소스 정보
- 저장소
- jeremylongshore/oss-agent-lab
- 최근 소스 활동
- 2026년 3월 17일 05:36
- 감지된 SKILL.md 언어
- 영어
- 스타
- 4
- 포크
- 0
설치 방법
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
소스 파일 검토
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
메뉴
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/jeremylongshore/oss-agent-lab --skill deer-flow명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
SKILL.md 표시 중
Self-improving research loops with hypothesis generation, experiment design, and result analysis
Natural language web UI control — element detection, targeted interaction, and automated form filling
Public opinion analysis and sentiment at scale — sentiment scoring, stance detection, and multi-dimensional bias measurement.
SOC 직업 분류 기준
| name | deer_flow |
| display_name | Deer Flow Specialist |
| description | Full-stack research and code generation pipeline - research, code, and create |
| version | 0.1.0 |
| source_repo | bytedance/deer-flow |
| license | MIT |
| tier | core |
| capabilities | ["research","code_generation","creation","summarize"] |
| allowed_tools | ["research_topic","generate_code","create_artifact"] |
| output_formats | ["python_api","cli","mcp_server","agent_skill","rest_api"] |
The Deer Flow specialist wraps the bytedance/deer-flow SuperAgent pattern into a composable OSS Agent Lab specialist. It chains three stages into a single pipeline:
Each stage is also independently callable as a tool, making the specialist useful for partial workflows (research-only, code-only, wrap-existing-content-in-artifact).
shallow, standard, and deep depth settings.| Tool | Description | Parameters | Side Effects |
|---|---|---|---|
research_topic | Research a topic and return findings, sources, summary, confidence | topic, depth, sources | None |
generate_code | Generate code, tests, and explanation from a specification | specification, language, style | None |
create_artifact | Package content into a versioned artifact with metadata | content, artifact_type, format | None |
research_topic
topic: str — subject to research (required)depth: str — "shallow" | "standard" | "deep" (default: "standard")sources: list[str] | None — explicit source list; auto-selected when Nonegenerate_code
specification: str — natural-language description of the code to produce (required)language: str — target language, e.g. "python", "typescript" (default: "python")style: str — "clean" | "verbose" | "minimal" (default: "clean")create_artifact
content: dict[str, Any] — pipeline outputs to embed (required)artifact_type: str — "report" | "notebook" | "package" | "summary" (default: "report")format: str — "markdown" | "json" | "html" (default: "markdown")SpecialistRequest
│
▼
research_topic(topic, depth, sources)
│
▼ (if code generation needed)
generate_code(specification, language, style)
│
▼
create_artifact(content, artifact_type, format)
│
▼
SpecialistResponse(result={research, code?, artifact})
Code generation is triggered when:
"code""code_generation"generate_code is truthy (default: True)from agents.specialists.deer_flow.agent import DeerFlowSpecialist
from oss_agent_lab.contracts import Intent, Query, SpecialistRequest
specialist = DeerFlowSpecialist()
request = SpecialistRequest(
intent=Intent(
action="research_and_build",
domain="code_generation",
confidence=0.95,
parameters={"depth": "deep", "language": "python"},
),
query=Query(user_input="async rate limiter with token bucket algorithm"),
specialist_name="deer_flow",
)
response = await specialist.execute(request)
print(response.result["artifact"]["artifact_id"])
print(response.result["code"]["code"])
oss-lab run deer_flow "async rate limiter with token bucket algorithm"
from agents.specialists.deer_flow.tools import research_topic
findings = research_topic(
topic="transformer attention mechanisms",
depth="deep",
sources=["arxiv", "github"],
)
print(findings["summary"])
print(f"Confidence: {findings['confidence']}")
from agents.specialists.deer_flow.tools import generate_code
result = generate_code(
specification="LRU cache with O(1) get and put operations",
language="python",
style="clean",
)
print(result["code"])
print(result["tests"])
Wraps bytedance/deer-flow — a full-stack multi-agent research framework featuring deep research, report generation, and podcast/presentation creation pipelines built on top of LangGraph.