Over-amplification of Subagents and Skills
zeeg · x · 2026-07-11
The author continues a discussion on the behavior of next-gen models/Agents: in actual use, models are aggressively invoking subagents and skills.
The scenario described:
- Prompt the model to generate code normally
- Attach a cleanup/review skill
- The model repeatedly auto-invokes this skill
- That single skill internally spawns about 20 subagents
This automation rapidly amplifies cost and complexity. He describes it as a "wealth transfer," citing another view that prompts and skills must be rewritten for every new model generation because default behaviors and usage patterns shift.
More from coding & agent
- Codex helps build Valdiluce, an open-world game with climbing, gliding and gondolas — Dimillian · 2026-07-22
- HeyGen adds a media-sourcing skill for coding agents with 75k images and 10k tracks — HeyGen · 2026-07-22
- Agent search bottlenecks are now about variance, not raw latency — rohanpaul_ai · 2026-07-22
- LangSmith adds tracing for Pipecat, LiveKit, OpenAI Realtime, and Gemini Live — LangChain · 2026-07-22
- An MCP server signs every AI agent tool call into a verifiable Merkle chain — Funky_Chicken_22 · 2026-07-22
- Annotated transcript of a Claude Code team interview is now available — trq212 · 2026-07-22