# feed7 Weekly #009

Six practical upgrades for safer, better-evaluated coding-agent sessions.

Published: 2026-09-07T00:00:00.000Z  
Selection: Six source-backed signals and one distraction to leave out.

## 1. SWE-Gate: Passing Functional Tests Is Not Enough for Software Engineering Agents

Treat green tests as one gate, not final acceptance, for coding-agent patches. Encode review expectations as executable checks where possible, and evaluate issue resolution separately from compliance with repository-specific requirements.

Source: https://arxiv.org/abs/2609.04167v1
Feed7: https://feed7.dev/p/swe-gate-passing-functional-tests-is-not-enough-for-soft-4951efd983
Trust: Needs Review

## 2. Clean Engineering, Unstable Measurement: A Preregistered Reliability Failure of Black-Box LLM Observers on Shared Endpoints

If an LLM judge gates agent releases or training data, test it as a measurement instrument first. Repeat identical requests, estimate the noise floor, retain execution records, and run a small pilot; the authors say roughly 2% of call volume would have revealed both unreachable gates.

Source: https://arxiv.org/abs/2609.04198v1
Feed7: https://feed7.dev/p/clean-engineering-unstable-measurement-a-preregistered-r-75afc6b794
Trust: Needs Review

## 3. Set per-user budgets on AI Gateway

Set a default allowance for every user, then override it for heavy agent users. Production or shared keys should be attributed to the team, and CLI management requires Vercel CLI 59.6.2+.

Source: https://vercel.com/changelog/set-per-user-budgets-on-ai-gateway
Feed7: https://feed7.dev/p/set-per-user-budgets-on-ai-gateway-34d93c0361
Trust: Official Source

## 4. How our agents build on-brand pages with design.md

For agent-built interfaces, encode reader intent and composition rules in context, move mechanical choices into reusable CSS, and rerun frozen prompts after every guidance change. Keep human review for hierarchy and deterministic checks for layout failures.

Source: https://vercel.com/blog/how-our-agents-build-on-brand-pages-with-design-md
Feed7: https://feed7.dev/p/how-our-agents-build-on-brand-pages-with-design-md-2f54eb4a2e
Trust: Official Source

## 5. openai/skills

Move future skill discovery and packaging work to the plugin path. Treat skills as reusable bundles of instructions, scripts, and resources, and verify installation guidance against the current documentation.

Source: https://github.com/openai/skills
Feed7: https://feed7.dev/p/openai-skills-4aff7db928
Trust: Needs Review

## 6. From coding to Knowledge work agents — Karan Vaidya, Composio

Treat prompts as guidance, not containment. Put permissions outside the model, log each tool action, test destructive operations against mocked tools, and require review before irreversible effects. The talk cites an outreach agent that sent mass email as instructed but without an adequate preflight check.

Source: https://www.youtube.com/watch?v=xxfMT-bPEmU
Feed7: https://feed7.dev/p/from-coding-to-knowledge-work-agents-karan-vaidya-compos-74c257483c
Trust: Source Linked

## 7. GPT-6 Astra: A new generation of intelligence

Builders should wait for task-level evidence before changing model defaults, then test Astra against their own agent workloads, especially tool use and repository work.

Source: https://openai.com/index/gpt-6-astra
Feed7: https://feed7.dev/p/gpt-6-astra-a-new-generation-of-intelligence-0f3b76bc48
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.