szaaaaaa
24 verified skills72 total stars
src/dynamic_os/skills/builtins/optimize_experiment
Evaluate `ExperimentResults` and emit an `ExperimentIteration` with optimization suggestions. This skill uses `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/reflect_on_failure
Analyze a failed skill execution to determine root cause and suggest fixes. The planner populates `ctx.goal` with failure context (failed skill ID, error message, observation details). This skill reads the failed skill's source code and uses LLM analysis to produce a `ReflectionReport` artifact containing root cause analysis and suggested fixes.
development3
src/dynamic_os/skills/builtins/review_artifact
Review an existing artifact and return a `ReviewVerdict`. This skill merges the earlier critique wrappers into one generic reviewer skill.
tools3
tests/fixtures/phase1_skills/restricted/network_probe
This fixture skill attempts a network call and should be blocked by its own permissions.
testing3
src/dynamic_os/skills/builtins/clarify_intent
Clarify the user's intent into one of three tiers: a confidently inferred goal, a default-filled goal with explicit assumptions, or a structured reask with option-style questions. The skill uses `ctx.tools.llm_chat()` with a structured JSON schema and emits either `ClarifiedIntent` or `ClarificationRequest`. When previous-round `ClarificationResponse` artifacts are present in `ctx.input_artifacts`, they are folded into the prompt so the LLM can incorporate prior answers.
tools3
src/dynamic_os/skills/builtins/plan_research
Produce a minimal topic brief and search plan from the current goal. This skill uses `ctx.tools.llm_chat()` and returns `TopicBrief` plus `SearchPlan`.
tools3
scripts/demo_assets/broken_skill
# extract_keywords_v0 (broken baseline) Demo asset for the `reflect_on_failure → optimize_skill` chain. See `run.py` for the deliberate bug.
data-ai3
src/dynamic_os/skills/builtins/run_experiment
Execute an `ExperimentPlan` and return `ExperimentResults`. This skill uses `ctx.tools.execute_code()`.
tools3
src/dynamic_os/skills/builtins/search_papers
Search for papers from a `SearchPlan` and return a `SourceSet`. This skill uses `ctx.tools.search()`.
tools3
src/dynamic_os/skills/builtins/generate_figures
模板驱动的图表生成技能。LLM 分析输入 artifact 数据并输出 JSON 图表规格， 由内置模板渲染器（matplotlib / networkx / mermaid-cli）生成 PDF + PNG 图表， 输出 `FigureSet` 供 `draft_report` 嵌入论文。 支持的图表类型：bar_chart, grouped_bar_chart, line_chart, scatter_plot, heatmap, pie_chart, network_graph, flowchart, sequence_diagram, class_diagram, timeline。 使用 `ctx.tools.llm_chat()` 获取图表规格。
tools3
src/dynamic_os/skills/builtins/extract_notes
Convert a `SourceSet` into structured `PaperNotes`. This skill uses `ctx.tools.index()` and `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/draft_report
Draft the final `ResearchReport` from existing artifacts. This skill uses `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/fetch_fulltext
Retrieve fuller text material for a `SourceSet`. This skill uses `ctx.tools.retrieve()`.
tools3
src/dynamic_os/skills/builtins/design_experiment
Design a bounded experiment and return an `ExperimentPlan`. This skill uses `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/compare_methods
Build a structured comparison of methods from `PaperNotes`, `EvidenceMap`, or `ExperimentResults` and produce a `MethodComparison`. This skill uses `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/analyze_trends
Analyze temporal research trends from `PaperNotes`, `SourceSet`, or `EvidenceMap` and produce a `TrendAnalysis`. This skill uses `ctx.tools.llm_chat()`.
tools3
src/dynamic_os/skills/builtins/build_evidence_map
Synthesize existing artifacts into `EvidenceMap` and `GapMap`. This skill uses `ctx.tools.llm_chat()`.
tools3
tests/fixtures/phase1_skills/invalid/broken_skill
Broken skill fixture.
testing3
src/dynamic_os/skills/builtins/create_skill
Generate a new skill from a task description when no existing skill fits the need. Uses LLM to produce all three required files (skill.yaml, run.py, skill.md), validates them, and writes to `evolved_skills/`. Optionally uses a `ReflectionReport` as context for why existing skills are insufficient.
testing3
src/dynamic_os/skills/builtins/analyze_metrics
Analyze `ExperimentResults` and produce `ExperimentAnalysis` plus `PerformanceMetrics`. This skill uses `ctx.tools.llm_chat()`.
tools3
tests/fixtures/phase1_skills/disallowed/custom_search
Custom search fixture.
testing3
tests/fixtures/phase1_skills/valid/search_papers
Use this skill to search papers.
testing3
src/dynamic_os/skills/builtins/aggregate_results
# aggregate_results 聚合多轮实验结果，生成跨迭代对比分析。 ## 功能 - 从所有 ExperimentResults 和 ExperimentIteration 产物中提取指标历史 - 构建迭代对比表（每轮指标 + 成功/失败状态） - 计算各指标的 min/max/avg/best 统计值 - 提取最佳配置和经验教训 - 调用 LLM 生成结构化分析文本 ## 适用场景 当实验循环结束（ExperimentIteration.should_continue=false）且存在多轮实验数据时， 由 analyst 角色调用此技能生成统一的分析报告，供 draft_report 使用。
data-ai3
src/dynamic_os/skills/builtins/optimize_skill
Generate an improved version of a failed skill based on a `ReflectionReport`. Reads the original skill source, sends it with the reflection analysis to LLM, validates the generated code, and writes the evolved skill to `evolved_skills/`. The evolved skill gets a `_evolved` suffix on its ID to avoid registry conflicts.
development3