Posts

Engineering ·
An External Monitoring Path That Works Even When the Node Fails
Internal alerts cannot watch over the complete failure of the node they run on. How PinLog separates internal monitoring from external HTTPS/TLS checks, and what that separation does not solve.
AI Agents ·
AI Processing of Incident Alerts, with Output Verified by Code
PinLog's Sentinel uses Hermes to process alert payloads, while trusted code revalidates mentions, URLs, and the final format. Critical, Warning, and RESOLVED policies, plus a closing one-line summary, make that division of responsibility concrete.
Retrospective ·
An AX Training Assistant’s Retrospective on Using AI After Training
Supporting corporate AX training led me to reflect on what helps people return to AI in their own work after the class ends.
AI Agents ·
Building a Personal Knowledge Assistant on Hermes
Introducing Hermes Agent to a personal knowledge store inspired by LLM Wiki: preserving existing knowledge and role boundaries while connecting retrieval, validation, and delegation.
Retrospective ·
Safely Removing the AlgoSu Blog’s Relation Graph and Search
Deleting a feature is scarier than adding one. You never know what else might break. Removing the decision-record (ADR) relationship graph and search from AlgoSu's blog taught me that safe deletion is a craft with its own method.
AI Agents ·
When to Bring In a High-End AI Model (Fable 5)
I switched to Fable 5, a new high-end model, the day it launched. Six sprints later, what I'd gained wasn't performance but strategy. A record of when to reach for a premium model, and why that judgment is the same thing as paying down technical debt.
Retrospective ·
Separating Markdown Decision Records for AI from HTML Versions for People
What I learned by splitting decision records (ADRs) into Markdown for AI agents and HTML for people.
AI Agents ·
Sliding-Window Context Management for AI Agents
Over 152 sprints, the decision records (ADRs), the work notes (MEMORY.md), and the agents' rules file (CLAUDE.md) all grew together. Making an AI agent read all of it every session was like pouring more water into a full bucket. This is how I trimmed agent context with one simple rule borrowed from the sliding window algorithm.
AI Agents ·
Introducing Critic, a Review Agent Running on Another Company’s Model (Codex)
I brought in Codex, another company's coding model, as a review agent, as an addition rather than a replacement. After 18 rounds of review, I realized what I needed wasn't a different model. It was a system not tied to any one model.
Engineering ·
A CI Refactor of Waiting PRs, Duplicated Config, and Coverage Gates
Starting from a Channel.io blog post: auto-merge, composite actions, coverage gates, and pre-checks before building, across five sprints. How one reference article turned into a real pipeline overhaul, and why two pieces of work that shipped zero lines of code were still wins.
Retrospective ·
Moving to Programmers After Baekjoon’s Shutdown, and Rethinking External Dependencies
On the day I saw Baekjoon's shutdown notice, migrating AlgoSu's problem source to Programmers taught me something I didn't expect. It wasn't a new technology but an attitude about building services that lean on external platforms.
AI Agents ·
Request Channels, Priorities, and Shared Rules for 12 AI Agents
As AI agents multiplied, control became the problem. One lead agent as the only channel, three priority tiers, one shared rules file: a firsthand account of building a governance structure solo.