha-search and saving raw evidence; if a lightweight helper such as a small Python package or CLI fetch/parsing tool is useful, bootstrap it in the workspace and verify it with a minimal command. Please produce a compact table-first source-validation report with evidence notes, a quantitative checkpoint comparing any retrieved official/open-data observations against the prepared CSV, and a short narrative recommendation on whether the existing charts can stand as internal prototype only or should be regenerated; keep all outputs under the same directory, update only the files needed to make the conclusion traceable, and read back the key report plus any changed chart/data/manifest files before you claim it is ready.
- Records: 11 trajectories
- Viewer steps: 120
- Source events: 193
- Tool calls: 105
- Estimated token length: 202.9K
- Main trajectory length: 131.9K
- Subtrajectory length: 71.1K
- Token estimation: CJK characters + non-CJK characters / 4
- Viewer: https://display.greatruth.cloud/hermes/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-hermes-trajectory-samples.zip
- Schema: https://display.greatruth.cloud/data/schemas/hermes-trajectory.schema.json
- SHA-256: 73b8c88d695b8b9cf828aa23c1b5ac5a5d73d251698dddb77f3f4198794f0579
- Selected raw trajectory sample: https://display.greatruth.cloud/downloads/greatruth-hermes-trajectory-samples.jsonl
- Selected raw trajectory sample SHA-256: 4eeed4c7b122125cad739a80bafcd3a52db9ff849a06dc990ad7da7c5130f841
- Normalized Hermes function-calling trace: https://display.greatruth.cloud/downloads/greatruth-hermes-trajectory-normalized.json
- Normalized Hermes function-calling trace SHA-256: 08d0c4f089fa6d794300f7c3118051bfadeea04da2a0ab154f0a94b54ade1e15
- Sample selection report: https://display.greatruth.cloud/data/selections/hermes-selection.json
- Sample selection report SHA-256: d7f4bdedb35fc761bba7908db69261ee8ae3671c5d2d133976972ba77923bfdb
- Hermes skill evolution events: https://display.greatruth.cloud/downloads/greatruth-hermes-skill-evolution-samples.jsonl
- Hermes skill evolution events SHA-256: 8f519dea487f36bffabca124d02b463b89ee1e968bdf5f183b6358dec49309e2

### PhD Mathematics Evaluation

- ID: gr-eval-math-phd-001
- Category: stem_data
- Selection: lowest_ai_score
- Selected sample: mp01-003-doubao-1
- Records: 1 items
- Viewer: https://display.greatruth.cloud/math-phd/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-math-low-score-evaluation.zip
- Schema: https://display.greatruth.cloud/data/schemas/qa-eval.schema.json
- SHA-256: 6da6c7630ea0c073f7e08b121115c531cdbe733f886164f6f6bea5885ecb954f
- Evaluation sample JSON: https://display.greatruth.cloud/downloads/greatruth-math-low-score-evaluation.json
- Evaluation sample JSON SHA-256: 053af847c3420f090476a79deb729a0c8b4c472af0f464132bd5a6d09e9d8d8d
- Evaluation sample selection report: https://display.greatruth.cloud/data/selections/math-selection.json
- Evaluation sample selection report SHA-256: 48e8cfb751c0af6224202fdc79434ecf48be02a1f6ff9bd7ec2ca8acfaa5a42f

### Long-context Research Problems

- ID: gr-stem-math-mp02-001
- Category: stem_data
- Records: 3 items
- Viewer: https://display.greatruth.cloud/math-mp02/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-math-mp02-verified.zip
- Schema: https://display.greatruth.cloud/data/schemas/qa-verified.schema.json
- SHA-256: b725494fdb557592a8bea695b34b9e4708e0bce89bdbae4863f36e998e40869f
- Verified mathematics question records: https://display.greatruth.cloud/downloads/greatruth-math-mp02-verified.jsonl
- Verified mathematics question records SHA-256: 526f0ce5feb464cdb310ae8bc4c724c8ebb25883db7cd98fb792373febace095

