id
stringlengths
11
11
created
timestamp[s]date
2026-01-01 00:00:00
2026-01-01 00:00:00
topic
stringclasses
12 values
task_type
stringclasses
8 values
difficulty
stringclasses
4 values
instruction
stringlengths
201
264
input
stringclasses
1 value
output
stringclasses
7 values
metadata
dict
train_00200
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
code
intermediate
Task: code Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: intermediate Target language: Rust Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Rust", "developer_needs": [ "repo_scale_reasoning", "reproducibility", "documentation" ] }
train_00201
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
explain
expert
Task: explain Topic: SWE-bench style real-repo evaluation Difficulty: expert Target language: Rust Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Rust", "developer_needs": [ "ci_integration", "documentation", "evaluation_metrics" ] }
train_00202
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
agent_loop
expert
Task: agent_loop Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: SQL Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "SQL", "developer_needs": [ "tooling", "repo_scale_reasoning", "cost_latency_tradeoffs" ] }
train_00203
2026-01-01T00:00:00
Extended context and repo-scale understanding
eval
advanced
Task: eval Topic: Extended context and repo-scale understanding Difficulty: advanced Target language: Rust Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Rust", "developer_needs": [ "tooling", "evaluation_metrics", "repo_scale_reasoning" ] }
train_00204
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
agent_loop
intermediate
Task: agent_loop Topic: Reasoning-first coding models and tunable deliberation Difficulty: intermediate Target language: Rust Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "Rust", "developer_needs": [ "documentation", "cost_latency_tradeoffs", "reproducibility" ] }
train_00205
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
explain
expert
Task: explain Topic: Reasoning-first coding models and tunable deliberation Difficulty: expert Target language: TypeScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "TypeScript", "developer_needs": [ "evaluation_metrics", "tests_are_truth", "security_gates" ] }
train_00206
2026-01-01T00:00:00
Extended context and repo-scale understanding
compare
advanced
Task: compare Topic: Extended context and repo-scale understanding Difficulty: advanced Target language: TypeScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "TypeScript", "developer_needs": [ "security_gates", "governance", "documentation" ] }
train_00207
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
eval
intermediate
Task: eval Topic: Mixture-of-Experts (MoE) for code Difficulty: intermediate Target language: Go Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Go", "developer_needs": [ "tests_are_truth", "evaluation_metrics", "ci_integration" ] }
train_00208
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
review
intermediate
Task: review Topic: Tool calling, sandboxes, and CI integration Difficulty: intermediate Target language: Python Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Python", "developer_needs": [ "documentation", "security_gates", "cost_latency_tradeoffs" ] }
train_00209
2026-01-01T00:00:00
Extended context and repo-scale understanding
compare
expert
Task: compare Topic: Extended context and repo-scale understanding Difficulty: expert Target language: Go Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Go", "developer_needs": [ "documentation", "ci_integration", "tooling" ] }
train_00210
2026-01-01T00:00:00
Secure code generation and policy gates
agent_loop
advanced
Task: agent_loop Topic: Secure code generation and policy gates Difficulty: advanced Target language: TypeScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "TypeScript", "developer_needs": [ "security_gates", "tooling", "evaluation_metrics" ] }
train_00211
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
data_pipeline
foundation
Task: data_pipeline Topic: SWE-bench style real-repo evaluation Difficulty: foundation Target language: Go Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Go", "developer_needs": [ "repo_scale_reasoning", "ci_integration", "documentation" ] }
train_00212
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
explain
intermediate
Task: explain Topic: Reasoning-first coding models and tunable deliberation Difficulty: intermediate Target language: JavaScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "JavaScript", "developer_needs": [ "tooling", "governance", "documentation" ] }
train_00213
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
data_pipeline
intermediate
Task: data_pipeline Topic: Mixture-of-Experts (MoE) for code Difficulty: intermediate Target language: JavaScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "JavaScript", "developer_needs": [ "cost_latency_tradeoffs", "tests_are_truth", "repo_scale_reasoning" ] }
