claude-reflect vs ECC
Claude Code plugin that learns from your corrections — hooks capture them in-session, /reflect syncs approved learnings to CLAUDE.md/AGENTS.md, /reflect-skills mines history into reusable commands. — versus — Cross-harness 'operating system' for coding agents — 268 skills, 66 agents, hooks, rules, memory persistence, instinct-based continuous learning and AgentShield security scanning. MIT.
Both close the learning loop from your corrections: claude-reflect is a focused Claude Code plugin syncing approved learnings to CLAUDE.md; ECC's instinct system does the same continuous-learning job with confidence scoring inside a much larger harness framework.
| claude-reflect | ECC | |
|---|---|---|
| Stars | 1.3k | 230k |
| Forks | 109 | 35k |
| Language | Python | JavaScript |
| License | MIT | MIT |
| Last activity | 4 months ago | 3 days ago |
| Topics | coding, memory | coding, skills |
| Curated connections | 9 | 3 |
claude-reflect — the curator's take
Install claude-reflect the third time you catch yourself typing the same correction into Claude Code. It's the pragmatic take on agent memory: no vector DB, no service — hooks queue corrections, you review, markdown files get smarter, and the AGENTS.md sync means Codex/Cursor/Aider benefit too. The /reflect-skills pattern-mining is the sleeper feature: 15 similar requests become one command. When NOT: if you expect actual memory infrastructure (semantic recall, knowledge graphs) — this is disciplined note-taking with AI triage, personal-scale by design. Everything lands via human review, which is a feature, not friction.
ECC — the curator's take
The kitchen-sink option, and the most popular one by a wide margin: one install gives Claude Code, Codex, Cursor, OpenCode and friends a full operating layer — session-memory hooks, learned instincts, quality gates, orch-* orchestrator commands, worktree lifecycle, security scanning — plus three genuinely good guides on token optimization, memory and agentic security. When NOT: it's maximalist. Hundreds of skills and global hooks add context weight, and the README's own top warning is about broken stacked installs — start with the minimal/no-hooks profile and never mix the plugin with the manual installer. If you only want one capability, take the surgical tool instead: claude-reflect for learning-from-corrections, pro-workflow for session memory, asm or skillkit for skill management. Watch the upsell surface (Pro app, sponsors) — the OSS core is MIT and complete.