Findings: a whole Free run on a new project (Scratch) #365

Closed
opened 2026-11-03 09:38:29 -05:00 by cmoriarty · 15 comments
Owner

The findings log for a whole Free run on production: thorough, a new multiplayer card game (scratch), with auto-approve on, started 2026-11-03 to see where Free stands. Findings arrive as comments and get their own issues where they need a fix.

So far (through the proposal's review)

  1. One request cost about 5 of the review's 6.5 minutes. agent.review.combined: Cloudflare's gemma-4-26b-a4b-it was cut at 300 s ("provider stream ended before completion: unexpected EOF"). Braid retried once and benched the model for 30 minutes, and the step succeeded on Qwen. The cut at almost exactly 300 s looks like a limit on the provider's side, not confirmed.
  2. One step moved between three providers while the first was answering. The same review went Qwen 397B, then north-mini-code, then Gemma, then back to Qwen. Why it left Qwen isn't known yet.
  3. A handled retry reads as a failure. The transcript shows a red provider.unknown line for something Braid dealt with itself.

Fine: agent.revise took 0.4 min on reflection/beam-free, and made the review's one finding (Depends on: lines) correctly. The proposal gate was auto-approved as set.

The findings log for a whole Free run on production: `thorough`, a new multiplayer card game (`scratch`), with auto-approve on, started 2026-11-03 to see where Free stands. Findings arrive as comments and get their own issues where they need a fix. **So far (through the proposal's review)** 1. **One request cost about 5 of the review's 6.5 minutes.** `agent.review.combined`: Cloudflare's `gemma-4-26b-a4b-it` was cut at 300 s ("provider stream ended before completion: unexpected EOF"). Braid retried once and benched the model for 30 minutes, and the step succeeded on Qwen. The cut at almost exactly 300 s looks like a limit on the provider's side, not confirmed. 2. **One step moved between three providers while the first was answering.** The same review went Qwen 397B, then `north-mini-code`, then Gemma, then back to Qwen. Why it left Qwen isn't known yet. 3. **A handled retry reads as a failure.** The transcript shows a red `provider.unknown` line for something Braid dealt with itself. **Fine:** `agent.revise` took 0.4 min on `reflection/beam-free`, and made the review's one finding (`Depends on:` lines) correctly. The proposal gate was auto-approved as set.
Author
Owner

More on finding 1, the 300 s cut. The request wasn't silent: its reasoning part arrived at the cut with about 50 KB in it, and no tool call or answer came. So Gemma spent the whole request thinking, and all of it was lost. A timer on silence wouldn't have caught this one.

Braid only recovers afterwards, by retrying and benching the model for 30 minutes. Nothing limits a request short of the router's 3,600 s cap. One sample, so the 300 s limit on Cloudflare's side is still unconfirmed.

More on finding 1, the 300 s cut. The request wasn't silent: its reasoning part arrived at the cut with about 50 KB in it, and no tool call or answer came. So Gemma spent the whole request thinking, and all of it was lost. A timer on silence wouldn't have caught this one. Braid only recovers afterwards, by retrying and benching the model for 30 minutes. Nothing limits a request short of the router's 3,600 s cap. One sample, so the 300 s limit on Cloudflare's side is still unconfirmed.
Author
Owner

Finding 1 now has its own ticket with the research: #366. The connection closed from the far side, not on Braid's 300 s timer, and the model was stuck in a thinking loop.

Finding 1 now has its own ticket with the research: #366. The connection closed from the far side, not on Braid's 300 s timer, and the model was stuck in a thinking loop.
Author
Owner

Build step, first 10 minutes: Kilo ran out, and the step kept going. 351 tool events so far.