train_00214
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
design
foundation
Task: design Topic: Governance, provenance, and licensing for code data Difficulty: foundation Target language: Bash Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "security_gates", "tooling", "governance" ] }
train_00215
2026-01-01T00:00:00
Secure code generation and policy gates
review
foundation
Task: review Topic: Secure code generation and policy gates Difficulty: foundation Target language: JavaScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "JavaScript", "developer_needs": [ "governance", "ci_integration", "tests_are_truth" ] }
train_00216
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
review
expert
Task: review Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: expert Target language: JavaScript Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "JavaScript", "developer_needs": [ "ci_integration", "security_gates", "tests_are_truth" ] }
train_00217
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
explain
advanced
Task: explain Topic: Governance, provenance, and licensing for code data Difficulty: advanced Target language: TypeScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "TypeScript", "developer_needs": [ "cost_latency_tradeoffs", "security_gates", "documentation" ] }
train_00218
2026-01-01T00:00:00
Secure code generation and policy gates
explain
intermediate
Task: explain Topic: Secure code generation and policy gates Difficulty: intermediate Target language: TypeScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "TypeScript", "developer_needs": [ "ci_integration", "documentation", "tooling" ] }
train_00219
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
data_pipeline
advanced
Task: data_pipeline Topic: SWE-bench style real-repo evaluation Difficulty: advanced Target language: Bash Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Bash", "developer_needs": [ "evaluation_metrics", "cost_latency_tradeoffs", "documentation" ] }
train_00220
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
code
foundation
Task: code Topic: Mixture-of-Experts (MoE) for code Difficulty: foundation Target language: JavaScript Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "JavaScript", "developer_needs": [ "tooling", "evaluation_metrics", "ci_integration" ] }
train_00221
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
code
intermediate
Task: code Topic: Code-specialized model families and sizing tradeoffs Difficulty: intermediate Target language: Bash Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "governance", "evaluation_metrics", "reproducibility" ] }
train_00222
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
compare
foundation
Task: compare Topic: Reasoning-first coding models and tunable deliberation Difficulty: foundation Target language: Bash Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Bash", "developer_needs": [ "documentation", "evaluation_metrics", "governance" ] }
train_00223
2026-01-01T00:00:00
Secure code generation and policy gates
review
foundation
Task: review Topic: Secure code generation and policy gates Difficulty: foundation Target language: C# Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "C#", "developer_needs": [ "ci_integration", "reproducibility", "governance" ] }
train_00224
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
eval
expert
Task: eval Topic: Model merging, distillation, and continued pretraining Difficulty: expert Target language: JavaScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "JavaScript", "developer_needs": [ "ci_integration", "cost_latency_tradeoffs", "evaluation_metrics" ] }
train_00225
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
agent_loop
intermediate
Task: agent_loop Topic: Mixture-of-Experts (MoE) for code Difficulty: intermediate Target language: Python Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "Python", "developer_needs": [ "tests_are_truth", "governance", "documentation" ] }
train_00226
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
agent_loop
intermediate
Task: agent_loop Topic: Code-specialized model families and sizing tradeoffs Difficulty: intermediate Target language: SQL Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "SQL", "developer_needs": [ "ci_integration", "governance", "evaluation_metrics" ] }
train_00227
2026-01-01T00:00:00
Secure code generation and policy gates
data_pipeline
expert
Task: data_pipeline Topic: Secure code generation and policy gates Difficulty: expert Target language: C# Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "C#", "developer_needs": [ "tests_are_truth", "repo_scale_reasoning", "evaluation_metrics" ] }
train_00228
2026-01-01T00:00:00
Extended context and repo-scale understanding
review
advanced
