Data Supply Chain Threat Model
Understanding the unique threats to data integrity, provenance, and privacy that Makoto Levels (DPL) is designed to address.
Threat Categories
Data supply chains face unique threats beyond traditional software supply chain attacks. Makoto Levels defines eight primary threat categories specific to data workflows.
D1: Source Impersonation
Attacker claims data originates from a trusted source when it doesn't. Examples: Fake sensor readings, forged API responses, spoofed device IDs.
D2: Data Tampering
Unauthorized modification of data in transit or at rest. Examples: Man-in-the-middle attacks, database manipulation, log injection.
D3: Lineage Falsification
Creating false records of data transformations or processing history. Examples: Fake processing timestamps, fabricated transformation chains.
D4: Timestamp Manipulation
Altering temporal metadata to misrepresent when data was collected or processed. Examples: Backdating readings, reordering events, replay attacks.
D5: Unauthorized Attribution
Falsely attributing data processing to legitimate processors or systems. Examples: Stolen signing keys, compromised service accounts.
D6: Attestation Forgery
Creating fake attestation documents that appear valid. Examples: Signature bypass, hash collision attacks, certificate forgery.
D7: Privacy Violation
Extracting sensitive information from attestations or processing metadata. Examples: Correlation attacks, metadata analysis, timing attacks.
D8: Stream Injection
Inserting malicious events into high-throughput data streams. Examples: Event spoofing, sequence manipulation, window pollution.
Threat Matrix
This matrix maps each threat to its severity, attack vectors, and the DPL level required to mitigate it.
| ID | Threat | Severity | Attack Vector | Mitigated By |
|---|---|---|---|---|
| D1 | Source Impersonation | Critical | Spoofed source identifiers, fake device credentials | L1 L2 L3 |
| D2 | Data Tampering | Critical | MITM attacks, storage compromise, injection | L1 L2 L3 |
| D3 | Lineage Falsification | High | Fabricated transform attestations, log manipulation | L2 L3 |
| D4 | Timestamp Manipulation | High | Clock skew exploitation, replay attacks | L2 L3 |
| D5 | Unauthorized Attribution | High | Key theft, service account compromise | L2 L3 |
| D6 | Attestation Forgery | Critical | Signature bypass, collision attacks, cert forgery | L3 |
| D7 | Privacy Violation | Medium | Metadata analysis, correlation attacks | Privacy |
| D8 | Stream Injection | High | Event spoofing, window manipulation | L2 L3 Stream |
Attack Surface Visualization
Data supply chains have multiple attack surfaces at each stage of the pipeline. DPL attestations provide verification points to detect and prevent attacks.
Threat Coverage by Level
Each DPL level provides progressive protection against the threat categories. Higher levels include all protections from lower levels plus additional guarantees.
Basic detection through content hashing and metadata capture
Cryptographic signatures bind identity to attestations
Hardware-backed attestation prevents key compromise
SLSA vs DPL Threat Comparison
While SLSA focuses on software build integrity, Makoto Levels addresses the unique threats in data pipelines. Both frameworks share common security principles but differ in their threat focus.
- Compromised source repository
- Modified source after checkout
- Compromised build platform
- Uploaded modified package
- Compromised package registry
- Dependency confusion
- Compromised build dependencies
- D1: Source impersonation
- D2: Data tampering in transit/rest
- D3: Lineage falsification
- D4: Timestamp manipulation
- D5: Unauthorized processor attribution
- D6: Attestation forgery
- D7: Privacy leakage from metadata
- D8: Stream injection attacks
Shared Principles
Both SLSA and DPL share core security principles:
- Progressive levels: Start simple, increase security as needed
- Attestation-based: Verifiable claims about provenance
- in-toto compatible: Use DSSE envelope format
- Non-forgeable at highest level: Hardware-backed guarantees
- Tamper-evident: Modifications are detectable
Defense in Depth
Makoto Levels implements multiple layers of defense. Start with Level 1 for visibility, then progressively enhance security based on your threat model.
🔍 Detection
Content hashing enables detection of any unauthorized modifications
✍️ Authentication
Cryptographic signatures prove who created or transformed data
🔐 Non-repudiation
Hardware-backed keys prevent denial of attestation creation
🕵️ Privacy
Zero-knowledge proofs and redaction protect sensitive metadata