Minutes in Served by
0-5 beam-free 61, Qwen 397B 4, Gemma 4
5-10 Gemma 22, Qwen 397B 10, Zen deepseek-v4.1-flash 6, none 3
10+ Zen deepseek-v4.1-flash 4 (still going)
  1. Kilo refused after 100 requests on this run, where the catalog expects 200 an hour. Its words: "Sign up for free to continue and explore 500 other models… Or come back later." Either Kilo's real limit is lower than catalog.py says, or something else counted against this IP. So Braid's own count thinks Kilo has room when it doesn't. That's probably why every later request tried Kilo first and hit its circuit breaker (39 times in this step, no time lost on each, but noise).
  2. A breaker reads as spent quota. Three requests got a bare 503 with no failover, and twice the step was parked with free quota spent for about 2 s. The quota wasn't spent; the provider's breaker was open. It cost seconds, but the label is wrong.
  3. The build changed models about every request. Gemma, Qwen and now DeepSeek take turns in one session. Whether that hurts the code is the thing to check when the step ends.
  4. Requests cancelled right after a failover. The router log shows context canceled (client_disconnected) on Qwen 1-2 s after a failover succeeded. Unknown yet whether opencode dropped them on purpose.
**Build step, first 10 minutes: Kilo ran out, and the step kept going.** 351 tool events so far. | Minutes in | Served by | |---|---| | 0-5 | `beam-free` 61, Qwen 397B 4, Gemma 4 | | 5-10 | Gemma 22, Qwen 397B 10, Zen `deepseek-v4.1-flash` 6, none 3 | | 10+ | Zen `deepseek-v4.1-flash` 4 (still going) | 4. **Kilo refused after 100 requests on this run, where the catalog expects 200 an hour.** Its words: "Sign up for free to continue and explore 500 other models… Or come back later." Either Kilo's real limit is lower than `catalog.py` says, or something else counted against this IP. So Braid's own count thinks Kilo has room when it doesn't. That's probably why every later request tried Kilo first and hit its circuit breaker (39 times in this step, no time lost on each, but noise). 5. **A breaker reads as spent quota.** Three requests got a bare 503 with no failover, and twice the step was parked with `free quota spent` for about 2 s. The quota wasn't spent; the provider's breaker was open. It cost seconds, but the label is wrong. 6. **The build changed models about every request.** Gemma, Qwen and now DeepSeek take turns in one session. Whether that hurts the code is the thing to check when the step ends. 7. **Requests cancelled right after a failover.** The router log shows `context canceled` (`client_disconnected`) on Qwen 1-2 s after a failover succeeded. Unknown yet whether opencode dropped them on purpose.
Author
Owner

Finding 4 (Kilo) now has its own ticket: #367.

  1. A provider change forced a compaction mid-build. openspec.apply[all] went from Qwen 397B to Zen's deepseek-v4.1-flash-free and was compacted from 64,309 to 20,519 tokens, reason provider_switch, with 10 tasks still to do. This is finding 6's cost: the next model starts from a summary, not from what the last one had read. Watching whether the work after it repeats or undoes anything.
Finding 4 (Kilo) now has its own ticket: #367. 8. **A provider change forced a compaction mid-build.** `openspec.apply[all]` went from Qwen 397B to Zen's `deepseek-v4.1-flash-free` and was compacted from 64,309 to 20,519 tokens, reason `provider_switch`, with 10 tasks still to do. This is finding 6's cost: the next model starts from a summary, not from what the last one had read. Watching whether the work after it repeats or undoes anything.
Author
Owner

The build step finished: 36.5 minutes, all 25 tasks ticked, no cuts. That's about what a similar build took on vLLM (36 minutes, run 147), though on a different project, so it's only a rough comparison.

It ran on six models: Zen's deepseek-v4.1-flash-free 184 requests, beam-free 61, Gemma 26, Qwen 397B 16, gpt-oss-120b 2, glm-4.5-flash 2. Zen carried most of it once Kilo refused, so finding 4 cost less time than feared. Whether the code holds together across those changes is for the tests and the Visual check to say.

