R139 · Harness tips › Harness tips

Rethinking Claude Code best practices: the five jobs that moved off your plate

Read the original

harness-configagent-harnesssandboxinghuman-oversightagent-reliabilitymulti-agentcode-reviewsdlc

TL;DR

The human-in-the-loop jobs of an agentic coding session (steer, approve, say no, check, start) have each moved to a specific layer: the model, the harness, or the environment. Lydia Hallie (Claude Code team, Anthropic) argues that the engineer's job is now to configure those layers well and review what comes out, not to supervise each step.

The load-bearing ideas for a staff engineer:

Nate Herk's post and video are a faithful, condensed retelling of the talk with a few product details added. The few-sentence version of this doc: CLAUDE.md shrinks to what the code cannot tell the model, irreversibility drives the deny list, and every long run needs pass/fail checks it cannot talk its way past.

Sources and how they relate

The primary source is Lydia Hallie's 17-minute AI Engineer talk; Nate Herk's post and 10-minute video retell it with his own examples. Read the talk if you only read one.

SourceWhoPublishedWhat it adds
Rethinking Claude Code Best Practices (talk, 17:32)Lydia Hallie, Claude Code team, via AI Engineer channelabout 2026-11-12 (re-recorded at home after on-stage technical problems)The argument, the sandbox and cloud-session detail, the review tiers, the orchestration model
Anthropic Just Released the New Rules of Claude Code (video, 9:40)Nate Herk, Uppit AI2026-11-17Clips from the talk, a knowledge-work framing, /auto-mode-setup, /goal, the Projects plan tiers; a gated free guide in his Skool community
LinkedIn postNate Herk2026-11-17The five-item summary; the video link is in his first comment

Fidelity check. Nate's retelling matches the talk on every substantive claim. Differences worth knowing:

One post comment (SJK Systems) adds a useful security footnote: approving a shell call that carries an inline token writes the literal key into .claude/settings.local.json, and it cites GitGuardian counts of 31,088 such secrets on GitHub and 613 in Docker images. Those numbers were not verified here.

Who does what now

Hallie's frame: everywhere the human steps out of the loop, something else must take over, and it is always one of three things: the model, the harness or the environment. The useful split for an engineer is which of those give guarantees.

Every loop job now has an owner; only some give guaranteesYOU USED TONOW HANDLED BYGUARANTEESteer and correctModel (Opus 5.5)Infers intent from the code; CLAUDE.md keepsonly what the code cannot showProbabilisticApprove each commandHarness: auto-mode classifierBlocks uncertain actions; trusts only thisrepo by default; blocked 89% in a studyProbabilisticSay noDeny rules + sandbox or cloud sessionDeny rules run before the classifier; the OSor a secret-free machine limits side effectsDeterministicCheck the outputTests, hooks and CIPass/fail checkpoints; hooks exit 2 to block;Claude reviews PRs first, humans by riskDeterministicStart every sessionProjects coordinatorFable 5.2 splits and judges; Opus 5.5 threadsdo the work; unfinished work goes backModel-judgedGreen rows are enforced by the harness or the operating system; amber rows depend on a model getting it right.
five loop jobs · who owns each now · what kind of guarantee

Read it top to bottom as a session's lifecycle. The two amber jobs are where the remaining risk sits, so they are where deny rules, checks and acceptance criteria earn their keep.

The five jobs in depth

Each job below names what moved, the mechanism, and what the docs say about it.

1. Steering moves to the model

Early-2026 sessions on Opus 4.5 stopped early and needed constant correction, so CLAUDE.md files filled with IMPORTANT and NEVER. On Opus 5.5 and Fable 5.2 that emphasis over-steers: Hallie reports more tool calls and worse results.

2. Approving moves to auto mode

Auto mode (the default since August) sends nearly every action except reading and editing project files through a separate classifier. Tests and installs run; uncertain actions block.

3. Saying no moves to deny rules and the environment

The classifier is a second gate. permissions.deny is evaluated first, and a matching call never reaches the classifier; neither the classifier nor user intent can override it (auto mode config).

4. Checking moves to tests, hooks and CI

Same prompt, different runs: migrating auth to a new API might leave the old API in place, migrate only new logins, or log everyone out. All three look reasonable at a glance.

5. Starting sessions moves to an orchestrator

The remaining human job is kicking off and tracking sessions, which Hallie calls bad for anyone who context-switches poorly.

Suggested first experiment (Hallie): pick one task far bigger than usual, make sure CI catches what you care about, configure the cloud environment (network access, secrets, setup script), run it in the cloud, and note where the session stalls or still needs you. That is the next thing to hand over.

Failure modes and gaps

The talk is a vendor's view of its own best case; these are the places it under-specifies or where teams will get hurt.

Review checklist

Use this when auditing a team's Claude Code setup or a repo before handing it long unattended runs.

Instructions

Permissions and environment

Verification

Orchestration

Related in this index

Sources

Connections

See R139 on the knowledge map

Part of the Research notes on nowucca.com. Figures and claims are the source's own unless the summary says otherwise.