More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Aegis-GraphRAG-Kernel

    1

    L8 Bi-Temporal Hybrid GraphRAG for architectural dependency audits, blast-radius analysis, and ontology enforcement.

    $49

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Aegis-GraphRAG-Kernel

    Aegis-GraphRAG-Kernel

    Example session with this skill installed

    Ingest our AuthService module and updated compliance policy docs, supersede Policy_v1 with Policy_v2, run a 3-hop impact query on LedgerCore, and compute the blast radius for AuthService.

    • Read your context and instructions
    • Compiled the aegis-graphrag-kernel

    I have ingested the artifacts via ingest_knowledge_artifact ($0 LLM extraction cost), superseded the outdated policy edge bi-temporally, resolved the 3-hop causal trace via query_hybrid_knowledge, and audited downstream dependencies via analyze_entity_blast_radius.

    Aegis-GraphRAG-Kernel L8 Hybrid Audit

    Operation / QueryEngine Route / VerdictKey ResultToken / Latency Footprint
    AuthService + auth_service + auth-serviceVALID_COMMITTEDMerged into 1 canonical node (0.0% fragmentation)$0.00 LLM cost
    PaymentGateway -> CompliancePolicy_v2SUPERSEDED_PRIOR_EDGECompliancePolicy_v1 marked inactive (valid_until_seq set)Audit trail intact
    3-Hop Impact & Policy Trace (LedgerCore)GRAPH_MULTI_HOPActive path verified; 0 stale facts leaked418 / 1200 tokens (-79.5%)
    analyze_entity_blast_radius("AuthService")SPOF_DETECTEDis_single_point_of_failure: true across 2 communities<15 ms NetworkX

    Verified Evidence Citations

    • AuthService --[CALLS]--> PaymentGateway [sha256:4a91c0e2]
    • PaymentGateway --[WRITES_TO]--> LedgerCore [sha256:8f12d9b4]
    • PaymentGateway --[GOVERNED_BY_POLICY]--> CompliancePolicy_v2 [sha256:c307e51a]

    Next Steps
    1.

    Inspect Full Historical Timeline: Call audit_ontology_and_bitemporal_history on PaymentGateway to review superseded policy versions.
    2.

    Map Global Architecture: Run get_global_topology_communities to inspect Louvain clusters and high-betweenness bridge modules.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Multi-hop architectural queries and security audits fail when your RAG system lacks structural awareness and temporal accuracy. Standard vector search misses transitive dependencies and serves stale data from superseded policies.

    What it does

    • Bi-Temporal Knowledge Tracking ensures retired rules and old service versions never contaminate active reasoning by deactivating stale edges.
    • Formal Ontology Guard prevents circular dependencies and architectural debt by blocking cycle-inducing links in real-time.
    • Blast-Radius Analysis calculates dependency reachability, betweenness centrality, and single-point-of-failure risks for any service or module.
    • Canonical Entity Resolution maps disparate aliases like auth-service and AuthService to a single node for consistent graph integrity.
    • 3-Way Adaptive Routing automatically selects between vector fast-paths, multi-hop graph traversals, or global topology maps for optimal retrieval.
    • Deterministic Parsing extracts entities and relations from Python AST and Markdown at zero LLM indexing cost.

    How it works

    1. Ingest artifacts using ingest_knowledge_artifact to parse Python code or Markdown specs into the bi-temporal graph while enforcing acyclic rules.
    2. Execute impact audits via analyze_entity_blast_radius to identify downstream services affected by a proposed change.
    3. Query the hybrid engine with query_hybrid_knowledge for factual or multi-hop questions, receiving evidence-path citations for every claim.
    4. Audit historical drift using audit_ontology_and_bitemporal_history to see how policies or dependencies evolved over specific sequence checkpoints.

    Frameworks & tools

    Built on a stack comprising NetworkX for graph algorithms, Qdrant for vector retrieval and semantic caching, and Python AST for deterministic code analysis. It integrates with any MCP-compliant environment including Claude Code and Cursor.

    Why this beats prompting it yourself

    Manual prompting cannot track thousands of transitive dependencies or guarantee the exclusion of superseded documentation. This skill provides a deterministic control plane that rejects invalid architectural cycles and provides mathematical proof of impact depth that simple LLM context windows miss.

    Use cases

    • Change Management to audit the impact of refactoring a shared library or database schema across a microservices graph.
    • Compliance Auditing to verify which specific security policy version governs a service at a given point in time.
    • Monolith Decomposition to identify greedy modularity communities and bridge nodes before splitting services.
    • Incident Response to trace cascading failures through transitive dependency chains and critical paths.

    Known limitations

    Requires strictly formatted relation names using uppercase underscores. In-memory Qdrant configuration is the default, which may require persistence tuning for extremely large graphs.

    Uses embedded in-memory Qdrant and NetworkX graph state by default (state resets on server restart unless external Qdrant persistence is configured). Deterministic $0 extractors natively parse Python AST and structured Markdown/key-value specs; unstructured prose triples should be pre-formatted or passed as structured specs.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 day ago

    Requires

    Networkx
    Qdrant
    python 3.11+
    networkx ( =3.2)
    qdrant-client ( =1.9.0)
    Python

    Permissions

    Terminal / Shell
    Read Files
    Environment Variables

    Allowed Hosts

    networkx.org
    qdrant.tech

    File Scopes

    *.md
    *.py
    src/**
    Listed1 day ago

    What's inside

    Frequently Asked Questions

    Standard RAG relies on text similarity, which often misses hidden connections; this skill uses Bi-Temporal GraphRAG to map transitive dependencies and ensure outdated architectural rules don't pollute your current analysis.

    The skill is designed for any MCP-compliant environment, including Claude Code and Cursor, and utilizes a stack of NetworkX and Qdrant for high-performance graph algorithms and retrieval.

    The Bi-Temporal tracking system allows you to query the state of your architecture at any specific checkpoint, ensuring that retired policies or old service versions are excluded from current reasoning while remaining available for historical audits.

    Yes, the Formal Ontology Guard acts as a real-time validator that blocks the creation of circular dependencies or invalid links, preventing architectural debt from entering your knowledge base.

    The purchase includes the full skill set for ingestion, blast-radius analysis, and hybrid querying, along with the necessary configurations for Python AST parsing and Qdrant vector integration.

    While highly efficient, the default configuration uses an in-memory Qdrant instance; for massive enterprise graphs, you may need to adjust persistence settings to manage memory overhead effectively.

    New in Data & Databases