What changed in ECC from 1 to 2

3 releases numbered after v1.10.0 up to and including v2.2.0, stable releases only. v1.10.0 and v2.2.0 are the newest stable releases of 1 and 2 we track; this page follows them as new ones ship.

41 changes across 3 releases

Added 19

v2.2.0

  • Add guided installer for Claude Code, Codex, and Kimi harnesses
  • Add native Antigravity 2.0 support with installation to .agents/{rules,workflows,skills,agents}
  • Add experimental, opt-in Nasiko CLI lifecycle bridge for inspecting, installing, and removing Nasiko v0.1.0
  • Add 68 agents, 286 skills, and 94 command shims in the shipped distribution
  • Add Pi, Itô, agent evaluation, multi-model council, dev-team, living docs, secure terminal, TasteForge, and skill-comply workflows

v2.1.0

  • Plan Canvas feature to review agent plans visually in the browser with the ability to click elements, select text, attach numbered annotations, and approve or request changes
  • Kimi harness with direct installation into Kimi Code via --target kimi flag
  • ecc ito find command as an opt-in bridge to Itô CLI for GPU capacity discovery
  • Hermes and OpenClaw install targets
  • Codex ECC navigation guide to find the right skill and command surface from inside Codex
  • GateGuard path exemptions via GATEGUARD_EXEMPT_GLOBS environment variable
  • Configurable instinct injection with count and confidence threshold settings

v2.0.0

  • 261 public skills across coding, research, security, media, enterprise ops, and agent workflows
  • 64 agents and 84 commands
  • Control-pane substrate with harness-neutral session adapters covering Claude Code, Codex, OpenCode, and dmux
  • MCP inventory with fragmentation/drift detection and secret redaction
  • Worktree-lifecycle service for deterministic merge-conflict prediction and safe garbage collection for parallel agent worktrees
  • Orchestrator skill family with dynamic workflow team orchestration for multi-agent fan-out
  • Rollout-derived optimization pack including parallel-execution-optimizer, benchmark-optimization-loop, data-throughput-accelerator, latency-critical-systems, and recursive-decision-ledger
Changed 9

v2.2.0

  • Improve installer to check every selected harness before writing files and apply setup as one operation
  • Improve release pipeline to require the exact npm archive to pass lifecycle tests on Linux, macOS, and Windows before latest tag moves
  • Improve uninstall reliability by maintaining a complete ownership ledger for selective installs
  • Improve migrations by preserving edited or untracked files during cleanup for OpenCode and Codex

v2.1.0

  • PostToolUse hooks consolidated into sync and async dispatchers with fewer processes per tool call
  • Plan Canvas renders Mermaid diagrams, tables, and task lists natively with live-reload on file edits
  • Pre-commit secret scan now catches Anthropic API keys with sk-ant- prefix

v2.0.0

  • Single-connector MCP default policy now includes only chrome-devtools with six other connectors moved to opt-in
  • Smaller OpenCode surface in default install
Fixed 9

v2.2.0

  • Fix legacy Antigravity ownership migration during installation

v2.1.0

  • Community fixes across OpenCode, Windows, bun lockfiles, the dashboard, and project detection

v2.0.0

  • Plugin hooks were silently no-ops on Node 21+ due to require.main dependency under node -e
  • Windows reliability with CLAUDE_PLUGIN_ROOT normalization, stdin prompt passing, and symlink/chmod test guards
  • Security issue where curl credentials could be exposed in argv
  • Session-end $-sequence corruption in sessions
  • Project-detect boundary matching correctness
  • Install manifest gaps
  • Legacy shim truncation
Removed 2

v2.0.0

  • Six MCP connectors retired from default set to opt-in after June 2026 audit
  • rules/zh removed from default install
Security 2

v2.2.0

  • Add SHA-256 verification for Nasiko CLI downloads

v2.1.0

  • Supply-chain hardening with strict vetting of the installer runtime

Original release notes, newest first

The list above is our reading of these notes; the originals from affaan-m are here, one fold per release.

v2.2.0ECC 2.2.0: Guided Setup, Antigravity 2.0, and the Nasiko CLI Bridge

ECC 2.2 is here, led by one guided installer for Claude Code, Codex, and Kimi. It also adds native Antigravity 2.0 support and an experimental, opt-in Nasiko CLI lifecycle bridge.

The release pipeline is stricter too: the exact npm archive must pass lifecycle tests on Linux, macOS, and Windows before latest moves.

One guided path into ECC

ECC 2.2 terminal welcome after a successful install

Start the guided installer with:

npx --yes ecc-universal@2.2.0 install --guided

The installer checks every selected harness before writing files, then applies the setup as one operation. Provider-owned model and authentication settings stay under the provider's control. The terminal welcome appears only after a successful interactive install.