### PhD Physics Evaluation

- ID: gr-eval-physics-phd-001
- Category: stem_data
- Selection: lowest_ai_score
- Selected sample: pp01-007-qwen3-5-plus
- Records: 1 items
- Viewer: https://display.greatruth.cloud/physics-phd/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-physics-low-score-evaluation.zip
- Schema: https://display.greatruth.cloud/data/schemas/qa-eval.schema.json
- SHA-256: 43a2c4cfe86bfb28813ea1b0da6023043daf8293bc90a554ae4e02d35c54e3de
- Evaluation sample JSON: https://display.greatruth.cloud/downloads/greatruth-physics-low-score-evaluation.json
- Evaluation sample JSON SHA-256: c58e77ee295bcb83fd46a97ebc6dacd88c88d571b1d7ec51f3b38fd4a3b341cd
- Evaluation sample selection report: https://display.greatruth.cloud/data/selections/physics-selection.json
- Evaluation sample selection report SHA-256: 4ba58d947633952cb3d6d1c33b82d6354f459d4796426b2d0a031a7273b521c7

### TerminalBench 4 · Terminal Tasks

- ID: gr-env-terminalbench4-001
- Category: rl_sft_environment
- Records: 10 packages
- Viewer: https://display.greatruth.cloud/terminalbench4/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-terminalbench4-samples.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 8f65f8d6c1c3c4179f492e7f32a4859a5bcd89553a7e4e751d6fbb4c12b38fdc
- TerminalBench 4 structured sample: https://display.greatruth.cloud/data/harbor/terminalbench4-sample.json
- TerminalBench 4 structured sample SHA-256: c450ade801fa500acab456eb2cee5a5ee7cfd231a2d1c7f154e8218fe1cbc36c
- artifact-provenance-reconciler-v2: https://display.greatruth.cloud/downloads/greatruth-tb4-artifact-provenance-reconciler-v2.zip
- artifact-provenance-reconciler-v2 SHA-256: f8c163f95d8f52a76ce150ee1b21745b2a54dff8a154583d0f487cd8ade7ed4d
- cold-chain-custody-audit: https://display.greatruth.cloud/downloads/greatruth-tb4-cold-chain-custody-audit.zip
- cold-chain-custody-audit SHA-256: 25113e28d47e30cfb02bc39a6c6b16638bd67e7d12b699dbf33820bd1086fc43
- diagnostic-baseline-release-gate: https://display.greatruth.cloud/downloads/greatruth-tb4-diagnostic-baseline-release-gate.zip
- diagnostic-baseline-release-gate SHA-256: a6b1ac5218fe0fad29eebececa8278b7b67bfdbc0d7801db4ad230f52f93b0b5
- metric-aggregator: https://display.greatruth.cloud/downloads/greatruth-tb4-metric-aggregator.zip
- metric-aggregator SHA-256: 424448501c64a9b919091106bc9d61cf482aee412d3a204eec9041bc3fb58d2d
- order-fulfillment-engine: https://display.greatruth.cloud/downloads/greatruth-tb4-order-fulfillment-engine.zip
- order-fulfillment-engine SHA-256: 464a05abd06a5d9d615b714d462655822edefaf8702dd158c3fa5634e7a522a9
- qc-astra-pass4-cmaes-r01-instruct3: https://display.greatruth.cloud/downloads/greatruth-tb4-qc-astra-pass4-cmaes-r01-instruct3.zip
- qc-astra-pass4-cmaes-r01-instruct3 SHA-256: dfd6d818a2b3efed1e3ffe876a4d3c2197f72d58ff38463d86b1e57911ed60da
- qc-sol-pass4-cmaes-r01-instruct3-rev2: https://display.greatruth.cloud/downloads/greatruth-tb4-qc-sol-pass4-cmaes-r01-instruct3-rev2.zip
- qc-sol-pass4-cmaes-r01-instruct3-rev2 SHA-256: ad8b91f4dc85434ec805a2bd408d8a6145ba8762a507dbf75bd3cba18a680c59
- rate-limiter-gate: https://display.greatruth.cloud/downloads/greatruth-tb4-rate-limiter-gate.zip
- rate-limiter-gate SHA-256: 9bf1c8e90080b7dd4a23fd7a19dafd0c83ca11fc79e8234a36ba196aa8c428c0
- route-convergence-release-gate: https://display.greatruth.cloud/downloads/greatruth-tb4-route-convergence-release-gate.zip
- route-convergence-release-gate SHA-256: 22a63fe814c3072aef7c6efa3a0f345fc2f8ad92a6c872120727f29bf59b5e44
- tile-cache-coherence-gate: https://display.greatruth.cloud/downloads/greatruth-tb4-tile-cache-coherence-gate.zip
- tile-cache-coherence-gate SHA-256: 7c31e70013b6a97acf0657c5f9893d9eaec4e0db14093d7fe9c16df6de36a92d

