Rage: The real story is combining lage's dev ergonomics with BuildXL's
correctness at scale. Ken has wanted this synthesis for years since
co-creating lage. Frame the series around ergonomics vs correctness.
Cloudpack benchmarking: synthetic codebases generated from observed
attributes of real shipping code. Pipeline: observe real repo → produce
BenchProfile JSON → corpus_gen creates deterministic synthetic codebase
(same seed = byte-identical) → corpus_verify confirms conformance →
benchmark at various scales (100 to 50k modules).
Three TypeScript module variants (barrel, intermediate, leaf) with
realistic import structures, side effects, and ambient refs. JSON
configs, markdown docs fill out the shape. Tolerances: 2% file count,
5% byte/line count.
This solves 'we can't share the real codebase' for public benchmarks.
Worth a blog post on its own.
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
The novel insight: traditional bundlers are compilers (source in, file out,
start over). Cloudpack is a database: the module graph is the schema, bundle
requests are queries, every output form is a materialization. One truth,
many views. No separate dev/prod pipeline. No divergence.
This concept is the unifying thread of the entire cloudpack series.
From the architect brief: 'The module graph is the software. The bundle is
a query.'
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Key principle added to standing rules: 'Lead with outcomes, not internals.
Nobody cares about a fingerprinting algorithm. They care that their builds
stop lying.'
Rage series retitled:
- 'Two-Phase Fingerprinting: BuildXL's Deep Magic' → 'How We Made the Cache Stop Lying'
- 'Hooking Syscalls Without a Kernel Driver' → 'Zero-Config File Watching Across Three Operating Systems'
- 'TypeScript 7 Makes the Build Harness More Important' → 'TypeScript 7 Is 10x Faster. Your Builds Aren't.'
- 'node_modules in a Content-Addressed Store' → '90 Seconds to 300 Milliseconds: Fixing CI Install Times'
- 'What I Learned Maintaining Lage and Why I Rewrote It' → 'The Architecture Mistake I Made in Lage'
Cloudpack series retitled:
- 'The Bundle Is Not a File' → '604ms Rebuilds at 50,000 Modules'
- 'The 2KB Module Summary' → '50KB Down to 2KB: Why We Stopped Parsing Source Files'
- 'Dead Code That Actually Stays Dead' → 'We Shipped 40% Less JavaScript and Nobody Noticed'
- 'The Service Worker as ld.so' → '50% Less Bandwidth Per Deploy'
- 'Signing Manifests, Not Code' → 'A Compromised Server Can\'t Inject Code'
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Cloudpack series maps native linker innovations to JS bundling at
Microsoft Teams scale (50k+ modules). Source: ~/workspace/wundler/
(kenotron-ms/cloudpack on GitHub).
Series plan:
1. The Bundle Is Not a File (ThinLTO/mold/BOLT analogy)
2. The 2KB Module Summary (25x compression, content-addressed)
3. Dead Code That Actually Stays Dead (two-layer tree-shaking)
4. The Service Worker as ld.so (ABS delta delivery)
5. Your Users Are Teaching the Bundler (PGO from browser sessions)
6. Signing Manifests, Not Code (ed25519, threat model)
Key receipts: 604ms warm rebuild at 10k modules, ~100MB for 50k
module summaries (fits L3), ~50% CDN bandwidth saved per deploy.
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Content model inspired by makingsoftware.com:
- Topics not dates (grouped by investigation theme)
- Homepage as pitch not index (show quality before navigation)
- Series are first-class (landing pages, progress, cross-links)
- Figure density: 1 visual per 300-500 words for technical posts
- No blog chrome (no comments, shares, newsletter popups)
Rage series plan (6 posts, all sourced from kenotron-ms/rage):
1. Your Build Cache Is Lying to You (the stale cache problem)
2. Two-Phase Fingerprinting: BuildXL's Deep Magic (WF/SF algorithm)
3. Hooking Syscalls Without a Kernel Driver (3 OS, 3 mechanisms)
4. TypeScript 7 Makes the Build Harness More Important (counter-intuitive)
5. node_modules in a Content-Addressed Store (CI kill shot)
6. What I Learned Maintaining Lage and Why I Rewrote It (meta narrative)
Each post has planned visuals (diagrams, flows, animations via Remotion).
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Content pipeline now starts with visual planning before drafting:
- What diagrams does this piece need?
- Static SVG? Animated (Remotion)? Interactive (d3-graphviz)?
- Plan alongside the outline, not as a decoration pass at the end
Cleaned up duplicate Project Context section in AGENTS.md.
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Every post must be considered for diagrams before publishing.
The question is not 'does this need a diagram?' but 'what would
illustrate this better than prose?'
Visualization approach:
- Claude hand-codes SVGs using primitives following design spec
- Warm palette (paper bg, brown-black ink, terracotta accent)
- Diagrams break wider than prose (840px breakout)
- DOT + d3-graphviz for interactive architecture graphs
- Image generation for concept art / visual targets
Rule: 1000+ words with zero visuals is a warning. The consideration
is mandatory. The diagrams are not.
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
- New lint check: 'doesn't just X. They Y' pattern (Claude rhetorical crutch)
- Fixed last instance in judgment essay: 'doesn't just lose income. They lose'
→ 'loses the acknowledgment... The income is secondary.'
- Updated voice-check skill dimension 2 with the pattern description
- Updated AGENTS.md with the new check
- Tightened regex to avoid false positives on legitimate 'isn't only' framing
🤖 Generated with Amplifier
Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
Ken