Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/ReflexioAI/claude-smartWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/plugins/reflexioai/claude-smart/plugin)<a href="https://agentmods.dev/plugins/reflexioai/claude-smart/plugin"><img src="https://agentmods.dev/badge/plugins/reflexioai/claude-smart/plugin/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/plugins/reflexioai/claude-smart/plugin"><img src="https://agentmods.dev/badge/plugins/reflexioai/claude-smart/plugin.svg" alt="Reviewed on agentmods" width="80" height="20"></a>Grade A, and why
claude-smart scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
{
"name": "claude-smart",
"version": "0.3.11",
"description": "Self-improving Claude Code, Codex, and OpenCode plugin that turns corrections into Preferences, Project-specific skills, and Shared skills",
"author": {
"name": "Yi Lu"
},
"keywords": [
"claude",
"claude-code",
"claude-code-plugin",
"codex",
"codex-plugin",
"opencode",
"opencode-plugin",
"memory",
"agent-memory",
"plugin",
"reflexio",
"self-improvement",
"self-improving-agents",
"skills",
"learning"
]
}
What it installs
The manifest is a name and a version. 1 skill, 5 commands, 7 hooks travel with it, and installing the plugin installs all of them — 155 tokens a session between them. Each is measured on its own page, and each can be installed alone.
- Skill claude-smart A 69 tokens
- Command learn A 14 tokens
- Command clear-all A 19 tokens
- Command dashboard A 29 tokens
- Command restart A 14 tokens
- Command show A 10 tokens
- Hook SessionStart A not measured
- Hook PostToolUse A not measured
- Hook Setup A not measured
- Hook Stop A not measured
- Hook UserPromptSubmit A not measured
- Hook SessionEnd A not measured
- Hook PreToolUse A not measured
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday Changed 110da60c6360
- 2d ago Changed 9cd8ab2d6151
- 3d ago Changed 27b3b9e658ff
- 5d ago Changed cbb169cc3b27
- 6d ago Changed 56a9abc12415
- 14d ago Changed 8ec98800e4d8
- 15d ago Changed 350b77e83b6f
- 17d ago Changed 3783f3b9fb7a
- 1mo ago First seen · 26 lines scan A 2da1401fe6c3
claude-smart is a plugin published in the GitHub repository ReflexioAI/claude-smart (783 stars, last pushed yesterday), licensed Apache-2.0. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other plugins, from other repositories
recall
Durable, fully-local project memory for Claude Code. Logs every session to .recall/history.md and condenses it into .recall/context.md with a local Python summarizer (TF-IDF + TextRank) — no API key, no external model, nothing leaves your machine. New sessions resume from the saved context.
toonify-mcp
Local context compression for Claude Code workflows, including structured data and debug-heavy output.
captain-memo
Cross-AI local memory layer for Claude Code, Codex, Gemini CLI, Antigravity, goose, Cursor, opencode, Kimi CLI and more: hybrid search, six summarizer providers, hosted or local embedder. Replaces claude-mem.
continuity
Durable project state across AI sessions: decisions, constraints, and rejected paths, versioned in git as plain markdown. Not a memory tool — it answers what is true about the project now, and what must not change.
handoff
Portable project memory + development workflow for coding agents: auto-loads STATUS+RULES each session, enforces folder-level CONTEXT.md on edit, nudges durable state updates at turn end, scaffolds the system into any repo via guided discovery, and ships a tiered 6-phase workflow skill set covering every phase (brief.
memory-loop
A memory lifecycle for Claude Code's native file-based memory: a save gate against hallucinated memories, tiered expiry (short-tier memories archived — never deleted — when they lapse), a propose-then-confirm consolidation pass (merge/dedupe long-tier memory), a correction-signal hook that logs likely corrections for…