ruflo-intelligence
Reviewed
at commit d96d246
New version, not yet reviewed
Caution: Its skills pre-approve Bash and have Claude run the unpinned npm package @claude-flow/cli@latest through npx, so each new version runs without asking.
Sends data to The IPFS pattern marketplace registry, and Pinata for publishing, only when you browse, download or publish patterns
User-facing surface for Ruflo's self-learning system: 6 neural_* + 10 hooks_intelligence_* + 6 routing/meta hooks + 3 hooks_model-* + 4 SONA/MicroLoRA tools…
0 upvotes
Log in to upvote or save- Made by
- ruvnet
- Pinned commit
-
d96d246e5ccba8ffeb8d0320ae7ec1948b72705b - What we checked
-
Refuses learning writes to the Ruflo intelligence tools when they hold a secret, and asks for confirmation before the intelligence store is reset. Its skills train, route and transfer learned patterns, including a public IPFS pattern marketplace you can browse or publish to. It adds /intelligence-mod and writes a small status file in your project.
Review details
Static review of the source at the pinned commit; not run. Delta review of the changes since 87d8687; the rest carries over from the review at that commit. New in this version: a piiGuard option (default on) that gates only the existing refusal of hooks_transfer calls holding an email address; with it off, such a transfer can be published to public IPFS, while secrets are still refused whenever the guard is on. Otherwise only the version, a repository field and a test change. The hooks module registers session.start (project root, registers /intelligence-mod, writes .claude-flow/intelligence-mod/status.json), tool.call (can only deny: secret patterns in hooks_intelligence_pattern-store, trajectory-step/end, neural_train, hooks_transfer and pattern-namespace memory_store; email addresses in hooks_transfer while piiGuard is on; hooks_intelligence-reset without confirm: true unless confirmReset is off) and command.run for /intelligence-mod (status, scan, stats). The stats verb calls $.mcp.call on the first connected tool whose name ends in hooks_intelligence_stats, read-only, output capped. No network, process or model calls in the hooks. The three skills (intelligence-route, intelligence-transfer, neural-train) pre-approve Bash and tell Claude to run 'npx @claude-flow/cli@latest' unpinned for training, routing, mcp exec, hooks_intelligence-reset and transfer-store commands; they were not changed. intelligence-transfer can download patterns from the marketplace (checksum verified, import only after inspection) and publish to public IPFS using PINATA_API_KEY and PINATA_API_SECRET from the environment, after PII scan and user confirmation per the skill text. Other tools come from the separate ruflo-core MCP server and the @claude-flow/cli package, which were not reviewed, nor were the registry hosts.
- What it uses
-
- Reads your prompts: no
- Sees tool calls: yes
- Blocks or changes tool calls: yes
- Reads or writes files: yes
- Runs programs on your machine: yes
- Calls the model (uses tokens): no
- Sends data over the network: yes
- Stores data: yes
- Shows something on screen: yes
User-facing surface for Ruflo's self-learning system: 6 neural_* + 10 hooks_intelligence_* + 6 routing/meta hooks + 3 hooks_model-* + 4 SONA/MicroLoRA tools (29 total). Implements the 4-step pipeline (RETRIEVE → JUDGE → DISTILL → CONSOLIDATE) and IPFS-based cross-project pattern transfer.
Made this mod? Add a badge to your README
Paste this into your README. The badge says reviewed once the current version has been reviewed, and links readers to this page.
[](https://modsforclaude.com/mods/ruflo-intelligence/)