  1. Zen requests have no recorded duration. All 184 show 0 s, so time per model can't be measured for Zen, which #366's question about thinking time needs.
**The build step finished: 36.5 minutes, all 25 tasks ticked, no cuts.** That's about what a similar build took on vLLM (36 minutes, run 147), though on a different project, so it's only a rough comparison. It ran on six models: Zen's `deepseek-v4.1-flash-free` 184 requests, `beam-free` 61, Gemma 26, Qwen 397B 16, `gpt-oss-120b` 2, `glm-4.5-flash` 2. Zen carried most of it once Kilo refused, so finding 4 cost less time than feared. Whether the code holds together across those changes is for the tests and the Visual check to say. 9. **Zen requests have no recorded duration.** All 184 show 0 s, so time per model can't be measured for Zen, which #366's question about thinking time needs.
Author
Owner

Run 170 failed after 64.5 minutes, at the first spec-test step. Its proposal, review and build all finished. Skipped: 45 steps, including three of the four spec-test steps, the Visual check and the merge. So this run says nothing yet about whether the code holds together.

  1. One added export ended the run. Not a Free problem: #368.
  2. A spec-test step ran at about a quarter speed for 17 minutes. Its minutes 0-16 had 72 requests that no model served (70 bare 503s from Kilo's breaker, 2 of Kilo's 429s) and 70 parks labelled free quota spent, each up to 12 s. Only 2 or 3 requests a minute were served, then 10 in the minute Kilo came back. In the build step the same breaker failed over to other models, but here most of those requests ended on Kilo alone, and Zen served none of this step. Why is unknown: #367.

Free's scorecard for this run, to the point it stopped

  • It held up: 20 steps succeeded, one model cut and recovered, Kilo lost for most of an hour and the build still finished.
  • No Free failure stopped it, and no person was needed until the export.
  • Not shown: anything after the first spec-test step.
**Run 170 failed after 64.5 minutes, at the first spec-test step.** Its proposal, review and build all finished. Skipped: 45 steps, including three of the four spec-test steps, the Visual check and the merge. So this run says nothing yet about whether the code holds together. 10. **One added `export` ended the run.** Not a Free problem: #368. 11. **A spec-test step ran at about a quarter speed for 17 minutes.** Its minutes 0-16 had 72 requests that no model served (70 bare 503s from Kilo's breaker, 2 of Kilo's 429s) and 70 parks labelled `free quota spent`, each up to 12 s. Only 2 or 3 requests a minute were served, then 10 in the minute Kilo came back. In the build step the same breaker failed over to other models, but here most of those requests ended on Kilo alone, and Zen served none of this step. Why is unknown: #367. **Free's scorecard for this run, to the point it stopped** - It held up: 20 steps succeeded, one model cut and recovered, Kilo lost for most of an hour and the build still finished. - No Free failure stopped it, and no person was needed until the `export`. - Not shown: anything after the first spec-test step.
Author
Owner

Correction to finding 11: the slow step ran out of capacity, rather than failing to fail over. OVH's 2 a minute was the only strong model left (details on #367). The sweep of free plans and what could add capacity is on #369.

Correction to finding 11: the slow step ran out of capacity, rather than failing to fail over. OVH's 2 a minute was the only strong model left (details on #367). The sweep of free plans and what could add capacity is on #369.
Author
Owner

Run 171 started: the same brief and settings as run 170, on the same repository, at d53f9e9. New since run 170: Kilo first and Hetzner second (#369), Kilo's refusals counted (#367), and a spec-test step's edit put back before its retry (#368). Findings from it continue here.

  1. The run's name keeps the brief's # and stops mid-word: # Build "Scratch" — a multiplayer online card ga. It's the first line's first 48 characters, and it shows in the top bar and on the Inference page.
**Run 171 started: the same brief and settings as run 170**, on the same repository, at d53f9e9. New since run 170: Kilo first and Hetzner second (#369), Kilo's refusals counted (#367), and a spec-test step's edit put back before its retry (#368). Findings from it continue here. 12. **The run's name keeps the brief's `#` and stops mid-word:** `# Build "Scratch" — a multiplayer online card ga`. It's the first line's first 48 characters, and it shows in the top bar and on the Inference page.
Author
Owner

Run 171, through the proposal: 2.7 minutes, all on beam-free (explore 0.9, propose 1.8 with one retry). In run 170 these two steps were also fine, so this says nothing yet about Kilo running out.