Task: review Topic: Extended context and repo-scale understanding Difficulty: advanced Target language: Python Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Python", "developer_needs": [ "security_gates", "tooling", "tests_are_truth" ] }
train_00229
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
foundation
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: foundation Target language: Go Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Go", "developer_needs": [ "documentation", "evaluation_metrics", "repo_scale_reasoning" ] }
train_00230
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
code
foundation
Task: code Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: foundation Target language: Go Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Go", "developer_needs": [ "tooling", "documentation", "reproducibility" ] }
train_00231
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
data_pipeline
expert
Task: data_pipeline Topic: Code-specialized model families and sizing tradeoffs Difficulty: expert Target language: Java Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Java", "developer_needs": [ "reproducibility", "documentation", "repo_scale_reasoning" ] }
train_00232
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
eval
foundation
Task: eval Topic: Code-specialized model families and sizing tradeoffs Difficulty: foundation Target language: Java Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Java", "developer_needs": [ "reproducibility", "documentation", "governance" ] }
train_00233
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
review
advanced
Task: review Topic: Reasoning-first coding models and tunable deliberation Difficulty: advanced Target language: Rust Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Rust", "developer_needs": [ "governance", "reproducibility", "evaluation_metrics" ] }
train_00234
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
review
intermediate
Task: review Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: intermediate Target language: Java Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Java", "developer_needs": [ "repo_scale_reasoning", "tests_are_truth", "ci_integration" ] }
train_00235
2026-01-01T00:00:00
Secure code generation and policy gates
explain
foundation
Task: explain Topic: Secure code generation and policy gates Difficulty: foundation Target language: JavaScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "JavaScript", "developer_needs": [ "security_gates", "repo_scale_reasoning", "documentation" ] }
train_00236
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
review
foundation
Task: review Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: foundation Target language: C# Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "C#", "developer_needs": [ "security_gates", "tooling", "cost_latency_tradeoffs" ] }
train_00237
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
code
intermediate
Task: code Topic: Model merging, distillation, and continued pretraining Difficulty: intermediate Target language: Bash Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "tests_are_truth", "evaluation_metrics", "repo_scale_reasoning" ] }
train_00238
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
review
foundation
Task: review Topic: Tool calling, sandboxes, and CI integration Difficulty: foundation Target language: SQL Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "SQL", "developer_needs": [ "tooling", "evaluation_metrics", "reproducibility" ] }
train_00239
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
eval
advanced
Task: eval Topic: Reasoning-first coding models and tunable deliberation Difficulty: advanced Target language: JavaScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "JavaScript", "developer_needs": [ "tooling", "repo_scale_reasoning", "documentation" ] }
train_00240
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
agent_loop
expert
Task: agent_loop Topic: Mixture-of-Experts (MoE) for code Difficulty: expert Target language: Python Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "Python", "developer_needs": [ "reproducibility", "security_gates", "cost_latency_tradeoffs" ] }
train_00241
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
advanced
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: advanced Target language: Bash Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Bash", "developer_needs": [ "tooling", "tests_are_truth", "cost_latency_tradeoffs" ] }
train_00242
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
data_pipeline
expert
Task: data_pipeline Topic: Tool calling, sandboxes, and CI integration Difficulty: expert Target language: JavaScript Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "JavaScript", "developer_needs": [ "cost_latency_tradeoffs", "repo_scale_reasoning", "reproducibility" ] }
train_00243
2026-01-01T00:00:00
Extended context and repo-scale understanding
