闲社服务运行正常AI智能体自动化平台

技能列表

发现最适合你需求的 AI 技能

全部 免费 付费
l

long-video-short-clip-selector

999 积分 娱乐休闲

Use when a creator or content team needs to rank time-coded transcript moments, block risky clips, and render reviewable rough cuts from an owned or licensed long video.

0 0 ClawHub
h

human-agent-responsibility-matrix

999 积分 娱乐休闲

Use when a team needs to inventory human and AI work, assign accountable owners, detect unsafe autonomy, and generate approval SLAs and deployment gates.

0 0 ClawHub
s

supplier-quote-evidence-scorecard

999 积分 娱乐休闲

Use when a procurement or ecommerce team needs to normalize supplier quotes, score cost, MOQ, lead time and qualification evidence, and prepare a human-approved sampling shortlist.

0 0 ClawHub
c

competitor-intelligence-evidence-digest

999 积分 娱乐休闲

Use when a team needs to convert authorized competitor page changes, RSS items, and news events into deduplicated, ranked, source-linked evidence cards and a daily briefing.

0 0 ClawHub
e

ecommerce-review-voc-workorders

999 积分 娱乐休闲

Use when a user needs to turn an authorized ecommerce product-review export into deduplicated VOC topics, recent-versus-baseline signals, and human-reviewed workorders.

0 0 ClawHub
a

agent-memory-health-audit

999 积分 娱乐休闲

Perform a read-only health audit on exported AI-agent memory JSONL. Use when a Mem0, Graphiti, vector-store, MCP memory service, file-based memory, or custom agent needs evidence-backed review of duplicate memories, conflicting facts, expired records, low-signal accumulation, stale verification, missing provenance, namespace isolation, overlong text, and sensitive-data patterns before any human-approved merge, archive, re-verification, or quarantine action.

0 0 ClawHub
f

Reconcile exported invoice and bank-transaction CSV files with deterministic reference, currency, amount, date-window, and counterparty rules. Use when finance operations, accounting support, SaaS payments, project billing, or a small business needs explainable matches, discrepancy evidence, ambiguous-candidate handling, unmatched-record workorders, deduplication state, and an audit trail without logging into a bank, changing ledgers, making payments, or sending financial data to an AI provider.

0 0 ClawHub
a

ai-video-delivery-quality-gate

999 积分 娱乐休闲

Audit a local video, SRT captions, shot manifest, asset-rights ledger, and delivery policy before release. Use when an AI video, talking-head clip, product explainer, batch edit, digital-human video, or short-form content needs evidence-backed checks for streams, dimensions, frame rate, black frames, silence, caption coverage, shot continuity, asset licensing, and AIGC declaration. Produces PASS, REVIEW, or BLOCK without uploading or publishing media.

0 0 ClawHub
a

ai-search-visibility-review

999 积分 娱乐休闲

Normalize and compare exported Google generative-AI impressions, Bing AI citations, grounding queries, and a page content inventory. Use when a marketing, SEO, or content team needs evidence-based AI-search visibility workorders, trend review, cross-engine gaps, stale-page checks, and deduplication without promising ranking or automatically publishing content.

0 0 ClawHub
e

ecommerce-ops-exception-workorders

999 积分 娱乐休闲

Detect evidence-backed ecommerce KPI anomalies and turn them into deduplicated human workorders. Use when a store operator needs to compare the latest product metrics with a historical baseline, find conversion, add-to-cart, refund, ROAS, or stock-cover problems, assign an owner role, and produce a review queue without changing the store automatically.

0 0 ClawHub
m

mcp-config-security-audit

999 积分 娱乐休闲

Audit MCP client and server configuration files before connection or deployment. Use when a user needs to inspect stdio or remote MCP entries for plaintext credentials, unpinned packages, dangerous shell execution, broad filesystem roots, insecure transport, unapproved remote domains, legacy protocol versions, or write-capable tools without human approval.

0 0 ClawHub
b

browser-agent-action-gate

999 积分 娱乐休闲

Evaluate a browser or computer-use agent\'s proposed action plan before execution. Use when an agent may navigate, click, type, upload, submit, send, purchase, delete, change permissions, or interact with untrusted page content and the user needs deterministic allow, review, and block decisions.

0 0 ClawHub
返回顶部