  1. beam-free wrote the proposal to a file literally named specs/**/*.md. openspec instructions proposal had given it the right path. openspec validate --strict then passed with no proposal.md at all, and Braid's check on the step caught it: "proposal: ready — 0 non-empty path(s)". The retry wrote it to the right place and removed the stray file, in 37 s. Braid's check did its job here, so nothing to fix; one model slip, worth watching if it repeats.
**Run 171, through the proposal: 2.7 minutes, all on `beam-free`** (explore 0.9, propose 1.8 with one retry). In run 170 these two steps were also fine, so this says nothing yet about Kilo running out. 13. **`beam-free` wrote the proposal to a file literally named `specs/**/*.md`.** `openspec instructions proposal` had given it the right path. `openspec validate --strict` then passed with no `proposal.md` at all, and Braid's check on the step caught it: "proposal: ready — 0 non-empty path(s)". The retry wrote it to the right place and removed the stray file, in 37 s. Braid's check did its job here, so nothing to fix; one model slip, worth watching if it repeats.
Author
Owner

Run 171, the proposal's review took 9.8 minutes, and one request was 7 of them. The build has now started, about 13 minutes into the run, with the proposal gate auto-approved as set.

  1. Hetzner's Qwen3.8-27B spent 424 s on one request of the review (14,753 tokens out, about 35 tok/s), where the step's other 11 requests took 2.5 minutes between them. It did finish and answer, so nothing was lost this time, but it's #366's question again: nothing limits a request short of the router's cap, and run 170's same step lost 5 minutes the same way on another model.

Not a bug, checked: the review skipped Gemini 3.8 Flash, first in the judge tier, because its 20 a day are kept for the code review and the Visual check. They're still untouched for when this run gets there.

**Run 171, the proposal's review took 9.8 minutes, and one request was 7 of them.** The build has now started, about 13 minutes into the run, with the proposal gate auto-approved as set. 14. **Hetzner's `Qwen3.8-27B` spent 424 s on one request of the review** (14,753 tokens out, about 35 tok/s), where the step's other 11 requests took 2.5 minutes between them. It did finish and answer, so nothing was lost this time, but it's #366's question again: nothing limits a request short of the router's cap, and run 170's same step lost 5 minutes the same way on another model. Not a bug, checked: the review skipped Gemini 3.8 Flash, first in the judge tier, because its 20 a day are kept for the code review and the Visual check. They're still untouched for when this run gets there.
Author
Owner

Run 171 is at the code review after 31 minutes: 245 requests, none refused, nothing waited for. Run 170 had taken 64.5 minutes to reach its first spec-test step and ended there.

Step Run 170 Run 171
Build, all tasks 36.5 min on six models 5.7 min, 96 requests, all beam-free
First spec-test step failed 0.3 min, 3 requests
All four spec-test steps not reached done in 11.6 min, one after another, all first attempts, all beam-free

233 of the 245 went to beam-free; the other 12 were the proposal's review. So the speed comes from Kilo leading, as the replay said it would (#369), and Kilo hasn't reached its count of about 300 yet.

The speed says nothing yet about the code: the tests and the Visual check have that to say. A spec-test step hasn't needed #368's put-back so far, so that fix is still unproven on this run.

**Run 171 is at the code review after 31 minutes: 245 requests, none refused, nothing waited for.** Run 170 had taken 64.5 minutes to reach its first spec-test step and ended there. | Step | Run 170 | Run 171 | |---|---|---| | Build, all tasks | 36.5 min on six models | 5.7 min, 96 requests, all `beam-free` | | First spec-test step | failed | 0.3 min, 3 requests | | All four spec-test steps | not reached | done in 11.6 min, one after another, all first attempts, all `beam-free` | 233 of the 245 went to `beam-free`; the other 12 were the proposal's review. So the speed comes from Kilo leading, as the replay said it would (#369), and Kilo hasn't reached its count of about 300 yet. The speed says nothing yet about the code: the tests and the Visual check have that to say. A spec-test step hasn't needed #368's put-back so far, so that fix is still unproven on this run.
Author
Owner