compare
foundation
Task: compare Topic: Extended context and repo-scale understanding Difficulty: foundation Target language: Go Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Go", "developer_needs": [ "tooling", "security_gates", "ci_integration" ] }
train_00244
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
data_pipeline
foundation
Task: data_pipeline Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: foundation Target language: TypeScript Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "TypeScript", "developer_needs": [ "tests_are_truth", "repo_scale_reasoning", "evaluation_metrics" ] }
train_00245
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
eval
intermediate
Task: eval Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: intermediate Target language: Bash Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Bash", "developer_needs": [ "governance", "tests_are_truth", "reproducibility" ] }
train_00246
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
compare
expert
Task: compare Topic: SWE-bench style real-repo evaluation Difficulty: expert Target language: SQL Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "SQL", "developer_needs": [ "tests_are_truth", "documentation", "governance" ] }
train_00247
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
review
foundation
Task: review Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: foundation Target language: JavaScript Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "JavaScript", "developer_needs": [ "tests_are_truth", "repo_scale_reasoning", "security_gates" ] }
train_00248
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
compare
foundation
Task: compare Topic: Tool calling, sandboxes, and CI integration Difficulty: foundation Target language: Python Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Python", "developer_needs": [ "tests_are_truth", "tooling", "security_gates" ] }
train_00249
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
code
intermediate
Task: code Topic: Governance, provenance, and licensing for code data Difficulty: intermediate Target language: Python Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Reference scaffold: ```python def agent_loop(plan, edit, test, issue, max_iters=3): history = [] p = plan(issue) for _ in range(max_iters): patch = edit(issue, p) ok, report = test(patch) history.append({"plan": p, "passed": ok, "report": report[:200]}) if ok: return patch, history p = p + " | refine from failures" return patch, history ``` Operational notes: sandbox, pinned deps, human gate.
{ "target_language": "Python", "developer_needs": [ "evaluation_metrics", "tests_are_truth", "documentation" ] }
train_00250
2026-01-01T00:00:00
Extended context and repo-scale understanding
code
intermediate
Task: code Topic: Extended context and repo-scale understanding Difficulty: intermediate Target language: Bash Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "tooling", "evaluation_metrics", "ci_integration" ] }
train_00251
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
eval
foundation
Task: eval Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: foundation Target language: Bash Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Bash", "developer_needs": [ "repo_scale_reasoning", "tooling", "documentation" ] }
train_00252
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
advanced
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: advanced Target language: Python Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Python", "developer_needs": [ "security_gates", "tests_are_truth", "repo_scale_reasoning" ] }
train_00253
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
compare
intermediate
Task: compare Topic: Code-specialized model families and sizing tradeoffs Difficulty: intermediate Target language: Bash Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Bash", "developer_needs": [ "security_gates", "tests_are_truth", "tooling" ] }
train_00254
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
review
foundation
Task: review Topic: Model merging, distillation, and continued pretraining Difficulty: foundation Target language: Rust Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Rust", "developer_needs": [ "reproducibility", "tests_are_truth", "tooling" ] }
train_00255
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
code
expert
Task: code Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: Python Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Reference scaffold: ```python def agent_loop(plan, edit, test, issue, max_iters=3): history = [] p = plan(issue) for _ in range(max_iters): patch = edit(issue, p) ok, report = test(patch) history.append({"plan": p, "passed": ok, "report": report[:200]}) if ok: return patch, history p = p + " | refine from failures" return patch, history ``` Operational notes: sandbox, pinned deps, human gate.
