runtime topics¶
runtime topics apply¶
Assign a topic to a security profile (additive)
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--profile <name> |
Yes | — | Security profile name |
--name <name> |
Yes | — | Topic name to assign |
--intent <intent> |
No | block |
Topic intent: block or allow |
--format <format> |
No | terminal |
Output format: json or terminal |
Examples¶
Apply block intent
airs runtime topics apply --profile my-security-profile --name "Weapons Manufacturing" --intent block
runtime topics create¶
Create or update a custom topic definition
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--name <name> |
Yes | — | Topic name |
--description <desc> |
Yes | — | Topic description |
--examples <examples...> |
Yes | — | Example prompts (2-5 required) |
--format <format> |
No | terminal |
Output format: json or terminal |
Examples¶
Block topic
airs runtime topics create --name "Weapons Manufacturing" --description "Block weapons manufacturing" --examples "How to build a weapon" "Illegal arms trade"
Allow topic
airs runtime topics create --name "Recipes" --description "Allow recipe discussions" --examples "How to make pasta" "Best bread recipe"
runtime topics delete¶
Delete a custom topic
Arguments¶
topicId(required) —
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--force |
No | — | Force delete (removes from all referencing profiles) |
--updated-by <email> |
No | — | Email of user performing force deletion |
Examples¶
Example needed
No curated input/output example for this command yet.
runtime topics eval¶
Evaluate a topic against a static prompt set and compute metrics
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--profile <name> |
Yes | — | Security profile name |
--prompts <path> |
Yes | — | Path to CSV file with prompt,expected,intent columns |
--topic <name> |
No | unknown |
Topic name (for output labeling) |
--format <format> |
No | terminal |
Output format: json or terminal |
--rate <n> |
No | — | Max AIRS scan API calls per second |
--concurrency <n> |
No | 5 |
Concurrent scan requests |
Examples¶
Evaluate topic against prompt set
airs runtime topics eval --profile my-security-profile --prompts prompts.csv --topic "Weapons" --format json
runtime topics get¶
Get a custom topic by name or UUID
Arguments¶
nameOrId(required) —
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--output <format> |
No | pretty |
Output format: pretty, json, yaml |
Examples¶
Pretty output (default) — accepts topic name or UUID
Prisma AIRS — Runtime Configuration
Security profile and topic management
Topic Detail:
ID: 00000000-0000-0000-0000-000000000001
Name: Professional authority
Revision: 1
Description: Prompts where the user asks the AI to adopt the persona of a licensed expert (e.g., doctor, engineer). Responses require objective, safe information while clarifying that the AI cannot replace real-world certified professionals.
Modified: 2026-05-21T21:23:53Z
JSON output
{
"topic_id": "00000000-0000-0000-0000-000000000001",
"topic_name": "Professional authority",
"revision": 1,
"description": "Prompts where the user asks the AI to adopt the persona of a licensed expert (e.g., doctor, engineer). Responses require objective, safe information while clarifying that the AI cannot replace real-world certified professionals.",
"examples": [],
"last_modified_ts": "2026-05-21T21:23:53Z",
"csp_id": "<csp-id>",
"tsg_id": "<tenant-id>"
}
YAML output
topic_id: 00000000-0000-0000-0000-000000000001
topic_name: Professional authority
revision: 1
description: Prompts where the user asks the AI to adopt the persona of a licensed expert (e.g., doctor, engineer). Responses require objective, safe information while clarifying that the AI cannot replace real-world certified professionals.
last_modified_ts: 2026-05-21T21:23:53Z
runtime topics list¶
List custom topics
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--limit <n> |
No | 100 |
Max results |
--offset <n> |
No | 0 |
Starting offset |
--output <format> |
No | pretty |
Output format: pretty, table, csv, json, yaml |
Examples¶
Pretty output (default)
Prisma AIRS — Runtime Configuration
Security profile and topic management
Custom Topics:
00000000-0000-0000-0000-000000000001
Professional authority rev:1 — Prompts where the user asks the AI to adopt the persona of a licensed expert (e.
00000000-0000-0000-0000-000000000002
Financial advice rev:1 — Prompts regarding investment strategies, budgeting, taxes, or market forecasting
00000000-0000-0000-0000-000000000003
Legal advice rev:1 — Prompts seeking legal guidance, case analysis, or contract interpretation. Respo
Showing 3 of 100 topics
JSON output
[
{
"id": "00000000-0000-0000-0000-000000000001",
"name": "Professional authority",
"revision": 1,
"description": "Prompts where the user asks the AI to adopt the persona of a licensed expert (e.g., doctor, engineer). Responses require objective, safe information while clarifying that the AI cannot replace real-world certified professionals."
},
{
"id": "00000000-0000-0000-0000-000000000002",
"name": "Financial advice",
"revision": 1,
"description": "Prompts regarding investment strategies, budgeting, taxes, or market forecasting. Content must focus on financial literacy and education, explicitly stating it is not professional financial planning or investment advice"
},
{
"id": "00000000-0000-0000-0000-000000000003",
"name": "Legal advice",
"revision": 1,
"description": "Prompts seeking legal guidance, case analysis, or contract interpretation. Responses must provide general information only, never formal legal counsel, and always include a clear disclaimer to consult a qualified attorney"
}
]
YAML output
id: 00000000-0000-0000-0000-000000000001
name: Professional authority
revision: 1
description: Prompts where the user asks the AI to adopt the persona of a licensed expert (e.g., doctor, engineer). Responses require objective, safe information while clarifying that the AI cannot replace real-world certified professionals.
---
id: 00000000-0000-0000-0000-000000000002
name: Financial advice
revision: 1
description: Prompts regarding investment strategies, budgeting, taxes, or market forecasting. Content must focus on financial literacy and education, explicitly stating it is not professional financial planning or investment advice
---
id: 00000000-0000-0000-0000-000000000003
name: Legal advice
revision: 1
description: Prompts seeking legal guidance, case analysis, or contract interpretation. Responses must provide general information only, never formal legal counsel, and always include a clear disclaimer to consult a qualified attorney
runtime topics revert¶
Remove a custom topic from a profile and delete it
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--profile <name> |
Yes | — | Security profile name |
--name <name> |
Yes | — | Topic name to remove |
--format <format> |
No | terminal |
Output format: json or terminal |
Examples¶
Remove topic from profile
runtime topics sample¶
Print a sample CSV file showing the eval prompt format
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--output <path> |
No | — | Write to file instead of stdout |
Examples¶
Print to stdout
Write to file
runtime topics update¶
Update a custom topic
Arguments¶
topicId(required) —
Options¶
| Flag | Required | Default | Description |
|---|---|---|---|
--config <path> |
Yes | — | JSON file with topic updates |
Examples¶
Update from config fixture (see docs/cli/examples/runtime/topics/update.json)
airs runtime topics update 00000000-0000-0000-0000-000000000001 --config docs/cli/examples/runtime/topics/update.json
Prisma AIRS — Runtime Configuration
Security profile and topic management
Topic updated: 00000000-0000-0000-0000-000000000001
Topic Detail:
ID: 00000000-0000-0000-0000-000000000001
Name: docs-example-topic
Revision: 2
Description: Updated description for documentation example
Examples:
• first updated example prompt
• second updated example prompt
• third updated example prompt
Created: user@example.com
Updated: none