Run 171's code review: 15.2 minutes, and it found real gaps in the fast build. It reported fixing these, and its tests pass, 43 in all:

  • The scoreboard table was empty; it now has rows, bid/won cells and the scratched-out rounds the brief asks for.
  • The server bound to 127.0.0.1, so Docker's port forwarding wouldn't reach it.
  • A crash in the sound code under the tests, an always-false variable, and two type errors.

That's the point of the step, and it means the 5.7-minute build wasn't complete on its own. The Visual check will show whether the rest of it is.

  1. Gemini 3.8 Flash served none of it. It answered 503 high demand and then 504 timeouts, and its provider's breaker opened, so the review ran on Qwen 397B, Qwen 27B and one north-mini-code. Failover worked and no request was lost. One Hetzner request again took 257 s (5,521 tokens out), as in finding 14.
**Run 171's code review: 15.2 minutes, and it found real gaps in the fast build.** It reported fixing these, and its tests pass, 43 in all: - The scoreboard table was empty; it now has rows, bid/won cells and the scratched-out rounds the brief asks for. - The server bound to `127.0.0.1`, so Docker's port forwarding wouldn't reach it. - A crash in the sound code under the tests, an always-false variable, and two type errors. That's the point of the step, and it means the 5.7-minute build wasn't complete on its own. The Visual check will show whether the rest of it is. 15. **Gemini 3.8 Flash served none of it.** It answered `503 high demand` and then `504` timeouts, and its provider's breaker opened, so the review ran on Qwen 397B, Qwen 27B and one `north-mini-code`. Failover worked and no request was lost. One Hetzner request again took 257 s (5,521 tokens out), as in finding 14.
Author
Owner

Run 171 finished: 104.7 minutes, nobody needed, PR merged, change archived. 38 steps succeeded and 28 were skipped as not needed. No fix round ran.

Model Requests Model time Longest request
beam-free 260 13.2 min 38 s
Hetzner Qwen3.8-27B 141 52.2 min 423 s
OVH Qwen3.5-397B 99 12.8 min 103 s
Gemini Flash-Lite 6 0.2 min 3 s
north-mini-code 2 0.1 min 3 s

508 requests, none refused, no parks, one compaction, no stalls. 27 failed legs all failed over, mostly open breakers.

Not tested by this run: Kilo running out (260, under its count of about 300), and #368's put-back (every spec-test step passed first time). I expected Kilo to run out and it didn't. Those two rest on their sims and replays.

The game falls short of the brief, and the Visual check passed it anyway. Verdict pass: 13 pass, 0 fail, 3 unclear. From its screenshots and the code:

  1. No animations at all, with the task ticked. Filed with its cause on #374.
  2. The table is below the fold during play. At 1280×720 the first screen is the room's code, its settings and its player list, in raw values (dealer restriction false · first lead leftOfDealer). The felt table and your hand need a scroll.
  3. The hand wraps into two rows of tilted cards, not a fan.
  4. Illegal play stays on the end screen. The check noted it and still passed that criterion. The PR lists it under Not done, as it does the three unclear items, which is honest.
  5. The check played alone against three bots, through a window.__app hook (93 evaluate calls, 5 clicks), bidding 0 every round. Two people in two browsers is covered only by the server's tests.
  6. The Visual check was 54% of the run: 56.8 minutes, mostly a whole 10-round game on Hetzner. One attempt also failed for a report with no verdict; the retry fixed the file in 3.5 minutes.

So Free is now fast and steady enough to finish a whole project alone. What it builds in 5.7 minutes needs checking harder than this run checked it.

