feat(agent): route slack-originated tasks to slack_app gateway product

[VojtechBartos]

Problem

The agent in the sandbox calls the LLM gateway as posthog_code regardless of how the task started. Slack-bot mentions are now billed differently from desktop tasks (gateway tags slack_app traffic with \$ai_billable=true), so the agent needs to send the correct product per task.

Changes

• Add "slack_app" to the GatewayProduct type union (packages/agent/src/utils/gateway.ts).
• Add "slack" to the Task.origin_product union (packages/agent/src/types.ts) to mirror the Python Task.OriginProduct.SLACK value already set on the backend.
• configureEnvironment now accepts originProduct?: Task["origin_product"] | null. Mapping: "slack" → slack_app, internal=true → background_agents, else → posthog_code. Call site reads preTask?.origin_product from the Task fetched at session init.
• Three new unit tests covering the slack → slack_app mapping, slack-over-internal precedence, and non-slack fallback.

Pairs with PostHog/posthog#59040 (gateway-side product registration + per-turn quota gate). Both ship behind the existing posthog-code-slack-availability flag — neither side has user-visible effect until the flag is enabled.

How did you test this?

I'm an agent. No manual testing.

• pnpm --filter @posthog/agent test src/server/agent-server.configure-environment.test.ts → 8 passed (5 pre-existing + 3 new).
• Typecheck error count unchanged from clean main (pre-existing module-resolution noise unrelated to this change).
• biome format --write clean on touched files.

Publish to changelog?

no

[greptile-apps]

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
packages/agent/src/server/agent-server.configure-environment.test.ts:92-127
**Prefer parameterised tests for product-routing cases**

The three new tests exercise the same `configureEnvironment` code path with different `(originProduct, isInternal, expectedProduct)` tuples — a textbook candidate for `it.each`. The codebase's own rule is to always prefer parameterised tests, and duplicating the arrange/assert boilerplate across three separate `it()` blocks makes it harder to see the full matrix at a glance and to add future cases. A single `it.each` table would also make it obvious that the `(originProduct="user_created", isInternal=true) → background_agents` cell is currently untested.

_{Reviews (1): Last reviewed commit: "feat(agent): route slack-originated task..." | Re-trigger Greptile}

[greptile-apps]

Prefer parameterised tests for product-routing cases

The three new tests exercise the same configureEnvironment code path with different (originProduct, isInternal, expectedProduct) tuples — a textbook candidate for it.each. The codebase's own rule is to always prefer parameterised tests, and duplicating the arrange/assert boilerplate across three separate it() blocks makes it harder to see the full matrix at a glance and to add future cases. A single it.each table would also make it obvious that the (originProduct="user_created", isInternal=true) → background_agents cell is currently untested.

Context Used: Do not attempt to comment on incorrect alphabetica... (source)

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/agent/src/server/agent-server.configure-environment.test.ts
Line: 92-127

Comment:
**Prefer parameterised tests for product-routing cases**

The three new tests exercise the same `configureEnvironment` code path with different `(originProduct, isInternal, expectedProduct)` tuples — a textbook candidate for `it.each`. The codebase's own rule is to always prefer parameterised tests, and duplicating the arrange/assert boilerplate across three separate `it()` blocks makes it harder to see the full matrix at a glance and to add future cases. A single `it.each` table would also make it obvious that the `(originProduct="user_created", isInternal=true) → background_agents` cell is currently untested.

**Context Used:** Do not attempt to comment on incorrect alphabetica... ([source](https://app.greptile.com/review/custom-context?memory=instruction-0))

How can I resolve this? If you propose a fix, please make it concise.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!