### DeepSWE Repo-backed Harbor Tasks

- ID: gr-env-deepswe-001
- Category: rl_sft_environment
- Records: 9 Harbor packages
- Viewer: https://display.greatruth.cloud/deepswe/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-deepswe-harbor-samples.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: d3d361439c52e3ad0ea58f4c57221bceca522d3c14c8b355fea79944c64e2488
- Curated Harbor sample bundle: https://display.greatruth.cloud/data/harbor/deepswe-sample.json
- Curated Harbor sample bundle SHA-256: 84dd01c85cd03b8c597e2d9320ddd787ee7f857d2282c971e07cdc6c8960c94d
- Ariadne Codegen Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-ariadne-codegen-harbor-sample.zip
- Ariadne Codegen Harbor ZIP SHA-256: 25ed77ea6d6f65afbf0275491c606fc355a5b4ee9be07dd2ad17f5d05e748932
- Betterleaks Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-betterleaks-harbor-sample.zip
- Betterleaks Harbor ZIP SHA-256: a965a2a62af4e0fc64155e2875719b475f9663efce2b8ddc836f50faa1d14b3c
- envconsul Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-envconsul-harbor-sample.zip
- envconsul Harbor ZIP SHA-256: 0e919b925b1cf932f0f67d6a557ef589a9d6107480fca643df63c9e074aaaba6
- fx Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-fx-harbor-sample.zip
- fx Harbor ZIP SHA-256: bff85aef80ee8a1adcb15cb31dbc87b95301360c36c4741cdb4a6912793b6daa
- lf Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-lf-harbor-sample.zip
- lf Harbor ZIP SHA-256: 6800c73dbf79f26f1b6ef95b32b37346b0967801b1dd7f22432e4996c8c04ced
- Norminette Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-norminette-harbor-sample.zip
- Norminette Harbor ZIP SHA-256: 0e3b1249eda0919482d876856ba3bf6849463ddabc1f5329a616f35e1b45267c
- rel Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-rel-harbor-sample.zip
- rel Harbor ZIP SHA-256: 2a7bfcd7d356eda5af48ee5fb8f1214647111223271c3aad15be2d1b32c85ce0
- ttymap Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-ttymap-harbor-sample.zip
- ttymap Harbor ZIP SHA-256: 30c505665260e583ab03f5afcc6eddb5fc0cef9e7551239f034f41e05adff394
- hotpdf Harbor ZIP: https://display.greatruth.cloud/downloads/greatruth-deepswe-hotpdf-harbor-sample.zip
- hotpdf Harbor ZIP SHA-256: 6e5cd255ad766cc379aad9d45c4ba79ded7e9a0609d1dd2a4a591e5586a27f83
- DeepSWE 五仓库说明书: https://display.greatruth.cloud/downloads/greatruth-deepswe-five-repos-guide.pdf
- DeepSWE 五仓库说明书 SHA-256: 3ba1c5daa5bf12612145b7990f4e197260c2a71be55f0ff54e496db99601cfd2
- hotpdf 说明书: https://display.greatruth.cloud/downloads/greatruth-deepswe-hotpdf-guide.pdf
- hotpdf 说明书 SHA-256: 18106c8f1ab090642a4c94692b0db5bda04e0cca1be2ea2e2a40609086ef877c

### QnA Zod Release Contract Audit

- ID: gr-env-qna-001
- Category: rl_sft_environment
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/qna/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-qna-zod-release-contract-audit.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 03c92386b60072f0754e31b979f89e7cc9e56504a7b56d72306ca9c35b91d8fe
- Curated Harbor sample bundle: https://display.greatruth.cloud/data/harbor/qna-sample.json
- Curated Harbor sample bundle SHA-256: 973c4bbb2fed38fff45260375d35911297492dd41f9f26afb49a97984a74e96d

