Govern your Slack bot with Talon

Route your Slack bot's LLM calls through Talon to cap what the bot can spend, block PII before it leaves your infrastructure, and get a signed audit trail of every request — all with a single config change. The same evidence records support compliance reviews when you need them. Allow about 10 minutes.

Prerequisites

• Talon installed and running with the gateway enabled
• A Slack bot that uses the OpenAI API (or another supported provider)
• Your real provider API key stored in Talon's vault

Steps

1. Install and start Talon with the gateway

talon serve --gateway --gateway-config=talon.config.yaml

Ensure your config includes a gateway block and that the bot's agent file carries a key binding (#266).

2. Store the real API key in Talon

talon secrets set openai-api-key

3. Point the Slack bot at the gateway

Most Slack bots use an environment variable or config for the API base URL. Set it to Talon's gateway:

OpenAI:

export OPENAI_BASE_URL=http://talon:8080/v1/proxy/openai/v1

If the bot runs on the same host as Talon:

export OPENAI_BASE_URL=http://localhost:8080/v1/proxy/openai/v1

Anthropic:

export ANTHROPIC_BASE_URL=http://talon:8080/v1/proxy/anthropic

No code changes are required — the bot still uses the same SDK; only the base URL changes.

4. Mint the bot's agent key

The Slack bot is one AI use case — one agent.talon.yaml with a vault-bound traffic key (#266). Mint it and configure the bot to send the VALUE as Authorization: Bearer <value> (or x-api-key for Anthropic):

talon secrets set support-slack-bot-talon-key "$(openssl rand -hex 24)"

Talon resolves that key to the agent (and its derived tenant), then uses the vault-stored provider key when calling the real provider.

5. Verify

Trigger the bot in Slack, then:

talon audit list

You should see gateway evidence for the bot's requests.

6. Add per-bot limits (optional)

In the bot's agent file, add overrides:

# support-slack-bot/agent.talon.yaml
agent:
  name: support-slack-bot
  version: 1.0.0
  key:
    secret_name: "support-slack-bot-talon-key"
policies:
  cost_limits:
    daily: 10.00
  models:
    allowed: ["gpt-4o-mini"]
  allowed_providers: ["openai"]
  data_classification:
    input_scan: true
    block_on_pii: true

Restart Talon after editing the config.

Summary

Item	Value
Gateway URL	http://<talon>:8080/v1/proxy/openai/v1 (trailing /v1 for correct paths; or anthropic/v1)
Auth	The bot's agent key (agent.talon.yaml + vault binding, #266)
Audit	talon audit list

---

You're done

You now have your Slack bot sending LLM calls through Talon. Talon is logging every request, scanning for PII, and applying per-bot limits you configured.

Next steps:

I want to…	Doc
Cap cost or restrict models for this bot	How to cap daily spend per team or application
Add Talon to another app (e.g. script or API)	Add Talon to your existing app
Export evidence for auditors	How to export evidence for auditors
Understand the request lifecycle	What Talon does to your request