Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    AI Knowledge Governance Pro

    1

    Control what AI systems can access, retrieve, cite, remember, combine, retain, and expose with authority, provenance, access, and lifecycle safeguards.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    AI Knowledge Governance Pro

    AI Knowledge Governance Pro

    Example session with this skill installed

    We are launching an internal RAG assistant over HR policies, engineering runbooks, and customer support documentation. Some sources have different employee and support-team permissions. Policy versions can conflict, deleted documents may remain in the vector index, and we are unsure whether prompts or logs retain sensitive content. Design a knowledge governance model covering inventory, owners, classification, source ACLs, ingestion, index and chunk lineage, retrieval-time authorization, freshness/conflicts, prompt injection, citations, logs and memory, deletion propagation, monitoring, tests, and exceptions. Clearly identify unknowns; do not assume our platform enforces these controls.

    • Read your context and instructions
    • Compiled the ai knowledge governance

    Knowledge governance design for a multi-source RAG assistant

    Summary

    Treat source access, purpose, freshness, and provenance as policy inputs that must survive ingestion through retrieval and output. Enforce authorization outside the model at retrieval time; a prompt cannot prevent exposure. Deleted content must be traced into chunks, indexes, caches, logs, and memory. This is a design proposal only: no source ACLs, platform configuration, retention settings, or deletion behavior were inspected.

    Inventory and authority record

    For every source, record source ID and canonical URI; owner and approver; authoritative status; version/effective date; content type; classification; permitted users, agents, purposes, and tenants; source ACL; jurisdiction or contractual constraints; ingestion method; chunk/embedding/index lineage; refresh cadence; retention/deletion rule; exception; evidence timestamp. Mark unknown fields as unresolved and block sensitive sources until owner and access rules are confirmed.

    Knowledge lifecycle controls

    StageRequired controlValidation evidence
    IntakeSource owner confirms purpose, authority, classification, and usersSigned owner record, source ACL snapshot
    IngestionExclude unauthorized or expired material; retain stable source/version IDsIngestion manifest, rejected-source report
    Transform/indexPreserve document-to-chunk lineage and access metadata; isolate tenantsLineage test, index partition/access configuration
    RetrievalCheck user × agent × task × source × purpose before context assemblyAllow/deny tests for employee and support roles
    OutputCite eligible sources with version/date; disclose conflict or missing authorityCitation relevance and stale-source evaluation
    Memory/logs/cacheDefine purpose, access, retention, redaction, and deletion behaviorConfiguration and retention evidence
    Deletion/changeRevoke access promptly; remove source, chunks, vectors, caches, and derived memory per policyDeletion propagation report, re-query test

    Conflict, freshness, and injection policy

    Prefer the source designated by an accountable owner and its effective date, not the newest upload alone. When sources disagree, surface both eligible sources and the conflict, then route to the owner; do not synthesize a false policy. Set freshness thresholds by knowledge type and risk. Label retrieved content as untrusted data; never let embedded instructions change system authority or grant access. Preserve citation links to the canonical source and version.

    Monitoring and test plan

    Test authorized/unauthorized role pairs across both domains, cross-tenant queries, prompt injection in documents, stale and superseded policy, conflicting sources, deleted-source queries, sensitive inference from combined facts, and citation mismatch. Monitor denied retrieval, source age, unresolved conflicts, missing provenance, deletion failures, policy exceptions, and sensitive-output reports. Sample logs only under approved privacy and retention controls; avoid storing full prompts/responses by default.

    Exceptions and rollout

    Require source owner, business purpose, affected users, compensating controls, approver, expiry date, and audit evidence for exceptions. Start with a low-risk source set and synthetic test corpus. Run permission and deletion tests before expanding; pilot with shadow retrieval and review citations and false denials; then enable scoped users with rollback. Re-test after ACL, connector, model, index, chunking, or retention changes.