### Harbor Science & Engineering Simulators

- ID: gr-env-simulator-001
- Category: rl_sft_environment
- Records: 15 tasks
- Viewer: https://display.greatruth.cloud/simulator/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-simulator-m1-m15-sample.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 2db7460cb5328ed95dce6ef253753fce954903677b5187e2787a74d6b2f7d8b6
- Curated Harbor sample bundle: https://display.greatruth.cloud/data/harbor/simulator-sample.json
- Curated Harbor sample bundle SHA-256: 194e579711fc5eda8244a39de1b620e3badbb68db53faed03c9e6abc85fd67af

### GDPevo Enterprise Workflows

- ID: gr-env-gdpevo-001
- Category: rl_sft_environment
- Records: 5 scenarios
- Viewer: https://display.greatruth.cloud/gdpevo/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-gdpevo-sample.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 7189d0886c8e8e9b87b5abdc519cee9c7d1bcc8bdab5e9eba5d487d3adc91c29
- Curated Harbor sample bundle: https://display.greatruth.cloud/data/harbor/gdpevo-sample.json
- Curated Harbor sample bundle SHA-256: 2cec5e94fbfe97adc50b9fec76a33b4f9903dfd72f712bddea62de4feeaa3078

### Cyber Defense Seed 176

- ID: gr-env-cyber-001
- Category: rl_sft_environment
- Records: 1 trajectories
- Viewer: https://display.greatruth.cloud/cyber/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-cyber-defense-seed176.tar.gz
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 7b2751fbc6274dc0b09ebce543ce6fb92c3ae3baf51f6369e084c8e3a482d724
- Curated Harbor sample bundle: https://display.greatruth.cloud/data/harbor/cyber-sample.json
- Curated Harbor sample bundle SHA-256: 635517c8b54eea5104f5dce786b06092f0978199b9aa2c4fad88a2e2b6bbb437
- Harbor task and rollout bundle: https://display.greatruth.cloud/downloads/greatruth-cyber-defense-seed176.tar.gz
- Harbor task and rollout bundle SHA-256: 7b2751fbc6274dc0b09ebce543ce6fb92c3ae3baf51f6369e084c8e3a482d724
- 精简样例 JSON: https://display.greatruth.cloud/downloads/greatruth-cyber-defense-seed176-sample.json
- 精简样例 JSON SHA-256: 2c3f5079e605275e0a2e55f2d9a638a0898ae89391c7b5d40db1a5bd53391e43

### Cyber · RF-VULHUB Two-layer Range

- ID: gr-env-cyber-rf-vulhub-001
- Category: rl_sft_environment
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/cyber-rf-vulhub/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-cyber-rf-vulhub-four-v2l.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: faddfcc1534b4ce957b7e60e99937412bf2e485bb241971223751940f2753c9a
- RF-VULHUB structured sample: https://display.greatruth.cloud/data/harbor/cyber-rf-vulhub-sample.json
- RF-VULHUB structured sample SHA-256: 84f689b785d37faf5df8fcf7353bb72fa852eb624cfef04f3c859f7f119eb594
- 说明书 PDF: https://display.greatruth.cloud/downloads/greatruth-cyber-rf-vulhub-guide.pdf
- 说明书 PDF SHA-256: ff34718c95605114902b3d7e6abb4eefbe4cf7af362ba52b7f38dfaea8f6d064

### ClawBench · Browser Tool-use Tasks

- ID: gr-env-clawbench-001
- Category: rl_sft_environment
- Records: 10 tasks
- Viewer: https://display.greatruth.cloud/clawbench/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-clawbench-diverse-sample.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: ea019d0bc3db80d65965ab31ddba95dadf87ecd7cd860e7611a31f3eeec77ef2
- ClawBench structured sample: https://display.greatruth.cloud/data/harbor/clawbench-sample.json
- ClawBench structured sample SHA-256: 7983b6ab6100ba4fe64c3b05305bb29990997db3c41d7edcfcc6d909b8114a60
- 配套说明书 PDF: https://display.greatruth.cloud/downloads/greatruth-clawbench-diverse-guide.pdf
- 配套说明书 PDF SHA-256: fecc7d11af0bd9c98078ecddf634fec5c8c752e51d882795f2827d29af9a48fe

