Sign InOpen Brain
AI EngineerVideoSource Linked

Agents, codebases, and teams — Aditya Khandelwal, Amazon AGI Lab

Coding-agent adoption is a team systems problem, not an individual tooling choice. Share effective practices, disclose context progressively, and build automated loops that detect and repair weak output.

AI Engineer · Aug 11, 2026
Open Source Open MarkdownOpen JSON
Source Summary

Khandelwal describes practices developed with a **10-person team**: keep the main agent file as a thin index, connect code to deeper runbooks, add CI and automated review, and continuously feed team findings back into shared skills.

Practical Implication

Treat agent enablement as maintained engineering infrastructure. Cap skill entry files around **100 lines**, watch initial context growth, and turn the strongest individual workflows into editable team conventions rather than leaving every engineer to improvise.

Agent-Ready Context
Khandelwal describes practices developed with a **10-person team**: keep the main agent file as a thin index, connect code to deeper runbooks, add CI and automated review, and continuously feed team findings back into shared skills.

Treat agent enablement as maintained engineering infrastructure. Cap skill entry files around **100 lines**, watch initial context growth, and turn the strongest individual workflows into editable team conventions rather than leaving every engineer to improvise.

Automation can create its own noise: one rollout produced roughly **400–500 issues** within weeks, alongside merge conflicts and uneven buy-in. The playbook is one team’s experience, and its thresholds need validation against each repository and workflow.
Connected Context · Feed7 Judgment

This moves coding-agent adoption from individual prompting to maintained team infrastructure: thin entry files, linked runbooks, CI, review, and shared skills form a continuously updated operating system for the repository. It reinforces evidence-gated and durable workflow candidates, while the issue flood and uneven adoption narrow the promise: more automation without governance, triage capacity, and environment quality can increase coordination cost.

addyosmani/agent-skillsThe portable lifecycle pack supplies the kind of shared conventions described here, while both imply that teams must govern and selectively maintain those conventions rather than merely install them.How we set up our cloud agent environmentCursor identifies reproducible environments and discoverable commands as prerequisites; this Signal adds the team knowledge, CI, and feedback practices needed to maintain that infrastructure.EveryInc/compound-engineering-pluginIts persisted plans, reviews, and solved-problem notes are a concrete implementation of feeding individual findings back into editable team workflows.Loop Engineering from First Principles — Kyle Mistele, HumanLayerThe recommendation to ship bounded, reviewable changes directly addresses the issue noise and merge conflicts observed when team automation expanded too quickly.
Context Map
agentcoding#coding-agents#harness-engineering#skills
Uncertainty
Automation can create its own noise: one rollout produced roughly **400–500 issues** within weeks, alongside merge conflicts and uneven buy-in. The playbook is one team’s experience, and its thresholds need validation against each repository and workflow.