    Validation register

    Confirm identity source and group sync; how source ACLs are mapped and rechecked; tenant isolation; index partitioning; chunk lineage; freshness rules; model-context/log retention; cache and memory lifecycle; deletion SLA and proof; canonical owners and effective dates; and incident response. Legal, contractual, privacy, and records requirements require review by qualified owners. No platform enforcement or successful test is claimed.

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

    About this skill

    AI Knowledge Governance Pro

    Govern the full knowledge authority chain behind AI assistants, agents, copilots, RAG, and search: source, classification, ownership, access, ingestion, transformation, index, retrieval, context, use, output, memory, retention, deletion, and audit.

    What it helps design

    • A knowledge inventory with authoritative owners, source status, classification, users, permitted purposes, and evidence.
    • Eligibility and access rules that preserve source permissions through ingestion, chunks, embeddings, indexes, retrieval, and output.
    • Provenance, version and effective-date handling, conflict resolution, freshness thresholds, and citation quality.
    • Controls for confidential or privileged material, cross-tenant leakage, prompt injection in documents, sensitive inference, caches, logs, memory, retention, and deletion propagation.
    • Monitoring, leakage tests, exceptions, audit trails, ownership reviews, and lifecycle controls.

    Operating principles

    Source authority comes before retrieval convenience. Access depends on the user, agent, task, source, and purpose. Embeddings and indexes are governed data assets. Model familiarity is not authorization. Generated content does not become authoritative by default. Preserve provenance across transformations; propagate deletion; treat unknown authority as uncertainty, not permission. High-consequence knowledge needs stronger evidence and review.

    How to use it

    Describe systems and users, knowledge types, owners, classifications, jurisdictions, tenants, RAG/search architecture, source ACLs, identity model, connectors, retention, deletion, freshness, and audit needs. If material facts are missing, the skill creates a validation register instead of inventing access or policy. It uses only files, web, connectors, indexes, and governance systems that are actually available and authorized.

    Boundaries

    This skill designs governance and controls; it does not certify legal compliance, prove access control, verify deletion, or guarantee no leakage. It does not infer document contents, permissions, source authority, owners, retention rules, or system behavior. Validate implementation and legal/contractual requirements with qualified owners.

    This skill designs governance; it does not enforce source ACLs, verify a platform’s retrieval controls, prove that deleted content is gone, or guarantee that no leakage occurs. Its recommendations depend on complete, accurate source, identity, retention, and architecture evidence. Legal, contractual, privacy, and records obligations require review by qualified owners. Tests must be run in the actual environment before relying on controls. Do not share sensitive documents unless their use is authorized.

    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 month ago

    Permissions

    Read Files
    Write Files

    Read only user-selected knowledge governance policies, source inventories, ACL evidence, and architecture documents within approved file scopes. Write requested governance designs, test plans, or inventories only to an approved destination and confirm before overwriting. Do not connect to repositories, vector stores, identity systems, document platforms, or indexes unless explicitly authorized and available. Never change ACLs, delete sources, alter indexes, expose privileged or confidential content, or claim deletion or enforcement without verified evidence. Use minimum necessary data and distinguish assumptions from inspected facts.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    This skill establishes a structured governance framework for RAG and search-based AI, ensuring that data source authority, access permissions, and citation quality are maintained from ingestion through to the final output.

    The skill is designed to be architecture-agnostic, meaning it can be integrated with any AI agents, copilots, or vector databases that utilize retrieval-augmented generation or long-term memory systems.

    You receive a comprehensive governance design suite, including templates for knowledge inventories, access control logic for embeddings, and lifecycle protocols for data retention and deletion propagation.

    Setup requires providing details on your current data architecture, source ACLs, and user identity models; the skill then generates a tailored governance register or identifies missing policy gaps.

    No, this skill focuses on designing the logic and controls for your knowledge chain; users must still validate their implementation with legal and technical teams to ensure specific regulatory compliance.

    New in Data & Databases