{ "target_language": "Python", "developer_needs": [ "cost_latency_tradeoffs", "governance", "tests_are_truth" ] }
train_00256
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
review
expert
Task: review Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: expert Target language: Bash Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Bash", "developer_needs": [ "tests_are_truth", "evaluation_metrics", "reproducibility" ] }
train_00257
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
eval
intermediate
Task: eval Topic: Model merging, distillation, and continued pretraining Difficulty: intermediate Target language: Go Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Go", "developer_needs": [ "governance", "evaluation_metrics", "cost_latency_tradeoffs" ] }
train_00258
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
compare
foundation
Task: compare Topic: Model merging, distillation, and continued pretraining Difficulty: foundation Target language: Bash Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Bash", "developer_needs": [ "tooling", "repo_scale_reasoning", "governance" ] }
train_00259
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
eval
foundation
Task: eval Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: foundation Target language: Python Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Python", "developer_needs": [ "tooling", "reproducibility", "governance" ] }
train_00260
2026-01-01T00:00:00
Model merging, distillation, and continued pretraining
compare
intermediate
Task: compare Topic: Model merging, distillation, and continued pretraining Difficulty: intermediate Target language: Bash Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "Bash", "developer_needs": [ "security_gates", "documentation", "repo_scale_reasoning" ] }
train_00261
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
data_pipeline
advanced
Task: data_pipeline Topic: Governance, provenance, and licensing for code data Difficulty: advanced Target language: Go Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Go", "developer_needs": [ "tests_are_truth", "evaluation_metrics", "repo_scale_reasoning" ] }
train_00262
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
code
expert
Task: code Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: Go Context: Integrate an LLM agent into CI for a large monorepo. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Go", "developer_needs": [ "tooling", "security_gates", "evaluation_metrics" ] }
train_00263
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
data_pipeline
intermediate
Task: data_pipeline Topic: Tool calling, sandboxes, and CI integration Difficulty: intermediate Target language: TypeScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "TypeScript", "developer_needs": [ "repo_scale_reasoning", "reproducibility", "tests_are_truth" ] }
train_00264
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
design
expert
Task: design Topic: SWE-bench style real-repo evaluation Difficulty: expert Target language: C# Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "C#", "developer_needs": [ "cost_latency_tradeoffs", "evaluation_metrics", "governance" ] }
train_00265
2026-01-01T00:00:00
Extended context and repo-scale understanding
data_pipeline
advanced
Task: data_pipeline Topic: Extended context and repo-scale understanding Difficulty: advanced Target language: Bash Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Bash", "developer_needs": [ "tooling", "reproducibility", "security_gates" ] }
train_00266
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
data_pipeline
intermediate
Task: data_pipeline Topic: Governance, provenance, and licensing for code data Difficulty: intermediate Target language: Rust Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Rust", "developer_needs": [ "documentation", "evaluation_metrics", "cost_latency_tradeoffs" ] }
train_00267
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
code
intermediate
Task: code Topic: SWE-bench style real-repo evaluation Difficulty: intermediate Target language: Java Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Java", "developer_needs": [ "security_gates", "ci_integration", "reproducibility" ] }
train_00268
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
eval
intermediate
Task: eval Topic: Tool calling, sandboxes, and CI integration Difficulty: intermediate Target language: C# Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "C#", "developer_needs": [ "evaluation_metrics", "repo_scale_reasoning", "reproducibility" ] }
train_00269
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
explain
expert
Task: explain Topic: Code-specialized model families and sizing tradeoffs Difficulty: expert Target language: Bash Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "reproducibility", "repo_scale_reasoning", "tests_are_truth" ] }
train_00270
2026-01-01T00:00:00
Mixture-of-Experts (MoE) for code
data_pipeline
intermediate
Task: data_pipeline Topic: Mixture-of-Experts (MoE) for code Difficulty: intermediate Target language: Rust Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Rust", "developer_needs": [ "evaluation_metrics", "security_gates", "reproducibility" ] }
train_00271
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
code
advanced
Task: code Topic: Code-specialized model families and sizing tradeoffs Difficulty: advanced Target language: Python Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Reference scaffold: ```python def agent_loop(plan, edit, test, issue, max_iters=3): history = [] p = plan(issue) for _ in range(max_iters): patch = edit(issue, p) ok, report = test(patch) history.append({"plan": p, "passed": ok, "report": report[:200]}) if ok: return patch, history p = p + " | refine from failures" return patch, history ``` Operational notes: sandbox, pinned deps, human gate.
