[changelog] [Announcement] Updated invocation limits

[nyxthedev]

GUID: CHANGE-3420
pubDate: Tue, 01 Sep 2026 03:27:49 GMT
Category: Announcement

What's changing

Atlassian is updating invocation rate limits for user-led invocations (UI and web triggers):

• Per app installation: 5,000 requests per minute (RPM) → 300 requests per second (RPS) or 7,000 RPM, whichever is hit first.
• Per environment: the previous global limit of 30,000 RPM is removed entirely.
• Per user: unchanged at 1,200 RPM.

Apps with high-burst activity may need retry logic, using the rateLimitProperties field returned from Forge bridge invoke, or web trigger rate-limiting response headers.

Impact on forge-sim parity

Simulator area: limits / platform behavior.

• Any documented or simulated invocation rate-limit values (per-installation 5,000 RPM, per-environment 30,000 RPM) are now stale — new numbers are 300 RPS / 7,000 RPM per installation, no per-environment cap.
• If forge-sim simulates rate limiting or 429 responses for invocations/web triggers, thresholds should reflect the new values. The dual-threshold semantics (burst RPS cap AND sustained RPM cap, whichever trips first) is a behavioral shape worth matching if limits are enforced.
• Also confirms two app-facing surfaces worth parity checks: the rateLimitProperties field on the bridge invoke response, and rate-limiting response headers on web triggers.

Link: https://developer.atlassian.com/platform/forge/changelog/#CHANGE-3420

• forge-spec: update affected requirement rows (behavior change)