Projects

A project bundles standing instructions and uploaded files into a named, reusable context. Reference a project from an agent run or a conversation, and that context is injected automatically.

Create a project

POST/v1/workspaces/{workspace_id}/projects —

fields: name (required), description, instructions (the standing guidance applied whenever the project is in scope).

bash

curl https://nexevo.ai/v1/workspaces/$NEXEVO_WORKSPACE/projects \
  -H "Authorization: Bearer $NEXEVO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Acme onboarding",
    "description": "Support context for the Acme account",
    "instructions": "Always cite the contract section. Currency is USD."
  }'

Files

POST/v1/workspaces/{workspace_id}/projects/{project_id}/files —

upload a file (multipart) into the project.

GET/v1/workspaces/{workspace_id}/projects/{project_id}/files —

list the project's files.

GET/v1/workspaces/{workspace_id}/projects/{project_id}/context —

the assembled context (instructions + file digest) the agent receives.

Using a project

Pass project_id when starting an agent run or creating a conversation — the project's instructions and files are injected server-side, so the agent reasons with that context without you re-sending it each call.

bash

curl -N https://nexevo.ai/v1/workspaces/$NEXEVO_WORKSPACE/agent/runs \
  -H "Authorization: Bearer $NEXEVO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"task": "Draft the renewal email", "project_id": "PROJECT_UUID"}'

Manage

GET/v1/workspaces/{workspace_id}/projects —

PATCH/v1/workspaces/{workspace_id}/projects/{project_id} —

DELETE/v1/workspaces/{workspace_id}/projects/{project_id} —

list, update, and delete projects.

Projects are for STANDING context you reuse. For one-off facts the workspace should remember across all tasks, use Memory instead.