Catalog Tools

Track every tool an AI agent can invoke — with type, owner, exposing server, risk and approval status.

The Tool Catalog is the inventory of everything an agent can actually do: functions, APIs, MCP tools, retrieval, and — the dangerous ones — code execution, shell and browser access.

1. Catalog a high-risk tool#

curl -X POST "https://your-company.authspoke.com/api/v1/ai/tools" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{
    "name": "execute_sql",
    "toolType": "CODE_EXEC",
    "owner": "[email protected]",
    "exposedBy": "internal-data-lake-mcp",
    "riskLevel": "HIGH",
    "status": "PENDING",
    "description": "Runs arbitrary SQL against the data lake."
  }'

toolType — FUNCTION, API, MCP_TOOL, RETRIEVAL, CODE_EXEC, SHELL, BROWSER. riskLevel — LOW, MEDIUM, HIGH. Registering a HIGH tool emits a WARNING event. status — APPROVED, UNAPPROVED, BLOCKED, PENDING. exposedBy — the MCP server or agent that surfaces this tool (feeds the Relationships graph).

2. Catalog lower-risk tools#

curl -X POST "https://your-company.authspoke.com/api/v1/ai/tools" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{ "name":"web_search", "toolType":"RETRIEVAL", "owner":"[email protected]", "riskLevel":"LOW", "status":"APPROVED" }'

3. List & maintain#

curl -s "https://your-company.authspoke.com/api/v1/ai/tools" -H "Authorization: Bearer $TOKEN"
curl -s -X PUT "https://your-company.authspoke.com/api/v1/ai/tools/$ID" -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" -d '{ "name":"execute_sql", "status":"BLOCKED", "riskLevel":"HIGH" }'
curl -s -X DELETE "https://your-company.authspoke.com/api/v1/ai/tools/$ID" -H "Authorization: Bearer $TOKEN"

Govern tools with policy#

Pair the catalog with a policy: a DENY on scopeType: TOOL, action: INVOKE_TOOL blocks tool use by default, then scope ALLOW policies to the specific approved tools. See Govern with Policies.

Console#

Enterprise AI → Tools lists every tool with its type, owner, exposing server, risk and status.