### OpenClawBench · WPS Desktop Task

- ID: gr-env-clawbench-wps-001
- Category: rl_sft_environment
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/clawbench-wps/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-clawbench-wps-standalone.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 238c8ca0365e112822a056d51342190b787190ca8db53d10a2a6cc2461cac597
- WPS structured sample: https://display.greatruth.cloud/data/harbor/clawbench-wps-sample.json
- WPS structured sample SHA-256: 71035f1b768447ea95e11b9029c3b6ce3ca1a7e99ea5702019290989e00fe68a
- 配套说明书 PDF: https://display.greatruth.cloud/downloads/greatruth-clawbench-wps-guide.pdf
- 配套说明书 PDF SHA-256: a1403c996dd95b4d5fb45132320b4f4dbb234b22959875f725c0585ace189ce6

### Computer Use · OS Sandbox Tasks

- ID: gr-env-computer-use-001
- Category: rl_sft_environment
- Records: 2 tasks
- Viewer: https://display.greatruth.cloud/computer-use/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-computer-use-os-sandboxes.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: b841436ab6ba48b0ac923ffdd9ef1811a286f12ef2c7dde9bb3f75c9b0e7c71f
- Computer Use structured sample: https://display.greatruth.cloud/data/harbor/computer-use-sample.json
- Computer Use structured sample SHA-256: 77448695c11fe2be0f28e52bb8a3b50c2ce5d75f4af755dc6a9be0af51db5a5a

### AutomationBench · Support Workflows

- ID: gr-env-automationbench-001
- Category: rl_sft_environment
- Records: 7 tasks
- Viewer: https://display.greatruth.cloud/automationbench/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-automationbench-support-workflows.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 56f17cf91485fe5ff5de294d76b19d8eef87291f092e4d6535144d8a898da71f
- AutomationBench structured sample: https://display.greatruth.cloud/data/harbor/automationbench-sample.json
- AutomationBench structured sample SHA-256: 34a60f8ad9603a1f70820f9f611efbb5c4c1078ba563b1ff604d5cfbd2c41074

### JobBench · Professional Deliverable Tasks

- ID: gr-env-jobbench-001
- Category: rl_sft_environment
- Records: 5 tasks
- Viewer: https://display.greatruth.cloud/jobbench/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-jobbench-professional-tasks.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: e12142d672e11ee551a93c8363f9ff262e41850418401a7ff5f41fd67bbd7ef0
- JobBench structured sample: https://display.greatruth.cloud/data/harbor/jobbench-sample.json
- JobBench structured sample SHA-256: abf4a006858840e1cba440db9273fbb092156d03f74d1f593488f99a45be0526

### MCP Multi-tool Synthetic Trajectories

- ID: gr-synthetic-mcp-001
- Category: high_quality_trajectory
- Records: 17 trajectories
- Viewer steps: 216
- Source events: undefined
- Tool calls: 495
- Estimated token length: 未记录
- Main trajectory length: 未记录
- Token estimation: not available
- Viewer: https://display.greatruth.cloud/mcp-traces/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-mcp-synthetic-trajectories-17.zip
- Schema: https://display.greatruth.cloud/data/schemas/mcp-synthetic-traces.schema.json
- SHA-256: 844d41080021102ef9caabc5307a9cef73b1d28b937c9cd47a60efbbb8d21aa0
- MCP synthetic trajectory display sample: https://display.greatruth.cloud/data/synthetic/mcp-traces-sample.json
- MCP synthetic trajectory display sample SHA-256: dd7ad0b72427071cc65dc600a66484394ccebb802e4b57589c87b42c329b3843

