[oddharsh]

Two releases shipped unlogged — Dyno (#295) and the lens docs fix (#288) — so /updates still names v179 as the running build while describing /perf, a path that now redirects. This stages the entry covering both, and records two findings that aren't derivable from the tree.

v180  dyno  The performance trend becomes Dyno and moves into the garage,
            /perf redirects to it, and the lens docs stop reading a refused
            Browser Run call as a verdict on the engine

Gotcha 26 — run_worker_first caps at 100 rules, and we're at exactly 100

Adding two paths for /garage/dyno took it to 102 and wrangler refused to start the Worker:

Error: Too many `run_worker_first` rules were provided;
102 rules provided exceeds max of 100.

Neither npm run build nor wrangler deploy --dry-run catches this. Both passed on a config that could not boot, because parseStaticRouting runs at session creation. npm run routes:check found it, since createTestHarness() starts a real Worker — a second reason the route oracle earns its keep beyond the routes it sweeps.

The note also records the two things that save the next person the afternoon:

• Check the existing wildcards first. /garage/*, /lwe/*, /access/*, /coffee/* and others are already there. The two /garage/dyno rules were redundant, which is how that change shipped at 100 rather than 102.
• Deduping buys nothing. The list holds 107 raw entries, 7 of them exact duplicates, and wrangler counts the 100 unique ones. Real headroom means folding paths onto wildcards, which changes what the Worker sees and isn't a cosmetic edit.

Gotcha 27 — a red github-advanced-security is usually not your diff

Seen twice on 2026-08-08, on two unrelated PRs (#289, #295), both times with the separate CodeQL check reporting "No new alerts in code changed by this pull request."

The signature is recognisable on sight: empty output title, and a single annotation pointing at .github:<line> — a path that's a directory in this repo and therefore can't have a line. The job log names the real actor (COPILOT_AGENT_INPUTS, COPILOT_JOB_NONCE, Preparing Copilot...).

Read the sibling CodeQL check before touching code. It's also not a required check: both of those PRs merged and shipped with it red.

Meanwhile, the pipeline closed itself

The nightly job ran for the first time and recorded a real measurement:

2026-08-09  36d3e55  worker_gzip=273039  pages_br=481886  assets_br=59413  nightly

aadhar.sh/garage/dyno.json now returns count=5, measured=1, so the chart has its first solid segment against the dashed hand-entered prefix. End to end, unattended.

Verification

• npm run checkpoints:check — projection agrees with D1 (115 released), v180 staged and unreleased
• npm run build clean · npm test 200 pass

🤖 Generated with Claude Code

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Updated (UTC)
✅ Deployment successful! View logs	aadhar-sh	29a2831	Aug 09 2026, 02:57 PM

[github-actions]

Wire-size diff vs merge base

36d3e55 → 08ddd65

Brotli q11, the quality the edge serves. Every number here is deterministic:
an unchanged file produces no row. Advisory — this check fails on nothing.

Client assets

No change. 11 files, 58.02 KiB Brotli total.

Pages

Total Brotli: 470.59 KiB → 470.66 KiB (+0.07 KiB, +0.0%) across 46 files, 2 changed.

No single file moved by 0.13 KiB or more.

_{Plus 2 files under the 0.13 KiB floor, net +0.07 KiB — usually content-hash churn in /a/ references, counted in the total above.}

Worker bundle

266.64 KiB → 266.73 KiB gzip (+0.09 KiB, +0.0%)

Largest module deltas (raw bytes in bundle):

file	base	head	Δ
.build/holding/_worker.js/checkpoints.json	20.29 KiB	20.56 KiB	+0.27 KiB (+1.4%)

Dictionary deltas (pd/)

134 → 134 deltas, 385.97 KiB → 386.37 KiB (+0.40 KiB).