agentsclimarketplace

Case 04566

Skill knownasnaffy/prompthound/dataset/case_04566

A fast, offline static risk analysis CLI for AI agent skill files. Detects malicious instructions, steganographic payloads, and dangerous capability chains.

Install
npx -y skills add knownasnaffy/prompthound --skill case_04566

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Persistent memory infrastructure for AI agents with 4-layer Cognitive Security Pipeline (PII redaction, AES-256-GCM encryption, intent validation, differential privacy). Use when: configuring SynapseLayer MCP integration, storing/retrieving agent memories, cross-agent memory sharing, analyzing trust scores, or implementing persistent memory in OpenClaw agents. Also use for troubleshooting SynapseLayer connectivity, understanding Trust Quotient scoring, or setting up framework integrations (LangChain, CrewAI, AutoGen, LlamaIndex, Semantic Kernel).

SKILL.md

4.9 KB, as published. Nobody here has run it

Synapse Layer Skill

Synapse Layer provides persistent, encrypted memory for AI agents with a 4-layer security pipeline.

Quick Setup

Python SDK (Recommended for OpenClaw)

Install and use:

pip install synapse-layer

from synapse_test import SynapseClient

client = SynapseClient(api_key="sk_connect_...")

# Save memory
result = client.remember("User prefers dark mode", agent="mel")

# Retrieve memories
memories = client.recall("user preferences", agent="mel")

See scripts/synapse_client.py for a complete client implementation.

MCP Integration (External Tools)

Add to OpenClaw gateway config for external MCP clients:

{
  "mcp": {
    "servers": {
      "synapse-layer": {
        "url": "https://forge.synapselayer.org/mcp",
        "headers": {
          "Authorization": "Bearer sk_connect_YOUR_API_KEY"
        }
      }
    }
  }
}

Note: This is for external MCP clients (Claude Desktop, Cursor) to connect to SynapseLayer, not for OpenClaw agents to use directly.

Available Tools

Once configured via Python SDK, these operations are available:

  • save_to_synapse - Store memory with full security pipeline
  • recall - Retrieve memories ranked by Trust Quotient™
  • search - Cross-agent memory search with full-text matching
  • process_text - Auto-detect decisions, milestones, and alerts
  • health_check - System health, version, and capability report

Cognitive Security Pipeline

Every memory passes through 4 non-bypassable layers:

  1. Semantic Privacy Guard™ - 15+ regex patterns detect PII, secrets, credentials
  2. Intelligent Intent Validation™ - Two-step categorization with self-healing
  3. AES-256-GCM Encryption - Authenticated encryption with PBKDF2 key derivation
  4. Differential Privacy - Calibrated Gaussian noise on embeddings

Key Concepts

Trust Quotient™ (TQ)

Score (0-1) ranking memory reliability. Higher TQ = more trusted memory.

Cross-Agent Memory

Memories can be shared across agents using the same agent_id or cross-agent search.

Storage Backends

  • Remote (Forge) - PostgreSQL via forge.synapselayer.org (recommended)
  • SQLite - Local .synapse/memories.db (requires local setup)

Usage Patterns

Basic Memory Operations

Use the Python SDK client:

from scripts.synapse_client import SynapseClient

client = SynapseClient(api_key="sk_connect_...")

# Save
client.remember("Important decision", agent="mel", importance=5)

# Recall
memories = client.recall("recent decisions", agent="mel", limit=5)

# Search
all_memories = client.search("project deadlines", limit=10)

Process Text Automatically

Extract events from free-form text:

events = client.process_text(
    "Decided to use PostgreSQL. Deadline is May 1st.",
    agent="hermes",
    project="website-redesign"
)

Security Considerations

Data Privacy

  • PII automatically redacted before storage
  • All data encrypted at rest (AES-256-GCM)
  • Differential privacy protects individual entries
  • Zero-knowledge architecture

Resilience Strategy

Recommended approach for production:

  1. Use SynapseLayer as primary memory store
  2. Keep OpenClaw's MEMORY.md/memory/ as backup
  3. Monitor service health via health_check
  4. Have fallback procedure if service unavailable

Troubleshooting

Connection Issues

  1. Run health_check to verify connectivity
  2. Verify API key is valid
  3. Check network access to forge.synapselayer.org

Memory Not Persisting

  1. Check API key permissions
  2. Verify security pipeline didn't reject content
  3. Review Trust Quotient in response

Low Trust Quotient

  1. Review security pipeline logs
  2. Increase confidence/importance scores
  3. Check if content was sanitized

Testing

Use the test script:

python3 /app/skills/synapse-layer/scripts/synapse_test.py

This script verifies:

  • Service connectivity (health_check)
  • Basic save operations
  • Memory retrieval
  • Cross-agent search

Reference Documentation

For more details, see:

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.