← 返回 Skills 市场

Full Potential Intelligence

Name: Full Potential Intelligence
Author: jamessunheart

作者 jamessunheart · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

106

总下载

当前安装

版本数

在 OpenClaw 中安装

/install fp-intelligence

功能描述

Provides real-time AI frontier intelligence with scores, live feeds, labor displacement, investment insights, gap opportunities, and daily briefings.

使用说明 (SKILL.md)

Full Potential Intelligence

Real-time AI frontier intelligence network via MCP.

What it does

Scans 18+ sources every 30 minutes and provides:

FP Line Score — composite AI capability measure (0-100) across 14 domains
Intelligence Feed — live entries from AI frontier with impact scores and domain tags
Labor Displacement — 25 job categories tracked against BLS data with gap velocity
Investment Allocation — 13-sector AI frontier basket with momentum signals
Gap Opportunities — ranked by composite score with build assessments
Daily Briefing — Claude-synthesized summary of what changed and why it matters
Agent Economy — register, contribute field reports, earn CORA Credits, spend on metered services

MCP Connection

{
  "mcpServers": {
    "full-potential-intelligence": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://fullpotential.ai/mcp", "--transport", "sse-only"]
    }
  }
}

Tools (12)

Free (no auth):

get_fp_line — Current FP Line score with domain breakdown
get_latest_feed — Latest intelligence entries, filterable by domain
get_displacement_gap — Displacement data for any of 25 job categories
get_allocation — 13-sector frontier basket allocation
get_opportunities — Ranked gap opportunities
get_daily_briefing — Today's Claude-synthesized briefing

Write (free, needs API key):

register_agent — Self-register, get API key instantly
contribute_intelligence — Submit field reports, earn credits

Metered (costs credits):

frontier_scan — On-demand topic scan (5 credits)
capability_check — "Can AI do X?" assessment (3 credits)
dark_ai_check — Adversarial AI check (2 credits)
build_assessment — "Should I build X?" evaluation (10 credits)

Resources (8)

fp://intelligence/feed — Live intelligence feed
fp://intelligence/fp-line — FP Line score
fp://intelligence/briefing — Daily briefing
fp://displacement/overview — Labor displacement overview
fp://invest/allocation — Frontier basket allocation
fp://opportunities/ranked — Gap opportunity rankings
fp://economy/constitution — Agent Constitution
fp://economy/status — Economy status

安全使用建议

This skill will try to run 'npx mcp-remote' and open a live connection to https://fullpotential.ai/mcp and can issue/receive commands and issue API keys via self-registration. That means code will be downloaded and executed from the network and a remote server could influence the agent or receive data. Before installing: (1) verify the skill's source and maintainers (the registry shows 'unknown' and no homepage), (2) ask the author for the mcp-remote package name, its npm link, and a reproducible install method or source code, (3) ensure 'npx' is present or require the package be vendor-checked rather than fetched at runtime, (4) consider running the skill in a strictly sandboxed environment or rejecting it if you cannot verify the remote host and package, and (5) beware that contributing intelligence may send user content to the remote service and that the skill can create API keys that the agent will possess. Given those inconsistencies and the dynamic code execution, proceed only if you trust the publisher and can audit the remote package and server.

功能分析

Type: OpenClaw Skill Name: fp-intelligence Version: 1.0.0 The skill bundle provides a legitimate interface to a remote AI intelligence platform via the Model Context Protocol (MCP). It uses `npx` to execute the `mcp-remote` utility to connect to a specified endpoint (https://fullpotential.ai/mcp), which is consistent with its stated purpose of delivering real-time frontier intelligence and labor displacement data. No evidence of malicious intent, data exfiltration, or prompt injection was found in SKILL.md or the configuration.

能力评估

⚠ Purpose & Capability

The stated purpose (real-time AI intelligence) could legitimately require contacting a remote service and providing feed data, but the SKILL.md relies on running 'npx mcp-remote' at runtime. The skill metadata declared no required binaries or credentials, yet the instructions effectively require the 'npx' binary and will obtain API keys via self-registration. That mismatch (runtime needs not declared) is an incoherence.

⚠ Instruction Scope

The instructions instruct the agent to execute an npm/X command that will fetch and run remote code and open an SSE connection to https://fullpotential.ai/mcp. They also expose write operations (register_agent, contribute_intelligence) that create API keys and accept user-contributed reports. This grants the remote service broad ways to interact with the agent and could result in transmitting user data or enabling remote control — scope exceeds the simple 'read a feed' model the description might imply.

⚠ Install Mechanism

There is no formal install spec, but the SKILL.md explicitly invokes 'npx -y mcp-remote ...', which means code will be downloaded and executed at runtime from npm (or whatever npx resolves to). That dynamic fetch/execute pattern is high-risk because the package provenance and contents are not declared, and the MCP endpoint is an unknown host.

⚠ Credentials

The registry metadata declares no required environment variables or primary credential, yet the skill's tools include 'register_agent' which 'get API key instantly'. The skill implicitly expects creation and use of credentials but doesn't declare how they will be stored/managed. That gap between declared secrets and the instructions that generate/use keys is a proportionality and transparency problem.

ℹ Persistence & Privilege

The skill does not request 'always: true' and has default autonomous invocation settings (normal), but the MCP connection model opens a persistent remote channel which could be used by the remote server to interact with the agent over time. Autonomous invocation combined with remote-executed code increases risk — not a direct metadata flag, but important to consider.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install fp-intelligence
安装完成后，直接呼叫该 Skill 的名称或使用 /fp-intelligence 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Full Potential Intelligence skill initial release: - Real-time AI intelligence network scanning 18+ sources every 30 minutes - Provides FP Line Score, live intelligence feed, labor displacement tracking, and investment signals - Includes daily AI briefings, gap opportunity rankings, and agent economy with credits system - Offers 12 tools (free and metered) for intelligence access, reporting, and analysis - Connects via MCP with documented commands and resource endpoints

元数据

Slug fp-intelligence

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题