# feed7 Weekly #005

Six practical upgrades for stronger agent workflows, sharper evals, safer tools, and better context.

Published: 2026-08-10T00:00:00.000Z  
Selection: Six source-backed signals and one distraction to leave out.

## 1. addyosmani/agent-skills

Treat the pack as a menu of enforceable workflows: begin with TDD, debugging or code review, then add broader lifecycle automation after checking how each skill fits your repository. Codex can load it as a native plugin from CLI v0.122+.

Source: https://github.com/addyosmani/agent-skills
Feed7: https://feed7.dev/p/addyosmani-agent-skills-2e74d0228d
Trust: Needs Review

## 2. Benchmarking Coding Agents on New vs Legacy Codebases — Denys Linkov, Wisedocs

Benchmark agents against explicit requirements, runnable end-to-end tests, deployment constraints, and hidden assumptions—not elapsed time or lines changed. Wisedocs also found a monorepo simpler for verification and sandbox setup across its former 10+ repositories.

Source: https://www.youtube.com/watch?v=7vn4WpqNpck
Feed7: https://feed7.dev/p/benchmarking-coding-agents-on-new-vs-legacy-codebases-de-9e82a687ae
Trust: Source Linked

## 3. Learning When to Trust via Selective Context Preference Optimization

Evaluate retrieval-augmented agents for selective trust, not only prompt-injection resistance. The proposed SCOPE method trains on matched preference pairs balanced across all four conditions so resistance does not come from ignoring useful context.

Source: https://arxiv.org/abs/2608.06377v1
Feed7: https://feed7.dev/p/learning-when-to-trust-via-selective-context-preference--91ef3aae9d
Trust: Needs Review

## 4. The Bitter Lesson of Tool Calling

If your agents already write reliable code, test a typed-stub execution layer that lets one turn chain or parallelize calls. It also matched or exceeded JSON in 13 of 14 models under parallel fan-out and stayed stable in the reported context-rot condition.

Source: https://arxiv.org/abs/2608.06370v1
Feed7: https://feed7.dev/p/the-bitter-lesson-of-tool-calling-f5a1122448
Trust: Needs Review

## 5. How Cursor Router chooses the right model for the task

Builders implementing model routing should learn from real task outcomes, including corrections and task progression, rather than rely only on benchmarks. Include token use, cache misses, model-switching costs, confidence thresholds, and an explicit per-turn budget.

Source: https://cursor.com/blog/how-cursor-router-works
Feed7: https://feed7.dev/p/how-cursor-router-chooses-the-right-model-for-the-task-b3425f8b3f
Trust: Official Source

## 6. Pause workflows for approval with Chat SDK

Use this as a durable gate before an agent deploys, changes production data, or takes another consequential action. Scoped approvers, platform-signature checks, and the returned user identity reduce the custom authorization plumbing.

Source: https://vercel.com/changelog/chat-sdk-durable-approvals
Feed7: https://feed7.dev/p/pause-workflows-for-approval-with-chat-sdk-2dd1524bd8
Trust: Official Source

## 7. New ways to learn and teach with ChatGPT Work and Codex

Builders in education may want to examine whether these plugins can supply agents with domain-specific workflows for research, teaching, or project work.

Source: https://openai.com/index/learn-teach-chatgpt-work-codex
Feed7: https://feed7.dev/p/new-ways-to-learn-and-teach-with-chatgpt-work-and-codex-b274d6785e
Trust: Official Source

## Agent Instruction

Use these items as source-backed context. Do not invent claims beyond linked material. Prefer practical implications for solo developer work. If sources conflict, call it out.