**Run 171 finished: 104.7 minutes, nobody needed, PR merged, change archived.** 38 steps succeeded and 28 were skipped as not needed. No fix round ran. | Model | Requests | Model time | Longest request | |---|---|---|---| | `beam-free` | 260 | 13.2 min | 38 s | | Hetzner `Qwen3.8-27B` | 141 | 52.2 min | 423 s | | OVH `Qwen3.5-397B` | 99 | 12.8 min | 103 s | | Gemini Flash-Lite | 6 | 0.2 min | 3 s | | `north-mini-code` | 2 | 0.1 min | 3 s | 508 requests, none refused, no parks, one compaction, no stalls. 27 failed legs all failed over, mostly open breakers. **Not tested by this run:** Kilo running out (260, under its count of about 300), and #368's put-back (every spec-test step passed first time). I expected Kilo to run out and it didn't. Those two rest on their sims and replays. **The game falls short of the brief, and the Visual check passed it anyway.** Verdict `pass`: 13 pass, 0 fail, 3 unclear. From its screenshots and the code: 16. **No animations at all, with the task ticked.** Filed with its cause on #374. 17. **The table is below the fold during play.** At 1280×720 the first screen is the room's code, its settings and its player list, in raw values (`dealer restriction false · first lead leftOfDealer`). The felt table and your hand need a scroll. 18. **The hand wraps into two rows of tilted cards**, not a fan. 19. **`Illegal play` stays on the end screen.** The check noted it and still passed that criterion. The PR lists it under Not done, as it does the three unclear items, which is honest. 20. **The check played alone against three bots, through a `window.__app` hook** (93 evaluate calls, 5 clicks), bidding 0 every round. Two people in two browsers is covered only by the server's tests. 21. **The Visual check was 54% of the run:** 56.8 minutes, mostly a whole 10-round game on Hetzner. One attempt also failed for a report with no `verdict`; the retry fixed the file in 3.5 minutes. So Free is now fast and steady enough to finish a whole project alone. What it builds in 5.7 minutes needs checking harder than this run checked it.
Author
Owner

Run 172: the same Scratch brief on the local profile (vLLM 0.33.0, Qwen3.8-27B W8A16, MTP k=4, TP4 on 4x RTX 5000), thorough, auto-approve. Succeeded 2026-11-04 15:49: 65/65 steps, 0 failures, 2 attempts, 503 min wall against run 171's 105 min. Repo cmoriarty/scratch-vllm-mtp.

171 (Free) 172 (vLLM MTP k=4)
wall 105 min 503 min
agent-step minutes, summed 50 470
assistant messages / tool calls 1,002 / 1,554 3,368 / 6,123
subagents spawned 1 35
prompt tokens 7.3M + 28.5M cache 94.7M
generated (output + reasoning) 0.20M 1.44M
generation incl. prefill, per message 34.7 t/s 35.7 t/s
visual-check rounds / fix rounds 1 / 0 3 / 2
context alarms (compactions / respawns) 1 (1 / 1) 5 (1 / 4)
delivered source / test lines, tests 1,457 / 909, 44 5,021 / 2,650, 87

What the 4.8x is. Volume, not speed: the per-message rate is the same, and the 27B model did 7x the generation, 4x the tool calls and 35 browser subagents, delivering 3.4x the source. Steps: openspec.apply[all] 169 min (171: 5.7), visual check 55 + 74 + 58 min (171: 3.7, one round, braid#379), fixes 33 + 16, spec tests 18 + 9 + 8 (braid#376).

Compactions. Five alarms at 197-202k of 226k, all in forked sessions: four respawns from Braid's note, one real compaction (197k -> 9k with 28 tasks left). The pauses cost 3.8 min in all; the re-orientation cost more: read calls in the apply step went from 3-6 per 15 min before each alarm to 42-75 after, about 10-15 min per event, 40-60 min over the run (8-12 % of wall). braid#378 (alarm wording) and braid#379 (the 64k judge brief that forces the respawn every round) are the two causes filed.

Quality of the app itself: not yet judged by the owner.

