362bf8d810
Extracted from: https://github.com/jwwelbor/AgentMap.git Score: 1.0
26 lines
1.5 KiB
JSON
26 lines
1.5 KiB
JSON
{
|
|
"name": "conditional-request-routing",
|
|
"version": "1.0.0",
|
|
"goal": "Classify an input request and route it to either an approval or rejection path, producing a corresponding result",
|
|
"inputs": [
|
|
"request: the input request or content to be reviewed and routed"
|
|
],
|
|
"steps": [
|
|
"Collect input: Use an input agent to capture the user's request into state field 'request'",
|
|
"Branch: Use a branching agent to evaluate 'request' and route to 'Approve' node on success or 'Reject' node on failure",
|
|
"Approve path: Use a default agent to format and output an approval message with the request",
|
|
"Reject path: Use a default agent to format and output a rejection message with the request"
|
|
],
|
|
"outputs": [
|
|
"result: the approval or rejection message containing the original request"
|
|
],
|
|
"failure_modes": [
|
|
"Branching agent cannot evaluate request and neither path is taken",
|
|
"Missing or empty input request",
|
|
"Prompt misconfiguration causing incorrect routing"
|
|
],
|
|
"confidence": 0.85,
|
|
"explanation": "Extracted from AgentMap's README example 'ReviewFlow' CSV workflow. This is a declarative LangGraph pattern using AgentMap's CSV format that implements conditional branching - a universally reusable pattern for request triage, content moderation, approval gates, or any two-path decision flow. It can be adapted by changing agent types (e.g., using LLM agents instead of default) and prompts.",
|
|
"source_repo": "https://github.com/jwwelbor/AgentMap.git",
|
|
"score": 1.0
|
|
} |