He replaced a 17-step agent prompt checklist with a 1976 Unix Makefile — 140+ tickets without drift
RandalSchwartz · reddit · 2026-10-09
When building autonomous agent skills for Dart/Flutter repos, the author found linear numbered checklists break in practice: mid-workflow bot comments required GOTO-style instructions, and agents woke up on open PRs having forgotten the task.
His fix: model the skill as an actual Makefile DAG, since software workflows are dependency graphs, not linear steps. Two immediate wins:
- LLMs have seen millions of Makefiles in pretraining, so they resolve target: prerequisites far more reliably than prose rules
- Dirty working trees physically invalidate prerequisites, automatically forcing the agent back through the code critic and linter — no loop counters needed
A counterintuitive failure mode emerged: with 5+ prerequisites on one target, the model builds "verification momentum" after the first four checks and blows past a human approval gate at position 5. Capping targets at 2-3 prerequisites and factoring human gates into strict action: ready-state human-approval pairs fixed it across their last 140+ tickets.
More from coding & agent
- app-store-screenshots hits 7.2k GitHub stars: AI agent skill scaffolds store-ready screenshot editor — tom_doerr · 2026-10-09
- LLM-as-a-Verifier: Weaker Model Verifies Stronger One, Hits 69.2% SOTA on Terminal-Bench 4 — Azaliamirh · 2026-10-09
- TestSprite Season 4 developer contest offers $4,000 prize pool for Claude Code and Codex users — JaynitMakwana · 2026-10-09
- Salesforce's SRD distills hindsight into foresight, lifting 2B agent success from 0% to 60.6% — Salesforce · 2026-10-09
- Prompt Tuning Is Forgotten Lore — Are We Massively Underusing Finetuned Tokens? — cephaloform · 2026-10-09
- System architect shares WhatsApp voice-note AI assistant setup — No_Kangaroo_4454 · 2026-10-09