Data Source Reference

All 8 read-only data sources. Use them to look up server-managed values (catalogs, statistics) without taking ownership of the underlying objects.

---

agentcy_agent_types

Catalog of valid agent_type values for agentcy_agent.

data "agentcy_agent_types" "all" {}

output "available_agents" {
  value = [for t in data.agentcy_agent_types.all.types : t.type]
}

Returns a list of { type, name, description } objects.

---

agentcy_permissions

RBAC permission strings recognized by agentcy_role. Useful for asserting that a role grants a specific known permission rather than a typo.

data "agentcy_permissions" "all" {}

resource "agentcy_role" "viewer" {
  name        = "viewer"
  permissions = [
    for p in data.agentcy_permissions.all.permissions :
    p if startswith(p, "graph:")
  ]
}

---

agentcy_realm_stats

Live node and relationship counts for a realm. Reads through to Neo4j, so it reflects the actual graph state.

data "agentcy_realm_stats" "engineering" {
  realm_id = agentcy_realm.engineering.id
}

output "engineering_size" {
  value = "${data.agentcy_realm_stats.engineering.total_nodes} nodes / ${data.agentcy_realm_stats.engineering.total_relationships} edges"
}

---

agentcy_running_tasks

Tasks currently executing. Useful for guard rails — e.g. don't redeploy while a long-running task is mid-flight.

data "agentcy_running_tasks" "now" {}

output "running" {
  value = length(data.agentcy_running_tasks.now.tasks)
}

---

agentcy_memory_search

Semantic search across the org's memory store. The query runs through the embedding model configured in agentcy_settings.embedding_model.

data "agentcy_memory_search" "house_style_hits" {
  query = "house style"
  top_k = 5
}

output "matches" {
  value = data.agentcy_memory_search.house_style_hits.results[*].title
}

---

agentcy_artifacts

The artifact registry — list-only, no body downloads. Useful for asserting that a known file exists, or for chaining a known artifact ID into another resource.

data "agentcy_artifacts" "all" {}

---

agentcy_conversations

Chat conversation metadata — IDs, titles, channel association, starred status. Bodies are excluded; this is metadata only.

data "agentcy_conversations" "all" {}

output "starred" {
  value = [
    for c in data.agentcy_conversations.all.conversations :
    c.title if c.is_starred
  ]
}

---

agentcy_pipeline_runs

History of ingestion pipeline runs. Each entry includes the source it ran against, the completion timestamp, and any error.

data "agentcy_pipeline_runs" "recent" {}

output "failed_recent" {
  value = [
    for r in data.agentcy_pipeline_runs.recent.runs :
    r if r.status == "failed"
  ]
}