--- title: "Who watches the coding agents?" newsletter: "MLOps Community" date: 2026-08-13 source: https://aaif.live/newsletters/mlopscommunity/2026-08-13-who-watches-the-coding-agents --- # Who watches the coding agents? *Plus evolving agent designs, MCP sprawl, and portable plugins* *MLOps Community — Agentic AI Foundation, 2026-08-13* When your agent sounds like a kid: "But you said I could!" [https://thehackernews.com/2026/08/atlassian-rovo-can-be-tricked-into.html] Somebody said it, and it wasn’t you. Our next reading group on 20 August takes that apart as a role confusion problem. Register here [https://home.mlops.community/home/events/prompt-injection-as-role-confusion-rethinking-agent-security-kzdnmrqqqz?agenda_day=6a7a2823abc455f7962051e5&agenda_track=6a7a2823abc455f7962051fc&agenda_stage=6a7a2823abc455f7962051eb&agenda_filter_view=stage&agenda_view=list]. ## Fresh Pair of AIs The best verifier is an agent that did not write the code. Who should check the work? Same context or Fresh context [SAME](https://go.mlops.community/NL_HT1_Aug13) [FRESH](https://go.mlops.community/NL_HT1_Aug13) ## Rules of engagement Close, but more of you back writing better rules over upgrading your model. ## Building voice agents for production ## VOICE AGENTS FORUM // SAN FRANCISCO Latency, interruptions, handoff, evaluation, observability, reliability. Voice agents bring a set of engineering and operational problems that only become more important once people start using them. Join us at Digital Jungle SF on September 16 for a day with engineers, founders, researchers, and operators building and deploying voice systems. Compare how teams are approaching the stack, where technical choices create tradeoffs, and what to consider as voice agents move from prototypes into production. Early Bird tickets end September 1. Register now [https://luma.com/voiceagentssf] ## Curated finds to help you stay ahead ## JOB OF THE WEEK ## 856 tools, 103 servers, one quality score ## LUNCH & LEARN · SESSION 21 Tomorrow, Friday 14 August, 16:00-17:00 UTC. Free and virtual. Om Shree is presenting TDQS (Tool Definition Quality Score), a framework for measuring MCP tool quality that came out of a study of 856 tools across 103 live MCP servers. He'll cover what separates a tool definition an agent can use reliably from one that produces confusing behavior, and what to check in your own tools. Community discussion on the MCP ecosystem follows the talk. Register here [https://go.mlops.community/L&L21_Live] The last session ran on a related theme. Leo Walker walked through Arcade's ToolBench benchmark for grading MCP servers, and why a strong compliance grade still doesn't guarantee a clean connection. John Dellenbaugh demoed the stateless protocol change that lets you scale MCP servers across instances without a sticky gateway or a session store, and Ben Kearns pushed back on what happens to authorization when a client carries its own identifier. Notes from the session [https://go.mlops.community/L&L_20_Asset] ## Why your AI bill will double before it gets better LLM costs are no longer falling fast enough to offset exploding usage, and hidden premiums can turn a viable feature into an expensive mistake. * Cost calculators built into evaluation workflows can flag unscalable experiments before teams spend weeks prototyping. * Token telemetry by service and feature makes spikes, model choices, and experiment costs visible. * Optimization needs researchers, engineers, and finance working from the same latency, quality, and cost data. Teams that price AI early can ship faster and avoid learning the economics after launch. [https://podcasts.apple.com/gb/podcast/why-your-ai-bill-will-double-before-it-gets-better/id1505372978?i=1000779713422](https://podcasts.apple.com/gb/podcast/why-your-ai-bill-will-double-before-it-gets-better/id1505372978?i=1000779713422) [https://home.mlops.community/home/videos/why-your-ai-bill-will-double-before-it-gets-better](https://home.mlops.community/home/videos/why-your-ai-bill-will-double-before-it-gets-better) [https://open.spotify.com/episode/24XVJn7m7VnJ6sZO0Wq7sx?si=1EfQdsUMRVSCN14nxpYApQ](https://open.spotify.com/episode/24XVJn7m7VnJ6sZO0Wq7sx?si=1EfQdsUMRVSCN14nxpYApQ) ## Why Supervising AI Code Is Harder Than Watching a Car Drive Itself Supervising a coding agent may be more cognitively demanding than watching a self-driving car because its intent and acceptable end state are far harder to observe. * Self-driving works within defined inputs, goals, and physical constraints; software development is open-ended, with much of its intent missing from the code. * Graph-based workflows can run independent agent tasks in parallel, then use fresh context to verify and merge results. * Growing skill libraries need evaluation, pruning, and reliable selection across models. Better agents still demand better ways to supervise their reasoning. Video [https://www.youtube.com/watch?v=J4wBMQLbJ9M&list=PLfUTNNsSsv89oCGrrvAy_aLktaI0fSYTk] · Spotify [https://open.spotify.com/episode/2wtAPiLCBsY0u165fLZeh2?si=7m8F12oASLmNRUa7-lT02A] ## AI Hype vs. Real Value AI projects can look productive until token and compute bills erase the gains. The discussion argues that real value comes from redesigning complete workflows, rather than scattering pilots across the business. * Compare AI-assisted teams with control groups and measure changes in speed, cost, revenue, or output. * Match