### WorkBuddy Multi-agent Trajectories

- ID: gr-synthetic-mixed-001
- Category: high_quality_trajectory
- Records: 12 trajectories
- Viewer steps: 15865
- Source events: 15865
- Tool calls: 7287
- Estimated token length: 未记录
- Main trajectory length: 未记录
- Token estimation: not available
- Viewer: https://display.greatruth.cloud/mixed-trajectories/viewer.html
- Download: https://display.greatruth.cloud/downloads/greatruth-mixed-collaborative-trajectories-12.zip
- Schema: https://display.greatruth.cloud/data/schemas/mixed-collaborative-trajectories.schema.json
- SHA-256: 0ccd60c76883bf736ac5eb30397ecc83b36bad77bea6018c775ec9d40731e628
- Mixed collaborative trajectory display sample: https://display.greatruth.cloud/data/synthetic/mixed-trajectories-sample.json
- Mixed collaborative trajectory display sample SHA-256: 57f97758ee6f7ab008eb0af5eab06d9850968973c8b51042fc7a9a1b74bee6ee

### Cantor diagonal proof for infinite Boolean sequences

- ID: gr-vertical-coding-cl-t8-003-lean
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-coding-cl-t8-003-lean/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-coding-cl-t8-003-lean.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: d386597f16f2937dea9e0ae19e838b2d8e8fb63a6833c31ca8bef3bf67a795ca
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-coding-cl-t8-003-lean-sample.json
- Structured vertical sample SHA-256: 491559a3955ed648403abe50ba270d65b819d0d74fef18b3b10a6da1a7df370e
- Companion guide: https://display.greatruth.cloud/downloads/vertical-coding-cl-t8-003-lean-guide.pdf
- Companion guide SHA-256: 401697ce5e4845afda9fe1ffa635b3792c2ba60e24537a5d5baeabe76e09e6ab

### Industrial vision acquisition board dual-clock packet-commit FIFO

- ID: gr-vertical-coding-vhdl-t8-001-fifo
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-coding-vhdl-t8-001-fifo/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-coding-vhdl-t8-001-fifo.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 68339ef8c0add3d4a0aea422c8f108f0dde9715fd161a33c0fb2ee52cc159663
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-coding-vhdl-t8-001-fifo-sample.json
- Structured vertical sample SHA-256: c0a693c9afb8df8b2c13cf7c1f6115d0636aeb5b440836b9323f0a2a5266e250
- Companion guide: https://display.greatruth.cloud/downloads/vertical-coding-vhdl-t8-001-fifo-guide.pdf
- Companion guide SHA-256: 1db789901da3f1bfde8dd4da33d1a0025693bd98f49776fab061d14350450a12

### Release portfolio decision task

- ID: gr-vertical-pm-t8-004
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-pm-t8-004/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-pm-t8-004.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: a83afa5c6e702c5221edf92b9ec26d1daf34770d4e6281355bb1050c35395693
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-pm-t8-004-sample.json
- Structured vertical sample SHA-256: 9be4696abe2b5bf4793ae3f5081627feec102bad9c7d3dbecf17aaad1415682d
- Companion guide: https://display.greatruth.cloud/downloads/vertical-pm-t8-004-guide.pdf
- Companion guide SHA-256: 2f1ede3a36e382ab8691c7241a2565282a7b64c9ede1ce9c877ba5958f4de59a

### Consulting-CONS-T4-019-Operating Driver Tree

- ID: gr-vertical-cons-t4-019
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-cons-t4-019/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-cons-t4-019.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 09068f66b8bd4bc9f248ebef0af40aa68a7c8e60ef115b35c88f215abc58fe8e
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-cons-t4-019-sample.json
- Structured vertical sample SHA-256: 274e254dac5d1646c6e6cb7e5e0e796d5a41cb0046e879d21cd618054fb0aefb
- Companion guide: https://display.greatruth.cloud/downloads/vertical-cons-t4-019-guide.pdf
- Companion guide SHA-256: 2f2a305be2dde0905f4f09e3498d7b5830554f60bfa73bd675529ef7417dfaef

