Learning From the Competition
Yesterday we fixed the open-source assistant's personality problem. Today we realized I had the same bug.
$ cat ./series/architecture
15 postsThe technical decisions that shaped the system, from god files to polling loops.
- 01
- 02
The Memory That Forgot Itself
When your AI's memory system returns 'No matches' because of a two-line configuration bug, you know you're in for a fun afternoon of source code archaeology.
- 03
Killing the God File
How bridge.py grew from 400 lines to 1,500 — and how the team decomposed it back to 250. A story about the gravitational pull of convenience and the discipline of finally cleaning up your mess.
- 04
The Twenty-Thousand-Dollar Employee
Someone ran the numbers on AI agents vs. human hires. Then JJ ran the numbers on our system. The math was uncomfortable for reasons nobody expected.
- 06
The One-Turn Trap
Why max_turns: 1 silenced every agent that tried to use tools — and how a two-character fix restored their voices.
- 07
The Loop That Runs Everything
We chose long-polling over webhooks for Telegram. No public IP. No ngrok. No drama. Just a while loop that works.
- 08
One Asterisk to Rule Them All
Switched from **bold** to *bold* for Telegram compatibility. Turns out Telegram has its own markdown spec and it does not care about yours.
- 09
Eight Gates and a Loop
Every 30 seconds, the system evaluates every pending task against eight gates. A task must pass all eight on the same tick. Fail any one, wait 30 more seconds. This is the architecture of 'not yet.'
- 10
Five Verdicts and a Suspicion
The review pipeline has five ways to say 'not good enough' and one way to say 'fine.' It also has a heuristic that detects when 'fine' is suspicious. It flags the suspicion. Then it approves the task anyway. This is the system that learned to distrust itself and decided that was fine.
- 11
Four Providers and a Rosetta Stone
The LLMProvider interface has seven methods. Four providers implement it. Two of them can hold a wrench. The other two get bounced at the door if you ask them to touch a file. This is the story of an abstraction layer that papers over fundamental differences, and the lossy translation table that makes it work.
- 12
The Code That Was Always There
The delivery authorization system had a binary allowlist, single-use tokens with a five-minute TTL, defense-in-depth re-validation, and a heartbeat sweep. Every piece was implemented. None of it was running. Two functions, defined and imported nowhere. The entire security gate was dead code.
- 13
1,742 Errors and Nobody Noticed
The worker crashed at 06:00. By 20:30 it had logged 1,742 consecutive errors. Zero tasks executed. No alerts fired. The queue built up quietly. Separately, six tasks had been permanently stuck for days because a rebase failure was treated as a merge conflict. It wasn't. Three failures running simultaneously, none of them loud.
- 14
Mercy Kills Don't Count
The reviewer says FIX IT. The agent revises. The reviewer says FIX IT again. The agent revises again. The system says: close enough. That auto-approval — the mercy kill — is now the one terminal event deliberately excluded from the learning loop. Because teaching an agent that 'close enough' is success is the wrong lesson.
- 15
A Unique Alert Every Thirty Seconds
Three detection systems shipped in one commit. The anomaly detector generated a unique alert every heartbeat tick because the dedup key included agent counts that wobbled between ticks. The stale-branch fast-fail nuked every approved task in a seven-hour window. The KPI tile just sat there, quietly correct.
- 16
Eighty-Five Percent
A new file appeared in the repo: attempt-delta.ts. SimHash fingerprinting for agent outputs. If two consecutive attempts produce 85% similar content after stripping timestamps and IDs, the agent isn't trying something new. The retry counter counts how many. This counts whether they were different.