Topic

context-engineering

3 min read

The agent drifts from your requirement while the context fills up

You asked for one behavior; twelve tool calls later it optimized the last output instead of the acceptance criteria. Context fills with intermediate logs, recent tokens dominate, and the original brief gets buried. Here is how drift happens and the cheap anchors that keep a task on rail.

Read article
4 min read

The agent argued against a file it had not re-read in twenty turns

The agent reads a file once and keeps that buffer across a long session. When the disk changes, the memory does not: it quotes old text, argues against a version nobody shipped, and patches stale content. The gate: invalidate the cache and re-read before editing.

Read article
4 min read

Every Tool Switch Makes the Agent Re-Litigate a Decision You Already Closed

An hour earlier we had chosen the retry queue over a direct write. Switching editors to run the migration, the fresh session saw no recorded decision and calmly proposed the direct write again. Forty minutes later, the closed decision had been reopened and compromised.

Read article
4 min read

When Your Agent Summarizes Your Codebase, It Quietly Drops the Types

An agent reads your file, writes a summary for the next session, and in that paraphrase the generic constraint, the discriminated union, the branded ID all collapse into a vague noun. The next session then builds on a version of your code that never existed.

Read article