Nasiko: experimental, opt-in, and explicit

Nasiko's open-source agent dashboard, shown as product context for the ECC CLI lifecycle bridge

ECC 2.2 can inspect, install, and remove one pinned Nasiko v0.1.0 CLI build:

npm install -g ecc-universal@2.2.0
ecc nasiko status --json
ecc nasiko install --version v0.1.0 --dry-run --json
ecc nasiko install --version v0.1.0 --yes --json

Downloads are content-addressed and SHA-256 verified, status is read-only, and uninstall protects modified or unqualified binaries. The dashboard above is Nasiko's own product. ECC does not connect agents, start a server, enable telemetry, share data, or operate that control plane. The bridge remains experimental and opt-in.

Learn more in the Nasiko repository.

More reliable across every harness
  • Native Antigravity 2.0: ECC installs to .agents/{rules,workflows,skills,agents} and safely migrates legacy ownership.
  • Reliable uninstall: repeated selective installs keep a complete ownership ledger, so unchanged ECC-managed files can be removed cleanly.
  • Safer migrations: OpenCode uses ~/.config/opencode, while OpenCode and Codex preserve edited or untracked files during cleanup.
  • More workflows: Pi, Itô, agent evaluation, multi-model council, dev-team, living docs, secure terminal, TasteForge, and skill-comply all expand the shipped distribution.

ECC 2.2 ships 68 agents, 286 skills, and 94 command shims, plus the rules, hooks, memory, continuous-learning, and installation surfaces around them.

Release assurance

The tagged commit must match origin/main. One npm archive is packed, hashed, and tested on all three operating systems, then published under a staging tag. Registry integrity must match before latest moves. The prior 2.1.0 package remains the immediate rollback target.

Install or upgrade

For the native Claude Code plugin:

/plugin marketplace add https://github.com/affaan-m/ECC
/plugin install ecc@ecc

Existing Claude installs can run /plugin update ecc.

Thank you

A special thank you to returning contributors @gaurav0107, @Seekers2001, and @t0kubetsu for substantial work on learning workflows, collaborative skills, and shared testing.

Among our first-time contributors, @renan-tiberio brought Pi support, @romanclaudersoai1 repaired /evolve, and @djpjronline-netizen hardened hook parsing.

Eighteen people made their first ECC contribution in 2.2. Thank you to everyone who tested the installer, reported issues, and helped harden the release across harnesses and operating systems.

New contributors

Join us in the ECC Discord.

Full changelog: https://github.com/affaan-m/ECC/compare/v2.1.0...v2.2.0

View originalPermalink

v2.1.0ECC 2.1.0: Plan Canvas, Kimi Harness, and Self-Hosted Compute

ECC 2.1 turns plan review into a visual loop and opens the harness to self-hosted models. Plan Canvas lets you review agent plans in the browser. Point at the part you mean instead of retyping it in chat. A new Kimi Code install target and a verified Itô GPU path make ECC + open-source models a first-class setup, backed by our public sponsors: Moonshot AI (Kimi), Itô, and Atlas Cloud.

Plan Canvas: review plans by pointing, not retyping

Plan Canvas demo: reviewing an ECC plan, attaching an annotation, chatting with the agent, and approving the plan

Download the MP4 demo

/plan ends with a confirm gate, and until now that review was a wall of markdown in the terminal. Now the agent opens the plan in a loopback-only browser canvas:

  • Click elements or select text to attach numbered annotations
  • Chat with the agent from a side rail while it works in the terminal
  • Approve plan / Request changes buttons map directly onto /plan's CONFIRM gate
  • Mermaid diagrams, tables, and task lists render natively; file edits live-reload the page
  • Model- and harness-agnostic: a plain CLI + JSON protocol (ecc-plan-canvas), no Claude-only dependency
Kimi harness: Moonshot AI partnership

ECC now installs directly into Kimi Code (--target kimi). Kimi Code discovers the installed .kimi/AGENTS.md instructions and .kimi/skills/ workflows natively:

bash ./install.sh --target kimi --profile minimal
npx ecc doctor --target kimi
kimi
Self-host on Itô GPUs

Run ECC against any self-hosted open-source model. If you need GPU capacity, Itô is ECC's preferred compute sponsor. The integration shipped guarded end to end:

  • ecc ito find: opt-in bridge to the canonical Itô CLI that submits a live, authenticated RFQ (it does not reserve capacity)
  • Guarded live node qualification and read-only compute handoff
  • Credential-bearing CLI shims are rejected outright

The sponsorship link itself is passive: it does not invoke an RFQ, reserve capacity, provision compute, or configure serving. Managed inference through Itô is not live yet. Any GPU provider works, and ECC stays provider-agnostic.

Partners

Moonshot AI (Kimi), Itô, and Atlas Cloud are now public sponsors of ECC. The README documents a recommended self-host path: Itô for GPU capacity, a Kimi checkpoint served behind a compatible endpoint, then Kimi Code + ECC on top. Each choice remains separate and swappable.

Also in 2.1
  • Hermes and OpenClaw install targets: two more harnesses join Claude Code, Codex, OpenCode, Cursor, Gemini, Zed, Copilot, and Kimi
  • Codex ECC navigation guide: find the right skill/command surface from inside Codex
  • GateGuard path exemptions (GATEGUARD_EXEMPT_GLOBS) and configurable instinct injection (count + confidence threshold)
  • PostToolUse hooks consolidated into sync/async dispatchers, with fewer processes per tool call
  • Supply-chain hardening: the installer runtime passes strict vetting, and the pre-commit secret scan now catches Anthropic API keys (sk-ant-...)
  • A long tail of community fixes across OpenCode, Windows, bun lockfiles, the dashboard, and project detection

The catalog now stands at 67 agents, 281 skills, and 94 command shims (2.0.0 shipped 64/261/84), plus hooks, rules, memory, continuous learning, and AgentShield.