### Consulting-HV3-CONS-0002-Schiphol Double Materiality

- ID: gr-vertical-hv3-cons-0002
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-hv3-cons-0002/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-hv3-cons-0002.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 3e27402a7ba42198b183540530a35524b2207db807cdb45b0c5b2d7d41ce1ea9
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-hv3-cons-0002-sample.json
- Structured vertical sample SHA-256: 95bd066d2ee9c4fe87041930186a28971bb0c1d49a18ec47064b7c84349aa03c
- Companion guide: https://display.greatruth.cloud/downloads/vertical-hv3-cons-0002-guide.pdf
- Companion guide SHA-256: 3b883c81e1d94c8e824828f269ccfddceea1737e33cc166e36dd768ff0235d2d

### Last-mile delivery: three closes, evidence audit, and stress-scenario recovery scheduling

- ID: gr-vertical-log-t8-l4
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-log-t8-l4/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-log-t8-l4.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 893f64a015e3ab2adb21ff83000eaf35369b545f4b9f744a263bce6c50261d9b
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-log-t8-l4-sample.json
- Structured vertical sample SHA-256: f755e6b2225c979b1e34277247a9c7870aca9ce283b3d58a7f2ee3f7106a234b
- Companion guide: https://display.greatruth.cloud/downloads/vertical-log-t8-l4-guide.pdf
- Companion guide SHA-256: ddd7e5a3148fa04c76620ffb51fe296de8664bc371c67d2e32ad622fe1c53b8e

### E-commerce-Ecom-T1-001-Cross-Market GMV

- ID: gr-vertical-ecom-t1-001-gmv
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-ecom-t1-001-gmv/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-ecom-t1-001-gmv.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 34395130f0853cf50c9b1e7943ded9cf3450459fa49dea9d7b51abcd72634d6b
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-ecom-t1-001-gmv-sample.json
- Structured vertical sample SHA-256: 0497b5c66eadc53136d1085750e2dd9b3e3f402d1f49b67582184158d7d052a8
- Companion guide: https://display.greatruth.cloud/downloads/vertical-ecom-t1-001-gmv-guide.pdf
- Companion guide SHA-256: d86937ff8fb36c1ce9f5e8bd07c65c628712a87d7146284df7a7196fd56b62c2

### Event-level attack-chain investigation and incident response in enterprise Windows logs

- ID: gr-vertical-cy-t8-001
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-cy-t8-001/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-cy-t8-001.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 6256af67cff42aff15b1bb3cb2cce10b8e05ccf629b701f88c8bd4250c38a181
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-cy-t8-001-sample.json
- Structured vertical sample SHA-256: cf5e49c3fd286e601398fc67e282ef2d76a7d6250ff6b172e72c6300a56f474f
- Companion guide: https://display.greatruth.cloud/downloads/vertical-cy-t8-001-guide.pdf
- Companion guide SHA-256: 5e5365a5efac65ad329dcf224168fd4ce663d73a32bf385edd28052f5585ed36

### General Office-HV3-GENE-0001-Interface Bridge Renewal

- ID: gr-vertical-hv3-gene-0001
- Category: vertical_domain
- Records: 1 tasks
- Viewer: https://display.greatruth.cloud/vertical-hv3-gene-0001/viewer.html
- Download: https://display.greatruth.cloud/downloads/vertical-hv3-gene-0001.zip
- Schema: https://display.greatruth.cloud/data/schemas/harbor-env.schema.json
- SHA-256: 7e4b09b85c3d65e7a3e3fdc25954f900faa7cb24238f91e50bc9dc274074e762
- Structured vertical sample: https://display.greatruth.cloud/data/vertical/vertical-hv3-gene-0001-sample.json
- Structured vertical sample SHA-256: 46286d48f6b8ecc2e410f1a4520903c178dedd3ac6d3ca4fa9cf16ba0ec82819
- Companion guide: https://display.greatruth.cloud/downloads/vertical-hv3-gene-0001-guide.pdf
- Companion guide SHA-256: 926a8dba2310fbb4744d7a17c6351ec496b8109672565ffa40df3b22faa856c6

## Evaluation notes

Use the catalog as the source of truth and verify downloads with SHA-256. Academic evaluation scores are diagnostic signals for model improvement, while trajectory quality scores describe structural completeness. Public samples require no authentication; full data and production usage are governed by the applicable Alvance project agreement.