Use when migrating an existing LangChain, LangGraph, LlamaIndex, or CrewAI system into Agently and mapping each source behavior to its actual request, runtime, context/storage, or TriggerFlow owner.
71
86%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Passed
No findings from the security scan
Use this Skill only when a source-framework system already exists. Do not translate class names one for one. Inventory the source's observable semantics, assign each to an Agently owner, and preserve those semantics with representative before/after evidence.
| Source behavior | Agently owner |
|---|---|
| One model call, prompt contract, structured result, or response stream | agently-request / ModelRequest |
| Model-callable tool, MCP server, sandbox, browser, database, or live client | agently-runtime / ActionRuntime + ExecutionResource |
| Conversation continuity | Session / SessionMemory |
| Durable records, links, retrieval, checkpoints, or snapshots | RecordStore |
| Cross-source task evidence and bounded reads | TaskContext + ContextSource + ContextReader |
| Stable multi-stage topology, branches, joins, loops, waits, resume, or runtime events | agently-triggerflow / TriggerFlow |
| Model-owned bounded planning, evidence, verification, and replan | AgentTask through AgentExecution |
Do not add TriggerFlow around a single request or Action. Do not flatten a real workflow into one prompt. A source “agent” is not automatically an Agently AgentTask, and a nominal source “workflow” is not automatically orchestration; classify its behavior.
5dc8d96
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.