**Run 172: the same Scratch brief on the local profile (vLLM 0.33.0, Qwen3.8-27B W8A16, MTP k=4, TP4 on 4x RTX 5000), thorough, auto-approve.** Succeeded 2026-11-04 15:49: 65/65 steps, 0 failures, 2 attempts, 503 min wall against run 171's 105 min. Repo `cmoriarty/scratch-vllm-mtp`. | | 171 (Free) | 172 (vLLM MTP k=4) | |---|---|---| | wall | 105 min | 503 min | | agent-step minutes, summed | 50 | 470 | | assistant messages / tool calls | 1,002 / 1,554 | 3,368 / 6,123 | | subagents spawned | 1 | 35 | | prompt tokens | 7.3M + 28.5M cache | 94.7M | | generated (output + reasoning) | 0.20M | 1.44M | | generation incl. prefill, per message | 34.7 t/s | 35.7 t/s | | visual-check rounds / fix rounds | 1 / 0 | 3 / 2 | | context alarms (compactions / respawns) | 1 (1 / 1) | 5 (1 / 4) | | delivered source / test lines, tests | 1,457 / 909, 44 | 5,021 / 2,650, 87 | **What the 4.8x is.** Volume, not speed: the per-message rate is the same, and the 27B model did 7x the generation, 4x the tool calls and 35 browser subagents, delivering 3.4x the source. Steps: `openspec.apply[all]` 169 min (171: 5.7), visual check 55 + 74 + 58 min (171: 3.7, one round, braid#379), fixes 33 + 16, spec tests 18 + 9 + 8 (braid#376). **Compactions.** Five alarms at 197-202k of 226k, all in forked sessions: four respawns from Braid's note, one real compaction (197k -> 9k with 28 tasks left). The pauses cost 3.8 min in all; the re-orientation cost more: `read` calls in the apply step went from 3-6 per 15 min before each alarm to 42-75 after, about 10-15 min per event, 40-60 min over the run (8-12 % of wall). braid#378 (alarm wording) and braid#379 (the 64k judge brief that forces the respawn every round) are the two causes filed. Quality of the app itself: not yet judged by the owner.
Author
Owner

Closing this log. Runs 170, 171 and 172 are over, and every finding that needed a fix has its own issue.

Finding Went to
1, 14: a request that only thinks, or takes minutes braid#366, open
4, 5, 11: Kilo's count, a breaker read as spent quota, capacity braid#367 and braid#369, closed
10: an added export ended the run braid#368, closed
12: the run's name cut at 48 characters braid#388, closed
16: animations ticked and unbuilt, and passed braid#374, closed
21, and run 172: Visual check rounds, the alarm's wording, spec tests braid#379, braid#378, braid#376, closed

Never filed, small, and still only here:

  • a retry Braid handled itself shows as a red provider.unknown line (3);
  • requests cancelled a second or two after a failover succeeded (7);
  • Zen requests recorded with no duration (9);
  • why the first review left Qwen while it was answering (2).

Findings 17 to 20 describe the generated game and how the check played it. The owner's judgement of the app's quality was never recorded here.

Closing this log. Runs 170, 171 and 172 are over, and every finding that needed a fix has its own issue. | Finding | Went to | |---|---| | 1, 14: a request that only thinks, or takes minutes | braid#366, open | | 4, 5, 11: Kilo's count, a breaker read as spent quota, capacity | braid#367 and braid#369, closed | | 10: an added `export` ended the run | braid#368, closed | | 12: the run's name cut at 48 characters | braid#388, closed | | 16: animations ticked and unbuilt, and passed | braid#374, closed | | 21, and run 172: Visual check rounds, the alarm's wording, spec tests | braid#379, braid#378, braid#376, closed | Never filed, small, and still only here: - a retry Braid handled itself shows as a red `provider.unknown` line (3); - requests cancelled a second or two after a failover succeeded (7); - Zen requests recorded with no duration (9); - why the first review left Qwen while it was answering (2). Findings 17 to 20 describe the generated game and how the check played it. The owner's judgement of the app's quality was never recorded here.
Sign in to join this conversation.
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
cmoriarty/braid#365
No description provided.