Everything Claude Code (ECC): Governable Agent Workflows

Overview

Everything Claude Code (ECC) is examined as an operating layer around coding-agent harnesses, not as a prompt collection. The series moves from component ownership and bounded context to plans, tests, review, evidence, memory, portability, security, visibility, and conditional autonomy. Its central thesis is that durable artifacts, scoped permissions, event checks, and explicit state make agent work more inspectable and recoverable after interruption. ECC does not guarantee correct code, universal harness parity, productive autonomy, or safe execution of untrusted candidates; each claim remains bounded by its contract and evidence.

Learning outcomes

ECC's core value is recoverable, inspectable agent work, not a guarantee that generated changes are correct.
Plans, tests, fresh-context review, hooks, and records preserve different kinds of evidence at distinct lifecycle boundaries.
Portability preserves shared workflow sources, but adapters can change activation, permissions, events, and enforcement.
Memory, autonomy, and security require explicit scope, visibility, approval, isolation, and stop paths; receipts do not prove safe execution.

Keywords

ECCworkflowssecurityClaude Codeagentscontextevidencememoryportabilityautonomy

Chapters

ECC: Recoverable Agent Workflows

Artifacts, checks, and boundaries that survive the chat

8 min · Artifacts, checks, and boundaries that survive the chat

8 min

ECC Component Boundaries

Choosing ownership, activation, and timing for each layer

9 min · Choosing ownership, activation, and timing for each layer

9 min

ECC Context Budgeting

Choosing what to read, delegate, and retain

10 min · Choosing what to read, delegate, and retain

10 min

ECC Evidence Lifecycle

From reviewed intent to tests, review, and bounded evals

10 min · From reviewed intent to tests, review, and bounded evals

10 min

ECC Workflow Control Model

How ownership, context, evidence, and authority connect

8 min · How ownership, context, evidence, and authority connect

8 min

ECC Memory Vault

Portable handoffs without automatic authority

10 min · Portable handoffs without automatic authority

10 min

ECC Portability and Installation

What crosses harness boundaries and what stays local

9 min · What crosses harness boundaries and what stays local

9 min

ECC Agent Security Boundaries

Least agency, isolation, approval, and observable stop paths

11 min · Least agency, isolation, approval, and observable stop paths

11 min

ECC Autonomy and Visibility

Status, checkpoints, rollback, and safe operator control

10 min · Status, checkpoints, rollback, and safe operator control

10 min

ECC: Evidence and Limits

Adoption decisions for recoverable, governable agent work

11 min · Adoption decisions for recoverable, governable agent work

11 min

Preview every chapter · Unlock full text, audio, and video from chapter 4

Plans

Ch.1-3 FreeCh.4-10 Premium

FAQ

What does Everything Claude Code (ECC) actually add to coding-agent work?

Everything Claude Code (ECC) adds durable workflows, scoped workers, checks, memory boundaries, adapters, and operational records around a coding-agent harness; it does not replace the model or guarantee correctness.

Does ECC make coding agents reliable across every harness?

No. ECC's support ranges from native to adapter-backed, instruction-backed, and reference-only, so hooks, permissions, events, and enforcement can differ by host.

How does ECC keep agent work recoverable after a session changes or ends?

ECC preserves intent, evidence, and unresolved state in plans, tests, reviews, handoffs, memory entries, and session artifacts, so a later operator can inspect what happened instead of relying on a transcript.

Can ECC safely run untrusted agent candidates automatically?

No. ECC records capsules, fixtures, and receipts but keeps candidate execution disabled because an independently reviewed operating-system containment backend is not yet available.