agents-and-societies
Build CAMEL ChatAgent workflows, BaseMessage setup, role-playing societies, Workforce orchestration, Task objects, and safe agent-loop debugging.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- agents-and-societies
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Build CAMEL ChatAgent workflows, BaseMessage setup, role-playing societies, Workforce orchestration, Task objects, and safe agent-loop debugging.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/api-reference.md10.7 KB
- 参考文档references/troubleshooting.md8.3 KB
- 参考文档references/workflows.md10.5 KB
03 SKILL.md 正文
Agents and Societies
Use this sub-skill when the user needs to assemble CAMEL-AI agents or multi-agent workflows: ChatAgent setup, system messages, tool-enabled agents at a routing level, RolePlaying societies, Workforce orchestration, Task objects, response terminators, callbacks, memory hooks, and loop debugging.
Start Here
- Read references/api-reference.md to choose the right CAMEL classes and constructor options.
- Read references/workflows.md for practical recipes: single agent, role-playing society, workforce, pipeline mode, and CI dry-run validation.
- Run scripts/inspect_agent_basics.py to inspect installed CAMEL imports, signatures, message constructors, and dry-run object setup without model credentials.
- Read references/troubleshooting.md before debugging loops, missing API keys, termination, timeouts, callbacks, memory state, or task failures.
Route By Task
- Single conversational agent: use
ChatAgentwith a string orBaseMessagesystem message, optional memory/window/token controls, and optional response terminators. - Message construction: use
BaseMessage.make_system_message,make_assistant_message, andmake_user_messagefor explicit roles, metadata, and multimodal attachments. - Tool-enabled agent design: attach
tools,external_tools, timeouts, and masking atChatAgentconstruction, then defer schema/toolkit details to../tools-runtimes-and-services/. - Two-role society: use
RolePlayingwhen an AI user and AI assistant should alternate turns, optionally with task specification, task planning, a critic, or pre-builtChatAgentinstances. - Team orchestration: use
Workforcewhen tasks need decomposition, assignment to workers, failure recovery, streaming callbacks, shared memory, or pipeline-mode dependencies. - Task state and decomposition: use
Taskfor structured objectives, subtasks, dependencies, result propagation, and Workforce entry points.
Boundaries And Cross-Links
- For provider credentials,
ModelFactory,BaseModelBackend, model enums, and backend-specific configuration, use../models-and-configuration/. - For
FunctionTool,MCPToolkit, schema synthesis, sandboxed execution, and external services, use../tools-runtimes-and-services/. - For
ChatHistoryMemory, retrieval-backed memory, vector stores, and embedding internals, use../memory-rag-and-data/. - For benchmark/data-generation recipes, use the sibling benchmark or datagen sub-skill if present; this sub-skill only covers agent workflow routing.
Safety Defaults
- Prefer credential-free dry runs first: instantiate messages, tasks, agents with explicit local/dummy model planning, and inspect signatures before calling
.step(). - Always bound loops with
max_iteration,response_terminators, role-play turn limits, ortask_timeout_secondsbefore running in CI. - Keep tool execution bounded with
tool_execution_timeout,mask_tool_output, and targeted tools; route tool schema failures to the tools sub-skill. - Treat
share_memory=True, custom memory, and callbacks as workflow state surfaces that require explicit reset and logging plans.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.