Posts 349 entriesPage 10 of 35

Posts

Technical Deep-Dive

Safety Model First: 16-Tool Ops MCP, One Day

Design a 7-point safety model before writing tools. How server-ops-mcp shipped 16 tools, 40 tests, and v0.1.0 in a single day.

Read
Technical Deep-Dive

FTS5 Fallback: How Zero Search Results Became Five (ICO Dogfood Day One)

First real dog-food run of Intentional Cognition OS scored 0/5 question engagement against a corpus that contained every answer. Root cause: AND-only FTS5 query construction plus a possessive normalization order-of-operations bug. Fix: strict-then-broad fallback. Result: 5/5, 28 citations, ~$0.20.

Read
Technical Deep-Dive

Ship Dormant, Wire Later — A Multi-Agent Slack Production Day

Ship infrastructure dormant behind feature flags. The activation day is wiring plus a CLI — eleven PRs, six dormant primitives going live in one day.

Read
Technical Deep-Dive

Five Tags, Zero Ships: How an Auto-Release Workflow Lied for a Whole Day

Five GitHub release tags created. npm registry unchanged. Three discrete bugs: tests silenced with || true, monorepo version drift, missing npm publish step.

Read
Technical Deep-Dive

A v1.0 Is a Gate, Not a Tag

Why release gates should accept GO with conditions, not binary GO/NO-GO. How ICO v1.0.0 shipped with documented gaps and a same-day v1.0.1.

Read
Technical Deep-Dive

Honest Perf Benchmarks for a Paid-API Compiler

Four PRs, three releases, and a benchmark suite that won't lie to you: seeded-RNG corpora, double-gated Claude scenarios, and skipped-but-recorded records.

Read
Technical Deep-Dive

Five Silent Failures in One Day

Five tools said PASS without doing the work — pr-prescreen, .gitignore, prettier, SSH deploy, and a regex. Each silent failure and the guard that catches it.

Read
DevOps

Deterministic First, LLM Second: An Advisory CI Pre-Screen

Wire an LLM into CI without veto power: the deterministic classifier is the product, the LLM is advisory polish, and the never-block contract retires the old system the same day.

Read
Technical Deep-Dive

Self-Improving Skills: Three Schema Versions in One Day

Backward-compatible schema upgrades for Claude skills: L0 metadata indexing, conditional visibility, and config prompts eliminate mid-task environment errors. 2783 skills.

Read
Technical Deep-Dive

Transitive CVE Clearance: The Dual-Layer Pattern

How v0.9.1 cleared 6 high-severity transitive CVEs in axios + fast-uri — and why the dep bump needed top-level overrides to actually stick.

Read