Install or upgrade
/plugin marketplace add https://github.com/affaan-m/ECC
/plugin install ecc@ecc

Existing installs: /plugin update ecc

Community

Join the ECC community for release announcements, questions, and Show and Tell: https://discord.gg/36yGMHGFbR

Full changelog: https://github.com/affaan-m/ECC/compare/v2.0.0...v2.1.0

New Contributors

Full Changelog: https://github.com/affaan-m/ECC/compare/v2.0.0...v2.1.0

View originalPermalink

v2.0.0ECC 2.0.0 — The Agent Harness Operating System

ECC 2.0.0 is the stable graduation of the 2.0 line: ECC as a cross-harness operating system for agentic work. Claude Code stays first-class; Codex, OpenCode, Cursor, Gemini, Zed, and terminal-only workflows share the same skills, rules, hooks, MCP conventions, release gates, and operator workflows. This is the months-in-the-making release the rc.1 candidate was building toward.

The control pane (early build)

ECC Control Pane — sessions, metrics, and the work-items board

Operator column — knowledge recall, connectors, and runnable agent actions

The ECC 2.0 operating surface, running locally (node scripts/control-pane.js): operator recall search, live session metrics, a work-items board with ready/running/blocked/done lanes, and an operator column that drives knowledge sync/recall, graph backfill, and the TUI. The session adapters and MCP inventory below feed this board. Build-in-public continues in the Discord #control-pane channel.

What is new
  • 261 public skills across coding, research, security, media, enterprise ops, and agent workflows (rc.1 shipped 243) — plus 64 agents and 84 commands.
  • The control-pane substrate — the foundation of the ECC 2.0 operating surface:
    • harness-neutral session adapters (ecc.session.v1) covering Claude Code, Codex, OpenCode, and dmux — one schema for "which agent is where, doing what"
    • MCP inventory (ecc.mcp.v1) — one normalized view of every MCP server config across harnesses, with fragmentation/drift detection and secret redaction (it caught a real arg-carried key leak during development)
    • worktree-lifecycle service — deterministic merge-conflict prediction and safe GC for parallel agent worktrees
  • orch-* orchestrator skill family plus dynamic workflow team orchestration — multi-agent fan-out as a first-class surface.
  • Single-connector MCP default policy — the default set is now exactly one connector (chrome-devtools); the other six retired to opt-in after a June 2026 audit. Policy + per-connector rationale: docs/MCP-CONNECTOR-POLICY.md.
  • Rollout-derived optimization pack: parallel-execution-optimizer, benchmark-optimization-loop, data-throughput-accelerator, latency-critical-systems, recursive-decision-ledger.
Hardening since rc.1 (~30 PRs)
  • Plugin hooks were silently no-ops on Node 21+ (#2184) — the hook runner depended on require.main under node -e; every plugin hook exited cleanly without running. If you are on Node 21+, update now.
  • Windows reliability: CLAUDE_PLUGIN_ROOT normalization (#2139), stdin prompt passing (#2174), symlink/chmod test guards (#2171, #2176).
  • Security: curl credentials out of argv (#2175), gateguard destructive-checkout gating (#2158) + env knobs (#2161), advisory intake hardening.
  • Correctness: session-end $-sequence corruption (#2180), project-detect boundary matching (#2181), install manifest gaps (#2172), legacy shim truncation (#2167).
  • Slimmer defaults: smaller OpenCode surface (#2140), rules/zh out of the default install (#2170).
The ECC Discord is live

https://discord.gg/36yGMHGFbR

  • #announcements — releases auto-post and pin there (this very release was announced by the workflow shipped in it, #2201)
  • #pr-and-issues — live feed of every PR, issue, and release from this repo
  • the ECC bot (built in this release, dependency-free gateway client) answers in-server:
    • /skill name:<query> — look up any of the 261 skills
    • /docs query:<terms> — search the docs
    • /release — latest release
    • /ecc, /help
  • #feedback / #feature-requests — read directly by the maintainer; #control-pane — build-in-public on the ECC 2.0 operating surface
  • 16 channels, 6 categories, community roles for Contributors, ECC Tools users, and AgentShield users
Install or upgrade
/plugin marketplace add https://github.com/affaan-m/ECC
/plugin install ecc

Existing installs: /plugin update ecc

Download
Verification

This release passed the full local validation gate before tagging: unicode safety, agent/command/rule/skill/hook validators, install-manifest checks, catalog and command-registry checks, and the complete test suite (SUITE EXIT: 0), plus markdownlint and the pre-push verification gate on every commit.

Credits

Authored by @affaan-m, built with Claude Code — including this release's Discord community, bot, and announce automation, which shipped themselves end to end.

View originalPermalink