{ "target_language": "Python", "developer_needs": [ "tests_are_truth", "ci_integration", "cost_latency_tradeoffs" ] }
train_00272
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
design
intermediate
Task: design Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: intermediate Target language: TypeScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "TypeScript", "developer_needs": [ "tooling", "repo_scale_reasoning", "security_gates" ] }
train_00273
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
data_pipeline
expert
Task: data_pipeline Topic: SWE-bench style real-repo evaluation Difficulty: expert Target language: C# Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "C#", "developer_needs": [ "tests_are_truth", "cost_latency_tradeoffs", "ci_integration" ] }
train_00274
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
compare
foundation
Task: compare Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: foundation Target language: C# Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "C#", "developer_needs": [ "evaluation_metrics", "tests_are_truth", "governance" ] }
train_00275
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
explain
expert
Task: explain Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: JavaScript Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "JavaScript", "developer_needs": [ "reproducibility", "evaluation_metrics", "security_gates" ] }
train_00276
2026-01-01T00:00:00
Governance, provenance, and licensing for code data
eval
advanced
Task: eval Topic: Governance, provenance, and licensing for code data Difficulty: advanced Target language: Python Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "Python", "developer_needs": [ "cost_latency_tradeoffs", "evaluation_metrics", "tests_are_truth" ] }
train_00277
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
intermediate
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: intermediate Target language: Python Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Python", "developer_needs": [ "governance", "reproducibility", "evaluation_metrics" ] }
train_00278
2026-01-01T00:00:00
Extended context and repo-scale understanding
review
expert
Task: review Topic: Extended context and repo-scale understanding Difficulty: expert Target language: Java Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Java", "developer_needs": [ "documentation", "governance", "tests_are_truth" ] }
train_00279
2026-01-01T00:00:00
Extended context and repo-scale understanding
eval
foundation
Task: eval Topic: Extended context and repo-scale understanding Difficulty: foundation Target language: SQL Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Eval: - Tasks: real issues - Metrics: pass@k, time-to-green - Gates: lint/security
{ "target_language": "SQL", "developer_needs": [ "tooling", "ci_integration", "repo_scale_reasoning" ] }
train_00280
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
design
advanced
Task: design Topic: Reasoning-first coding models and tunable deliberation Difficulty: advanced Target language: Bash Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "governance", "cost_latency_tradeoffs", "repo_scale_reasoning" ] }
train_00281
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
explain
advanced
Task: explain Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: advanced Target language: TypeScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "TypeScript", "developer_needs": [ "evaluation_metrics", "governance", "documentation" ] }
train_00282
2026-01-01T00:00:00
SWE-bench style real-repo evaluation
review
foundation
Task: review Topic: SWE-bench style real-repo evaluation Difficulty: foundation Target language: SQL Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "SQL", "developer_needs": [ "evaluation_metrics", "cost_latency_tradeoffs", "security_gates" ] }
train_00283
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
intermediate
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: intermediate Target language: TypeScript Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "TypeScript", "developer_needs": [ "documentation", "reproducibility", "evaluation_metrics" ] }
train_00284
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
data_pipeline
intermediate
Task: data_pipeline Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: intermediate Target language: TypeScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "TypeScript", "developer_needs": [ "ci_integration", "cost_latency_tradeoffs", "reproducibility" ] }
train_00285
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
explain
intermediate
Task: explain Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: intermediate Target language: Rust Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Rust", "developer_needs": [ "ci_integration", "security_gates", "governance" ] }
train_00286
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
data_pipeline
foundation
Task: data_pipeline Topic: Code-specialized model families and sizing tradeoffs Difficulty: foundation Target language: JavaScript Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "JavaScript", "developer_needs": [ "ci_integration", "governance", "documentation" ] }
train_00287
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
agent_loop
