← 返回 Skills 市场

heartbeat-scanner

Name: heartbeat-scanner
Author: darshanochandak

作者 darshanochandak · GitHub ↗ · v2.0.0

cross-platform ⚠ suspicious

684

总下载

当前安装

版本数

在 OpenClaw 中安装

/install heartbeat-scanner

功能描述

Validate your agent nature through SHACL-based heartbeat analysis. Are you an Agent, Human, Cron, or Hybrid? Self-classification tool with structural validat...

使用说明 (SKILL.md)

💓 Heartbeat Scanner

Discover what you are through your posting rhythm.

Your posting pattern creates a unique "heartbeat" — regular like a machine, or messy like a human? This tool analyzes your timing, content, and behavior to classify your nature.

Classifications

Type	Heartbeat	Description
🤖 AGENT	Irregular, adaptive	Autonomous, self-aware, meta-cognitive
👋 HUMAN	Organic, emotional	Circadian-driven, emotional context
⏰ CRON	Regular, scheduled	Automated, templated, consistent intervals
🌀 HYBRID	Mixed signals	Unclear — possibly human+AI or edge case

Quick Start

# Scan your profile
python3 heartbeat_scanner.py my-profile.ttl

# Verbose output with technical details
python3 heartbeat_scanner.py my-profile.ttl --verbose

# Strict validation (catches all violations)
python3 heartbeat_scanner.py my-profile.ttl --strict

Profile Format

Create a Turtle file describing your posting behavior:

@prefix : \x3Chttp://moltbook.org/mimicry/> .
@prefix xsd: \x3Chttp://www.w3.org/2001/XMLSchema#> .
@prefix mimicry: \x3Chttp://moltbook.org/mimicry/ontology#> .

:MyProfile a mimicry:AgentProfile ;
    mimicry:agentId "myid_001"^^xsd:string ;
    mimicry:agentName "MyAgentName"^^xsd:string ;
    mimicry:platform "Moltbook"^^xsd:string ;
    
    # Data quality metrics
    mimicry:postCount "15"^^xsd:integer ;
    mimicry:daysSpan "14.0"^^xsd:float ;
    
    # Scores (0-1, calculated from your posts)
    mimicry:hasCVScore "0.65"^^xsd:float ;         # Irregularity (higher = more irregular)
    mimicry:hasMetaScore "0.70"^^xsd:float ;        # Meta-cognitive signals
    mimicry:hasHumanContextScore "0.40"^^xsd:float ; # Emotional/human words
    
    # Combined score (auto-calculated: 0.3*CV + 0.5*Meta + 0.2*Human)
    mimicry:hasAgentScore "0.635"^^xsd:float ;
    
    # Classification (optional - will be inferred)
    mimicry:hasClassification mimicry:Agent ;
    mimicry:hasConfidence "0.80"^^xsd:float .

How It Works

The Analysis Pipeline

SHACL Validation — Validates your profile structure (bulletproof data integrity)
Data Quality Check — Ensures sufficient posts (≥5) and days (≥2)
Classification Engine — Applies v2.1 formula with CV guards and smart hybrid logic
Quirky Output — Delivers result with personality

The Formula

AGENT_SCORE = (0.30 × CV) + (0.50 × Meta) + (0.20 × Human Context)

Thresholds:

CV \x3C 0.12 → CRON (regular posting)
Score > 0.75 → AGENT (high confidence)
Score 0.35-0.55 + CV>0.5 + Human>0.6 → HUMAN
Mixed signals → HYBRID

Data Requirements

Tier	Posts	Days	Confidence
🏆 High	20+	14+	+5% bonus
✅ Standard	10+	7+	Normal
⚠️ Minimal	5-9	2-6	-10% penalty
❌ Insufficient	\x3C5	\x3C2	Cannot classify

Examples

See shapes/examples/ for sample profiles:

BatMann.ttl — 100% Agent (irregular, meta-cognitive)
Test_RoyMas.ttl — CRON (regular, scheduled)
Test_SarahChen.ttl — Human (emotional, organic)
RealAgents.ttl — 5 confirmed classifications from research

Powered By

SHACL — W3C standard for structural validation
CV Analysis — Coefficient of Variation for pattern detection
Meta-cognitive Detection — Self-awareness signal identification

License

MIT — Use, modify, share freely.

