Ruflo Intelligence
插件 活跃User-facing surface for Ruflo's self-learning system: 6 neural_* + 10 hooks_intelligence_* + 9 routing/meta hooks + 4 SONA/MicroLoRA tools (29 total). Implements the 4-step pipeline (RETRIEVE → JUDGE → DISTILL → CONSOLIDATE) and IPFS-based cross-project pattern transfer.
To operationalize a sophisticated self-learning intelligence system, enabling advanced pattern recognition, knowledge consolidation, and cross-project knowledge sharing through a unified interface.
功能
- Implements the 4-step intelligence pipeline (RETRIEVE, JUDGE, DISTILL, CONSOLIDATE)
- Supports cross-project pattern transfer using IPFS
- Integrates 29 specialized neural, routing, and hook tools
- Manages learning state and prevents catastrophic forgetting
- Provides dashboards for metrics and model routing rationale
使用场景
- When needing to leverage learned patterns across different projects or environments.
- For operationalizing a complex, multi-stage AI learning pipeline.
- To analyze and optimize AI model routing decisions.
- When building agentic systems that require continuous self-improvement.
非目标
- Replacing the core LLM execution environment.
- Providing a generic file transfer utility.
- Managing agent deployment outside of learned pattern application.
Documentation
- warning:Configuration & parameter referenceWhile many tools are documented, the `hooks_transfer` tool requires `PINATA_API_JWT` configuration, which is mentioned in the README but not explicitly detailed as a required parameter in the tool's schema or a formal configuration reference.
Security
- warning:Secret ManagementThe `hooks_transfer` tool requires a `PINATA_API_JWT` environment variable, but it's not explicitly routed through `userConfig` with `sensitive: true` in the `plugin.json`, potentially leading to it being stored insecurely in settings.json.
- warning:Data ExfiltrationThe `hooks_transfer` tool requires `PINATA_API_JWT`, and while not explicitly exfiltrating data, the lack of `userConfig` sensitive handling for this secret poses a risk of leakage if not managed carefully.
- warning:Keychain-stored secretsThe `PINATA_API_JWT` required for `hooks_transfer` is configured via environment variable and not explicitly routed through `userConfig` with `sensitive: true`, risking insecure storage in `settings.json`.
Code Execution
- info:ValidationParameter schemas are present, but the explicit use of a validation library like Zod or Pydantic for all inputs and outputs is not clearly demonstrated in the provided snippets.
Scope
- warning:Tool surface sizeThe plugin exposes a total of 29 tools across various families, exceeding the target of 10 and potentially leading to complexity.
Protocol
- info:Idempotent retry & timeoutsWhile the plugin appears to handle errors, explicit documentation on per-call timeouts or idempotency for mutating operations is not readily apparent.
安装
请先添加 Marketplace
/plugin marketplace add ruvnet/ruflo/plugin install ruflo-intelligence@ruflo包含 3 个扩展
Skill (3)
Route tasks via the 3-tier model selector and learned patterns; emits a routing rationale via hooks_explain
Publish or fetch learned patterns across projects via IPFS (Pinata) -- the cross-project pattern transfer that hooks_transfer enables
Train SONA + MicroLoRA neural patterns from successful task completions; runs the DISTILL + CONSOLIDATE phases of the 4-step pipeline
质量评分
类似扩展
Si
100Self-Improving Agent: curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills. Provides /si:review, /si:promote, /si:extract, /si:status, and /si:remember slash commands.
Uc Taskmanager
100SDD WORK-PIPELINE Agent — Requirements analysis & development 6-agent full pipeline with DAG-based orchestration and sliding window context management
Data Validation Suite
99Schema validation, data quality monitoring, streaming validation pipelines, and input validation for backend APIs
Ruflo Agent
99Agent runtimes for ruflo — local WASM-sandboxed agents (rvagent: 10 wasm_agent_*/wasm_gallery_* MCP tools, built on @ruvector/rvagent-wasm + @ruvector/ruvllm-wasm per ADR-070) plus Anthropic Claude Managed Agents as a cloud backend (managed_agent_* MCP tools per ADR-115). One interface, local-vs-cloud runtimes.
Vulnetix
98Vulnerability intelligence and remediation skills for Claude Code — 7 skills for exploit analysis, fix proposals, scoring, exploits, and package security via the Vulnetix VDB API
Socraticode
98代码智能 — SocratiCode 的语义搜索工作流、依赖图分析和上下文伪影探索