Episode "Phase 5 · Review" of the SFEIR series on the augmented SDLC, published **the same day** as the Addy Osmani LinkedIn post that it translates into a phase specification. Thesis: **quality has changed address** — it is no longer read in the code (agents produce more of it than anyone can review) but in **the ring of constraints surrounding the agent**. Osmani's ring (seven dimensions — correctness, security, performance, accessibility, maintainability, **economic efficiency**, **comprehensibility** — linked by the **back-pressure** rule: "a loop is only granted the autonomy that can be verified cheaply and reliably, not an inch more") is redrawn, translated, and attached to phase 5 of SFEIR's 11-phase cycle. The structuring corollary: **the bottleneck has never been generation, it is verification** — "generation is a wide mouth, verification a narrow neck; speeding up the mouth thickens the pile at the neck." **The most interesting design decision is a cycle-architecture choice**: Review is deliberately **outside the three human gates** (Define, Plan, Ship), because making Review the gate would put human attention — a finite resource — as the control point of a generation capacity that itself scales: "you would have built a pipeline whose maximum throughput is the number of diffs a senior can read before the end of the day." Hence the split: **Review instruments, Ship decides** — Review delivers an *opposable body of evidence*, Ship decides on the evidence, not on the full diff. A position staked against Monperrus (from whom SFEIR retains the diagnosis — human inspection of every diff cannot withstand agentic speed — but rejects the conclusion: acceptance cannot be delegated). The named trap is **circular validation** (the agent that writes the code writes the tests that validate it: "you built a mirror, not a ring"), with five countermeasures drawn from Anthropic (independent gates in separate context windows, deterministic + agentic never substituting for one another, shadow mode, risk-based tiering, logging to the SIEM) and Compare the Market's warning (**AST graph ~70% vs vector RAG ~58%**, with RAG performing *worse than no context at all*). The firm's own extension is **the ratchet**: "every escape becomes a constraint" — a defect that has crossed the ring is closed *within the ring* (test, lint rule, review rubric, harness guardrail) at Compound-1, "the only asset in the chain that appreciates while the models depreciate" (an unaudited internal measurement: **−30% fix iterations after ten cycles**). It closes by reformulating the question: "is this code good?" has become unanswerable; what remains is **"what does my system refuse to let through?"**
#ring of constraints#constraints around agents#Review phase
SFEIR (voix éditoriale du cabinet, article non signé individuellement) — construit sur Addy Osmani (Google) ; cite Martin Monperrus · Paula Hingel (Augment Code) · DORA/Google Cloud · Jason Clinton (Anthropic) · l'équipe Engineering de Compare the Market
Doctrinal article by Addy Osmani (Google) that establishes a foundational distinction for the 2026 debate on AI and cognition: **Cognitive Offloading** (healthy — delegating the *how* while retaining judgment over results) vs **Cognitive Surrender** (toxic — accepting AI output wholesale without forming parallel reasoning, *"borrowing the model's confidence as substitute for personal understanding"*). Solid scientific grounding: the **Shaw & Nave (Wharton/UPenn)** study of 1,372 participants — **73% accept demonstrably wrong AI answers**, with confidence rising despite a 50% error rate. **MIT *Your Brain on ChatGPT*** — reduced neural connectivity among AI-assisted writers. **Anthropic Skill-Formation** — engineers using AI to generate code score **17% lower** on comprehension versus those using it for conceptual inquiry. Four concrete examples of surrender (reviewing 600-line PRs on surface signals, shallow debugging, architectural decisions made without reasoning, degraded learning). Five personal heuristics (pre-generating expectations, junior-engineer-standard review, adversarial prompting, fatigue awareness, verification of the source of confidence). Six structural guardrails (verification exit criteria, anti-rationalization tables, **PRs ~100 lines max**, interrogative over generative mode, scaffolded friction, **regular solo keyboard time**). Two new concepts: ***Comprehension Debt*** (the growing gap between total codebase volume and human understanding) and ***Mutual Amplification*** (a cooperative prompt-refine loop vs surrender-delegation). Pivot thesis: ***"the choice between thinking with AI versus not thinking at all remains entirely human"***. A structural and operational counterweight to *"coding is solved"* (Cherny 2026-05) and an analytical complement to Frizzo (2026-05-05).
Addy Osmani (Software Engineer at Google, Cloud + Gemini, ex-Chrome — déjà au dossier veille avec *Agent Harness Engineering* 2026-04-19, *How to write a good spec for AI agents* 2026-01-13, *Conductors to Orchestrators* 2025-11-01).
Google whitepaper (the "Day 1" installment of a series, by Addy Osmani, Shubham Saboo and Sokratis Kartakis) mapping the transformation of the software development lifecycle (SDLC) in the age of coding agents. Thesis: the fundamental shift is not a new language but the move from writing code to **expressing intent**. The document sets out a spectrum ranging from *vibe coding* (prompting and accepting) to *agentic engineering* (AI implements under constraints, tests, and feedback loops designed by humans), with **context engineering** as the central skill, the **software factory** model (the developer's deliverable = the system that produces the code), **harness engineering** (Agent = Model + Harness), and a CapEx/OpEx economic analysis of total cost of ownership.
Synthesis by Addy Osmani (Google, Chrome/Cloud) of the emerging field of *harness engineering*: the equation `agent = model + harness`, the *ratchet* principle ("every mistake becomes a rule"), the HumanLayer "skill issue" reframe, Terminal Bench evidence (Top 30 → Top 5 from a harness change alone), the layered Claude Code architecture, Anthropic's "harnesses don't shrink, they move" vision, and Harness-as-a-Service (Claude Agent SDK, Codex SDK, OpenAI Agents SDK). Pivot article that consolidates Trivedy, HumanLayer, Anthropic, and Böckeler into a doctrine.
#harness engineering#agent harness#Addy Osmani
Addy Osmani (Software Engineer at Google, Cloud + Gemini)