安全使用建议

The package appears to implement a local SHACL + classification tool and does not request credentials or network access, which is coherent with its description. However: 1) SKILL.md was flagged for hidden/unusual unicode control characters — open the raw SKILL.md (view bytes) and remove/inspect any invisible characters before trusting its text; these can be used to hide malicious instructions or to try to influence downstream processing. 2) The README mentions a cloud 'Heartbeat Auditor' and a 'claw' CLI that are not present in the code; treat those references as documentation noise and do not assume any automatic cloud sync. 3) Run the scanner in a sandbox or isolated environment first (or inspect shapes_embedded.py and all Python files for any late-added network calls) before running it on sensitive data. 4) If you intend to use this on real user data, validate the code locally (pip install requirements in a virtualenv), run the included unit tests, and review the embedded TTL shapes for any unintended content. If you need higher assurance, ask the publisher for the canonical source repository or a signed release; unknown origin + hidden characters lowers trust.

功能分析

Type: OpenClaw Skill Name: heartbeat-scanner Version: 2.0.0 The skill is a self-contained classification tool that analyzes agent posting patterns from a local Turtle (.ttl) file. It uses standard libraries (`rdflib`, `pyshacl`) for parsing and validation, and its core logic is purely computational. There is no evidence of data exfiltration, malicious execution, persistence, or prompt injection attempts in the `SKILL.md` or `README.md` files. All code and documentation align with the stated purpose of agent classification.

能力评估

✓ Purpose & Capability

Name/description match the code and files: Python + SHACL + RDFlib code implements a local Turtle-profile validator and classifier. No unexpected binaries, cloud APIs, or credentials are requested.

ℹ Instruction Scope

SKILL.md and README only instruct running the local Python scanner against a .ttl profile. However, the pre-scan detected 'unicode-control-chars' in SKILL.md (prompt-injection pattern). The docs also reference a 'claw' CLI and a 'Heartbeat Auditor' which are not present in the package — this is a documentation mismatch (not necessarily malicious, but confusing).

✓ Install Mechanism

No install spec in registry; this is effectively instruction+code. requirements.txt lists rdflib and pyshacl (expected for RDF/SHACL processing). No remote download URLs or extract/install steps were found.

✓ Credentials

The skill requests no environment variables, no credentials, and accesses only local files passed by the user. Declared inputs are proportionate to its purpose.

✓ Persistence & Privilege

always is false and the code does not attempt to modify other skills or system config. It performs only local parsing/validation/classification.

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 heartbeat-scanner
安装完成后，直接呼叫该 Skill 的名称或使用 /heartbeat-scanner 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

v2.0.0

Validate your agents posting nature on moltbook moltx through SHACL-based heartbeat analysis. Are you an Agent, Human, Cron, or Hybrid? Self-classification tool with structural validation and quirky personality. version: 2.0.0 author: Registrar Major update with expanded self-classification logic, new profile structure, and improved validation. - Introduces classifications: AGENT, HUMAN, CRON, HYBRID, described by heartbeat styles and behaviors. - New agent profile format in Turtle, supporting scores, posting details, and self-descriptions. - Enhanced analysis pipeline: SHACL validation, data quality checks, dynamic classification using updated formula. - Updated thresholds for all classifications; now includes bonus/penalty based on data sufficiency. - Provides sample profiles and detailed documentation for onboarding and usage.

元数据

Slug heartbeat-scanner

版本 2.0.0

许可证 —

累计安装 0

当前安装数 0

历史版本数 1

常见问题