models to tasks, limit token use, and apply FinOps at workflow level. * Keep humans involved in sensitive decisions, then reduce oversight as accuracy improves. AI pays off when workflow gains survive cost and risk checks. [https://home.mlops.community/home/videos/ai-hype-vs-real-value](https://home.mlops.community/home/videos/ai-hype-vs-real-value) [https://open.spotify.com/episode/6QLuRQLZQN8A6Jm3L1I823?si=moS-J2ssQf-qb7ATzq0E5Q](https://open.spotify.com/episode/6QLuRQLZQN8A6Jm3L1I823?si=moS-J2ssQf-qb7ATzq0E5Q) ## Agent evolutions: Stop guessing the design - evolve it The winning design cut CLI startup from 142.0 ms to 98.6 ms, yet it did not exist when the search began. * A genetic coding-agent loop generates diverse approaches, keeps strong performers, and mutates or combines them across generations. * Binary gates exclude broken variants, while a frozen numeric rubric prevents the scoring criteria shifting after results appear. * File-based results preserve audit trails, survive crashes, and keep parallel sub-agents isolated. Measured evolution can find designs that intuition and one-shot sampling miss. [Read the blog](https://mlops.community/blog/agent-evolutions-stop-guessing-the-design----evolve-it) ## The enterprise MCP scaling paradox: why every agent needs a tailored backend Successful MCP adoption could leave an enterprise running hundreds of servers, each adding compute, maintenance, security review, and on-call responsibilities. * Broad vendor tool catalogs increase token use, latency, and attack surface while exposing far more capabilities than individual agents need. * Internal servers fragment identity, deployment, and telemetry unless teams share a clear platform contract. * Gateways, generated tool definitions, and agent-specific backends each reduce some duplication but introduce their own tradeoffs. MCP scale depends on governing the full estate before server sprawl sets in. [Read the blog](https://aaif.io/blog/the-enterprise-mcp-scaling-paradox-why-every-agent-needs-a-tailored-backend) ## From Skills and Tools to Portable Agent Plugins A deployment assistant combining reusable instructions and live tools can now travel across compatible agent clients without being repackaged for each one. * Agent Plugins 1.0 places native Agent Skills and MCP configuration inside one predictable directory structure and manifest. * Client-specific features remain in namespaced folders that unsupported clients can safely ignore. * Installation, permissions, trust, distribution, and runtime behavior stay under each client’s control rather than being forced into the shared format. The result is a narrow portability layer that preserves room for clients to differ. [Read the blog](https://aaif.io/blog/from-skills-and-tools-to-portable-agent-plugins) ## IN-PERSON EVENTS * New York [https://luma.com/kylt79cf] - August 14 * Silicon Valley [https://luma.com/ey4era2u] - August 27 * Melbourne [https://luma.com/et8icca1] - September 2 * Shanghai [https://luma.com/m4ygi0b2] - September 5 * AGNTCon + MCPCon China [https://www.lfopensource.cn/mcp-dev-summit-shanghai/] - September 6 * Toronto [https://luma.com/p6vg7zag] - September 10 * AGNTCon + MCPCon Japan [https://events.linuxfoundation.org/agntcon-mcpcon-japan/] - September 10 * Pune, India [https://luma.com/4h0mjfzr] - September 12 * New York [https://luma.com/aaif-kc6u] - September 15 * San Francisco, Voice Agents Forum [https://luma.com/voiceagentssf] - September 16 ## MCP RELEASE PARTIES * Amsterdam [https://luma.com/xpbgu0t7] - July 29 * London [https://luma.com/klu2tegz] - July 28 * New York [https://luma.com/aaif-zwdl] - July 28 * San Francisco [https://luma.com/ugfl1mn2?tk=C9eRhQ] - July 28 * Seattle [https://luma.com/aaif-znfn] - July 28 ## VIRTUAL EVENTS * Coding Agents Lunch & Learn Session 21 [https://home.mlops.community/home/events/coding-agents-lunch-and-learn-session-21-evaluating-mcp-tools-for-real-world-agent-workflows-7tfxwl1r8t?agenda_day=6a743839b54ca3c602a460bd&agenda_track=6a74383ab54ca3c602a460d4&agenda_stage=6a743839b54ca3c602a460c3&agenda_filter_view=stage&agenda_view=list] - August 14 * Reading Group: Prompt Injection as Role Confusion [https://home.mlops.community/home/events/prompt-injection-as-role-confusion-rethinking-agent-security-kzdnmrqqqz?agenda_day=6a7a2823abc455f7962051e5&agenda_track=6a7a2823abc455f7962051fc&agenda_stage=6a7a2823abc455f7962051eb&agenda_filter_view=stage&agenda_view=list] - August 20 ## A Discussion on Loop Engineering A coding loop can launch 75 agents and burn through its token budget in minutes if its controls are wrong. Making agents autonomous requires more than repeating prompts until the tests pass. * Machine-readable plans, tests, and independent reviewers give each loop a measurable target. * Observability and evals show where systems drift, stall, or waste tokens. * Persisted failure modes and learnings let later runs avoid repeating the same mistakes. The useful loop is the one that knows when to stop. [Watch the reading group](https://home.mlops.community/home/videos/a-discussion-on-loop-engineering-aaif-reading-group-july) ## Predicting the Past [https://docs.google.com/forms/d/e/1FAIpQLScLyUmV7B_fvmiQni6cF0iIFMEvjrL5fivzUjdFci14G_nQCA/viewform](https://docs.google.com/forms/d/e/1FAIpQLScLyUmV7B_fvmiQni6cF0iIFMEvjrL5fivzUjdFci14G_nQCA/viewform) --- Source: https://aaif.live/newsletters/mlopscommunity/2026-08-13-who-watches-the-coding-agents