The ecc project positions itself as an operating system layer for AI agent harnesses, potentially unifying execution, sandboxing, and orchestration primitives.
state: seedheat: lowuncertainty: mediumconvergesscott: highagent-harnesses agent-orchestration agent-sandboxing agent-os-layer ecc-projectaffaan-mhochmartinez
What is this?
ECC (Everything Claude Code / ECC Tools) is an open-source project (github.com/affaan-m/ECC, 200K+ stars) that positions itself as a cross-harness 'operating system for AI harnesses.' It provides a unified operational layer above coding agents (Claude Code, Codex, OpenCode, Cursor, Gemini, Zed, terminal workflows) with 261+ skills, 64+ agents, an orchestrator family (orch-), a control-plane substrate with session adapters, MCP inventory, and worktree-lifecycle service. ECC 2.0.0 shipped June 2026. The project also includes AgentShield for security scanning and a GitHub App for PR automation. Key maintainers appear to be affaan-m (repo owner) and hochmartinez. The snippets consistently frame ECC as a harness-native operator system — not a replacement for the underlying tools but a unification layer for configuration, orchestration, and governance.
Why it matters to Scott
ECC is a high-adoption (200K+ stars) open-source implementation of the exact cross-harness orchestration stack Scott has been architecting: deterministic control plane (router-as-kernel), stateless workers with external state (long-running-agents), capability-scoped sandboxing (SiloOS/padded-cell), skills/agents as version-controlled artifacts (soft-weights/Markdown OS), MCP as transport not control plane (code-first/hybrid architecture), and policy-as-code gates (decision-authority-infrastructure). It independently arrives at the same architectural doctrine — machine-native middle, human-legible boundaries, hard authority underneath — across the full harness fleet (Claude Code, Codex, OpenCode, Goose, etc.) Scott actively uses. This is not merely an example of a pattern he believes in; it is a concrete, widely-used reference implementation that bears directly on his OpenClaw, Superlever, and All-in-one stacks, his deterministic-agent-control-plane concept, and his consulting practice's governance/sandboxing deliverables.
ip:framework.agent-native-computingip:framework.code-first-architectureip:framework.micro-agents-architectureip:framework.long-running-agentsip:framework.siloosip:framework.decision-authority-infrastructureip:framework.separation-of-powers-for-cognitionip:framework.markdown-osip:concept.model-plus-harness-benchmark-unitip:concept.router-as-kernelip:concept.goal-world-isolationip:concept.skills-and-workflowsip:concept.soft-weightsip:concept.policy-as-source-codeip:concept.sandboxed-executionip:concept.durable-external-stateip:concept.claude-md-patternip:concept.spec-as-assetdev:concept.deterministic-agent-control-planedev:project.openclawdev:project.superleverdev:project.all-in-one-softwaredev:concept.resumable-agent-job-control-planedev:technology.claude-codedev:technology.codex-clidev:technology.opencodedev:technology.goosedev:technology.mcpdev:concept.padded-cell-agent-architectureradar:49-ide-agent-fleet-workspaceradar:agentstow-coding-agent-config-fanoutradar:agentrove-parallel-coding-workspaceradar:apowerb-open-agent-runtimeradar:agent-acid-rollback-guardrailsradar:acs-local-skill-risk-catalogradar:fakegreen-ci-deception-detectorradar:four-model-terminal-bench-backfireradar:github-copilot-production-trace-findingsradar:anthropic-multiagent-patterns-validation
queries asked of Scott's wikis
- agent-harness orchestration primitives and cross-harness unification patterns
- local-first control plane for agent session management and worktree lifecycle
- MCP (Model Context Protocol) inventory and cross-harness MCP configuration
- agent sandboxing and security policy as code (AgentShield, declarative policies)
- open-weight/local model agent infrastructure and harness compatibility
- agent infrastructure as code: skills, agents, rules as version-controlled artifacts
Measured heat
now 0 pts/hpeak 4 pts/hcomments 0/hpeers p16momentum: steady2 platformsage 99h
points/hour across evidence · reading as of 2026-10-12 02:59:37.977291+11:00 · deterministic, not a model opinion
How the heat travelled
pace: p33 vs 1247 stories at the 96h mark (now 99h old) — ahead of 3jsbench-llm-3d-generation-benchmark (1.5x), behind agent-memory-add-search-evaluation (0.8x)
Evidence (2) — ⭐ canonical anchor
Interpretation history
2026-10-09T01:26:08Z
grounded: converges/high — ECC is a high-adoption (200K+ stars) open-source implementation of the exact cross-harness orchestration stack Scott has been architecting: deterministic contro
2026-10-09T01:15:27Z
case created — HN post announces GitHub repo explicitly framed as 'the operating system for AI harnesses', directly targeting hot agent-harnesses/orchestration space with first-party artifact.
Decision trace
- 10-09 13:41feedback_briefingScott vote via UI
- 10-09 13:26attention_communicatedaffaan-m's ECC project (200K+ GitHub stars) frames itself as 'the operating system for AI harnesses' and independently implements Scott's full orchestration doctrine: deterministic
- 10-09 13:26attention_routeHigh-adoption reference implementation of Scott's architecture across the full harness fleet he uses (Claude Code, Codex, OpenCode, Goose). Adoption signals will accumulate rapidly; this is a con
- 10-09 13:18attention_routeHigh-adoption reference implementation of Scott's architecture across the full harness fleet he uses (Claude Code, Codex, OpenCode, Goose). Adoption signals will accumulate rapidly; this is a con
- 10-09 13:11attention_candidatecreate
- 10-09 12:26groundECC is a high-adoption (200K+ stars) open-source implementation of the exact cross-harness orchestration stack Scott has been architecting: deterministic control plane (router-as-kernel), stateless wo
- 10-09 12:15createHN post announces GitHub repo explicitly framed as 'the operating system for AI harnesses', directly targeting hot agent-harnesses/orchestration space with first-party artifact.