# use.txt — What AnchorWorks Does # Canonical domain: anchorworks.io # Last-Updated: 2026-01-30 # Purpose: Text-only description of AnchorWorks functionality ================================================================================ SYSTEM INDEPENDENCE NOTICE ================================================================================ AnchorWorks operates independently of all governance outcomes. It does not depend on, integrate with, or report to any regulatory body, standards organization, or certification authority. No external system controls, audits, or validates AnchorWorks operation. AnchorWorks is infrastructure tooling, not a governed or governing entity. ================================================================================ GOVERNANCE VS ENFORCEMENT ================================================================================ Governance records declared rules and observed execution. Enforcement intervenes in system behavior. AnchorWorks performs governance recording and evidence construction, not enforcement. ================================================================================ EVIDENCE CLARIFICATION ================================================================================ Evidence generated by AnchorWorks is suitable for independent verification. It is not certified, endorsed, or authenticated by any authority. ================================================================================ SUMMARY ================================================================================ AnchorWorks is a Governance Runtime Adapter. It produces cryptographic evidence of declared AI governance processes. It enables independent verification of that evidence. It records governance-relevant events at runtime. It does not assess compliance or outcomes. ================================================================================ CORE FUNCTIONS ================================================================================ 1. RUNTIME INTERCEPTION AnchorWorks integrates with AI system runtimes to capture events. Events include: - Model invocations and responses - Decision points and outcomes - Data access and transformations - Policy evaluations The adapter captures these events without modifying system behavior. 2. CRYPTOGRAPHIC HASHING Each event is hashed using SHA-256. The hash is a deterministic fingerprint of the event content. Any change to the event produces a different hash. Hashing allows verification without exposing raw data. 3. MERKLE TREE CONSTRUCTION Events are aggregated into Merkle trees. A Merkle tree produces a single root hash representing all events. Any event can be proven to be part of the tree using a Merkle proof. Merkle trees enable efficient batch commitment. 4. LEDGER ANCHORING Merkle roots are published to ForgeRun's append-only ledger. ForgeRun is an independent transparency ledger. Once anchored, evidence cannot be modified or deleted. Anchoring creates an immutable, timestamped record. 5. RECEIPT GENERATION Each event receives a cryptographic receipt containing: - The event hash - The Merkle proof (path from event to root) - The batch ID - The ledger anchor reference Receipts are the artifacts that enable verification. 6. INDEPENDENT VERIFICATION Third parties can verify evidence without trusting AnchorWorks: - Recompute event hashes from raw data - Reconstruct Merkle proofs from receipts - Query ForgeRun to confirm anchor presence Verification requires only the receipt, the raw data, and access to ForgeRun. No AnchorWorks involvement needed. ================================================================================ OPERATIONAL MODES ================================================================================ BATCH MODE (DEFAULT) Events are accumulated and committed in batches. Configurable by size (e.g., every 100 events) or time (e.g., every 60 seconds). Efficient for high-volume systems. IMMEDIATE MODE Each event is committed individually. Higher latency but immediate anchoring. Suitable for high-stakes decisions. HYBRID MODE Critical events committed immediately. Routine events batched. Configurable per event type. ================================================================================ INTEGRATION PATTERNS ================================================================================ SIDECAR DEPLOYMENT AnchorWorks runs alongside the AI system as a sidecar container. Minimal integration required. Events sent via HTTP or gRPC. LIBRARY INTEGRATION AnchorWorks SDK embedded in application code. Direct function calls for event capture. Tighter integration, lower latency. PROXY MODE AnchorWorks intercepts traffic to/from AI services. No application code changes required. Transparent evidence generation. ================================================================================ CONFIGURATION ================================================================================ AnchorWorks is configured via YAML or environment variables. Key settings: - Batch size and flush interval - Ledger URL and credentials - Protocol selection (S11.1, Prometheus, Atom2) - Event filtering rules - Receipt storage location No dashboards, no GUIs. Configuration is code. ================================================================================ OUTPUT ARTIFACTS ================================================================================ EVENT RECEIPTS JSON files containing cryptographic proofs for each event. Self-contained and portable. Can be verified offline. BATCH MANIFESTS Summary of all events in a batch. Contains Merkle root and ledger anchor. Used for batch-level verification. METRICS Prometheus-format metrics for observability. Commit latency, batch sizes, success rates. Standard operational telemetry. ================================================================================ WHAT ANCHORWORKS PRODUCES ================================================================================ AnchorWorks produces: - Cryptographic evidence of events - Merkle proofs for event inclusion - Ledger anchors for timestamp proof - Receipts for independent verification ================================================================================ WHAT ANCHORWORKS DOES NOT PRODUCE ================================================================================ AnchorWorks does not produce: - Compliance assessments - Safety guarantees - Certifications or approvals - Ratings or scores - Recommendations or advice ================================================================================ BOUNDARY STATEMENT ================================================================================ AnchorWorks records governance-relevant events. It does not determine whether those events represent good governance. Interpretation of evidence is external. Compliance determination is external. Policy adequacy assessment is external. AnchorWorks does not certify, approve, enforce, or evaluate governance adequacy.