[codex] Raise MiniMax instant admit threshold#544
Conversation
Greptile SummaryThis PR bumps the Confidence Score: 5/5Safe to merge — single numeric constant change with no logic impact. The only finding is a P2 suggestion (env-var–driven capacity) that doesn't affect correctness. No P0/P1 issues were found. No files require special attention.
|
| Filename | Overview |
|---|---|
| web/src/server/free-session/config.ts | Raises minimax/minimax-m2.7 instant-admit capacity from 200 → 1000; all other logic is unchanged. |
Prompt To Fix All With AI
This is a comment left during a code review.
Path: web/src/server/free-session/config.ts
Line: 50-53
Comment:
**Consider env-var–driven capacity for zero-deploy tuning**
The rest of the file already reads tuneable knobs from `env` (e.g. `FREEBUFF_SESSION_LENGTH_MS`, `FREEBUFF_SESSION_GRACE_MS`). Driving `INSTANT_ADMIT_CAPACITY` the same way would let you raise or lower the threshold without a code deploy — the same benefit the PR description cites for keeping this server-side in the first place.
```suggestion
const INSTANT_ADMIT_CAPACITY: Record<string, number> = {
'z-ai/glm-5.1': env.FREEBUFF_INSTANT_ADMIT_ZAI_GLM ?? 50,
'minimax/minimax-m2.7': env.FREEBUFF_INSTANT_ADMIT_MINIMAX_M2_7 ?? 1000,
}
```
How can I resolve this? If you propose a fix, please make it concise.Reviews (1): Last reviewed commit: "Raise MiniMax instant admit threshold" | Re-trigger Greptile
| const INSTANT_ADMIT_CAPACITY: Record<string, number> = { | ||
| 'z-ai/glm-5.1': 50, | ||
| 'minimax/minimax-m2.7': 200, | ||
| 'minimax/minimax-m2.7': 1000, | ||
| } |
There was a problem hiding this comment.
Consider env-var–driven capacity for zero-deploy tuning
The rest of the file already reads tuneable knobs from env (e.g. FREEBUFF_SESSION_LENGTH_MS, FREEBUFF_SESSION_GRACE_MS). Driving INSTANT_ADMIT_CAPACITY the same way would let you raise or lower the threshold without a code deploy — the same benefit the PR description cites for keeping this server-side in the first place.
| const INSTANT_ADMIT_CAPACITY: Record<string, number> = { | |
| 'z-ai/glm-5.1': 50, | |
| 'minimax/minimax-m2.7': 200, | |
| 'minimax/minimax-m2.7': 1000, | |
| } | |
| const INSTANT_ADMIT_CAPACITY: Record<string, number> = { | |
| 'z-ai/glm-5.1': env.FREEBUFF_INSTANT_ADMIT_ZAI_GLM ?? 50, | |
| 'minimax/minimax-m2.7': env.FREEBUFF_INSTANT_ADMIT_MINIMAX_M2_7 ?? 1000, | |
| } |
Prompt To Fix With AI
This is a comment left during a code review.
Path: web/src/server/free-session/config.ts
Line: 50-53
Comment:
**Consider env-var–driven capacity for zero-deploy tuning**
The rest of the file already reads tuneable knobs from `env` (e.g. `FREEBUFF_SESSION_LENGTH_MS`, `FREEBUFF_SESSION_GRACE_MS`). Driving `INSTANT_ADMIT_CAPACITY` the same way would let you raise or lower the threshold without a code deploy — the same benefit the PR description cites for keeping this server-side in the first place.
```suggestion
const INSTANT_ADMIT_CAPACITY: Record<string, number> = {
'z-ai/glm-5.1': env.FREEBUFF_INSTANT_ADMIT_ZAI_GLM ?? 50,
'minimax/minimax-m2.7': env.FREEBUFF_INSTANT_ADMIT_MINIMAX_M2_7 ?? 1000,
}
```
How can I resolve this? If you propose a fix, please make it concise.
Raises the freebuff MiniMax M2.7 instant-admit capacity from 200 to 1000. This lets more MiniMax users skip the waiting room before FIFO backpressure starts. The change is limited to the server-side free-session config knob, so CLI-facing shared model definitions are unchanged. Validated with
bun test web/src/server/free-session/__tests__/config.test.tsusing dummy required env vars for config initialization.