All Articles
Guard Your Agent's Destructive Operations with Pre- and Post-condition Contracts
A design for wrapping an autonomous agent's writes in deterministic pre- and post-condition checks. A contract gate stops the destructive operations that better prompts can never reliably prevent.
Before You Send Reviews and Crash Logs to the Claude API: A Reversible PII Masking Design
When you run App Store reviews and Crashlytics logs through the Claude API, the personal data buried in the text is unavoidable. Here is a reversible masking design that lets you trace the model's output back to the real record, plus the pitfalls I hit in production, with code.
Fixing 403 Write Access to Repository Not Granted When Pushing from Claude Code
When Claude Code automation pushes to GitHub and hits a 403 'Write access to repository not granted', the cause usually lies in how fine-grained PATs differ from classic ones. Here's how to diagnose repository access, Contents permissions, and org approval.
Moving Firebase From CocoaPods to SPM With Claude Code as a Partner: An Implementation Memo
A record of migrating Firebase from CocoaPods to Swift Package Manager in an indie iOS app. What I handed to Claude Code, what I decided myself, and the Crashlytics dSYM trap I hit along the way.
Grouping Crashes by Root Cause: A Triage Design Built on the Claude API
Crashlytics 'Issues' often scatter the same root cause across separate entries. After years of running apps with 50M+ cumulative downloads, here is how I use the Claude API to regroup crashes by actual root cause and rank them, with working code and real numbers.
Isolating Poison Messages in a Claude Async Pipeline: A Dead-Letter Queue Implementation Note
How one broken input can stall an entire batch — and how to isolate these 'poison messages' with a Cloudflare Queues dead-letter queue. Covers classifying Claude API failures and safe redrive, all from production experience.
What to Do When Claude Code Hits the Context Window Limit
Long Claude Code sessions eventually hit the 200k token limit and stall. Here's how to use /compact, --continue, and session design patterns to keep development moving without losing context.
Claude Code vs. Claude in Chrome: Where I Draw the Line in Daily Ops
Running apps solo means constantly hopping between an editor and a browser, and that hopping quietly drains your focus. Here is how I split work between Claude Code and Claude in Chrome over a month, plus the rule I use when a task straddles both.
Make Agent Failures Reproducible: Deterministic Replay and Event Sourcing
An autonomous agent that fails at 2 a.m. can't be reproduced by simply running it again. Record every nondeterminism boundary as an append-only event log and replay the failed run deterministically — with working code and operational lessons.
Designing Skills That Stabilize Output: Template Fixing and Decision Guides in Claude Code
A practical design guide for stabilizing wobbly Claude Code skill output: stating when to use it, decision-guide tables, fixing quality with references templates, and traceability rules, with code from real operations.
The Second Half of My SKILL.md Wasn't Being Read: Keeping It Under 200 Lines
Sparked by an observation that Codex CLI stops reading SKILL.md at around 220 lines, a look at how long an agent actually reads, and how to keep SKILL.md under 200 lines by offloading detail into references.
Continuing past max_tokens in the Claude API without duplicated text or broken code fences
Detect stop_reason: max_tokens, continue the generation with an assistant prefill, and stitch the parts back together without duplicated seams or broken code fences. A production-tested continuation pattern in TypeScript.