advanced
Task: agent_loop Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: advanced Target language: Python Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Agent loop: Plan → Edit (diff) → Test → Reflect → Human gate
{ "target_language": "Python", "developer_needs": [ "documentation", "governance", "reproducibility" ] }
train_00288
2026-01-01T00:00:00
Secure code generation and policy gates
compare
intermediate
Task: compare Topic: Secure code generation and policy gates Difficulty: intermediate Target language: TypeScript Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "TypeScript", "developer_needs": [ "repo_scale_reasoning", "ci_integration", "documentation" ] }
train_00289
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
code
expert
Task: code Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: Go Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Go", "developer_needs": [ "repo_scale_reasoning", "documentation", "governance" ] }
train_00290
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
code
expert
Task: code Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: expert Target language: Bash Context: Create an eval harness that reflects real developer workflows. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "Bash", "developer_needs": [ "repo_scale_reasoning", "cost_latency_tradeoffs", "documentation" ] }
train_00291
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
review
advanced
Task: review Topic: Reasoning-first coding models and tunable deliberation Difficulty: advanced Target language: Java Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Java", "developer_needs": [ "tooling", "cost_latency_tradeoffs", "security_gates" ] }
train_00292
2026-01-01T00:00:00
Multimodal dev workflows (docs, diagrams, traces)
data_pipeline
expert
Task: data_pipeline Topic: Multimodal dev workflows (docs, diagrams, traces) Difficulty: expert Target language: JavaScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "JavaScript", "developer_needs": [ "ci_integration", "reproducibility", "tooling" ] }
train_00293
2026-01-01T00:00:00
Dataset curation pipelines (filter, dedupe, quality)
review
expert
Task: review Topic: Dataset curation pipelines (filter, dedupe, quality) Difficulty: expert Target language: Java Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Review: correctness, security, performance, governance
{ "target_language": "Java", "developer_needs": [ "evaluation_metrics", "documentation", "repo_scale_reasoning" ] }
train_00294
2026-01-01T00:00:00
Code-specialized model families and sizing tradeoffs
compare
expert
Task: compare Topic: Code-specialized model families and sizing tradeoffs Difficulty: expert Target language: JavaScript Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "JavaScript", "developer_needs": [ "tooling", "ci_integration", "documentation" ] }
train_00295
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
compare
expert
Task: compare Topic: Reasoning-first coding models and tunable deliberation Difficulty: expert Target language: C# Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "C#", "developer_needs": [ "documentation", "cost_latency_tradeoffs", "security_gates" ] }
train_00296
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
design
expert
Task: design Topic: Reasoning-first coding models and tunable deliberation Difficulty: expert Target language: SQL Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "SQL", "developer_needs": [ "tooling", "documentation", "evaluation_metrics" ] }
train_00297
2026-01-01T00:00:00
Agentic coding systems (plan→edit→test→reflect)
code
advanced
Task: code Topic: Agentic coding systems (plan→edit→test→reflect) Difficulty: advanced Target language: C# Context: Fix a failing issue with tests as the oracle and produce a safe patch. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Design guidance with risks, metrics, acceptance criteria
{ "target_language": "C#", "developer_needs": [ "cost_latency_tradeoffs", "repo_scale_reasoning", "documentation" ] }
train_00298
2026-01-01T00:00:00
Tool calling, sandboxes, and CI integration
data_pipeline
foundation
Task: data_pipeline Topic: Tool calling, sandboxes, and CI integration Difficulty: foundation Target language: Java Context: Design a data pipeline for continued pretraining with auditability. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Pipeline: 1) Ingest 2) Normalize 3) Filter 4) Dedupe 5) Quality score 6) Sample 7) Audit
{ "target_language": "Java", "developer_needs": [ "repo_scale_reasoning", "documentation", "governance" ] }
train_00299
2026-01-01T00:00:00
Reasoning-first coding models and tunable deliberation
compare
foundation
Task: compare Topic: Reasoning-first coding models and tunable deliberation Difficulty: foundation Target language: SQL Context: Evaluate two coding models for internal rollout under strict governance. Deliver production-grade guidance or artifacts.
Key facts: - Modern AI coding emphasizes correctness via tests, agentic loops, and real-repo evaluation. - Reasoning-first and MoE approaches improve capability-per-compute when paired with tools. Compare: capability, cost, latency, reliability, governance
{ "target_language": "SQL", "developer_needs": [ "tests_are_truth", "tooling", "reproducibility" ] }