Test only what you own
Use local fixtures, local files, and systems where you have explicit permission.
Safe Testing and Non-Claims
Security and boundaries
Latticra treats security-facing work as evidence-bound: no claim without evidence, no execution before a contract, no mutation before visibility, and no hidden effects.
Current Posture
Defensive model, not production protection.
Current security work is mostly parser validation, deterministic diagnostics, source spans, denied requests, metadata reports, effect classification, and public non-claims. Latticra does not currently provide a hardened sandbox, malware prevention, ransomware prevention, kernel isolation, production runtime, or security certification.
Report privately
Do not publish exploit details, secrets, payloads, or reproduction steps in public issues.
Prefer minimal fixtures
Parser and source-handling reports should use small deterministic cases.
Keep effects visible
Unknown, ambiguous, or unclassified effects should remain denied.
Security-Relevant Areas
What reports are most useful right now.
The current project is early, but security reports are useful when they find a concrete mismatch between code, tests, contracts, and public claims.
Parser safety
Malformed source, oversized input, unchecked buffers, string escapes, literal NUL handling, and source-span confusion.
Diagnostics and reports
Cases where rejected behavior becomes unclear, ambiguous, hidden, or reported as accepted.
Boundary bypass
Unknown requests, unknown effects, future-gated actions, or operator confirmation being treated as permission.
Claim overreach
Documentation, status, examples, or UI language that overstates security maturity or production readiness.
Effect Gates
Classify before handling.
Latticra's effect model is conservative: no hidden mutation, no hidden network behavior, and no implied runtime authority.
none
Fixtures, validation, reports, classification, and preview metadata.
read
Approved local inspection only after a source contract and visible report path.
mutation
Local or host mutation needs explicit gates, policy, reports, and rollback/failure behavior.
network / hardware / boot
External, hardware, boot, recovery, and server behavior remain denied until future evidence gates exist.
Runtime Boundary
The current runtime posture is disabled-by-default.
The runtime boundary is the line between validated metadata/report/classification surfaces and future behavior that could execute, mutate, contact networks, affect recovery, touch hardware, or claim authority.
Allowed now
Parsing, validation, classification, previews, deterministic reports, no-effect task records, and report-only dry runs.
Denied now
Command execution, Lat/LIR execution, file I/O, network I/O, server behavior, recovery, rollback, hardware, boot, and sandbox claims.
Operator confirmation
May be metadata in future work, but it must not override runtime policy in the current boundary.
Non-Claims
Do not describe previews as protection.
These are intentionally blocked as public claims until contracts, implementation, tests, status records, and reproducible evidence support them.
No hardened sandbox
Current metadata and validation surfaces are not process isolation or sandbox escape resistance.
No malware or ransomware prevention
Denied reports are not host protection, malware containment, or ransomware resistance.
No production cryptography claim
Seal and key records are evidence surfaces, not production cryptographic enforcement.
No OS completeness
Latticra is not a kernel, bootable image, daily-driver platform, or production systems platform.
Source Records
Use exact records when discussing security posture.
Security policyReporting path, safe testing rules, and current security scope.
High-assurance baselineNSA, CISA, FBI, and NIST source-tracked allocation for future security work.
Memory-safety roadmapComponent inventory, C/C++ mitigation posture, and promotion blockers.
Supply-chain gatesReader-facing baseline for CI, dependency, SBOM, update, runtime-authority, and release blockers.
Cryptographic assurance and key managementReader-facing guide to FIPS/CMVP claim gates, key lifecycle, randomness, and no production crypto claims.
Non-claimsUnsupported security, OS, hardware, and production-readiness claims.
Runtime boundaryDefault-deny runtime classification and no-effect Nucleus reports.
Nadia boundariesOffline AI safety posture, tool denial, and runtime non-claims.
Local validationInstall evidence, package guards, and disposable VM gates.
Threat modelDefensive vocabulary, assets, abuse cases, and evidence expectations.
Incident response boundaryReader-facing guide to reporting routes, evidence preservation, response gates, and incident-response non-claims.
Cyber incident reporting and response baselineExact source record for reporting routes, response gates, and closed response authority.
Vulnerability management release gateReader-facing guide to KEV/NVD review, exception records, release blocking, and product-security non-claims.
Vulnerability management release gate baselineKEV/NVD review, disclosure handling, release blocking, and no product-security claims.
Cryptographic assurance and key management baselineFIPS/CMVP claim gates, key lifecycle, randomness, and no production crypto claims.
Identity, credential, and access managementReader-facing guide to privileged access, phishing-resistant MFA planning, account lifecycle, service identity, and no hosted access claims.
Identity, credential, and access management baselinePrivileged access, phishing-resistant MFA planning, account lifecycle, service identity, and no hosted access claims.
Security logging, monitoring, and detectionReader-facing guide to event-source inventory, audit events, redaction, retention, triage, and no monitoring-service claims.
Security logging, monitoring, and detection baselineEvent-source inventory, audit events, redaction, retention, triage, and no monitoring-service claims.
Backup, recovery, and cyber resilienceReader-facing guide to backup scope, restore testing, recovery prioritization, rollback planning, RTO/RPO, and no recovery-service claims.
Backup, recovery, and cyber resilience baselineBackup scope, restore testing, recovery prioritization, rollback planning, and no recovery-service claims.
Secure configuration and change managementReader-facing guide to configuration inventory, secure baselines, approved changes, rollback planning, drift detection, exceptions, and no hardening claims.
Secure configuration and change management baselineConfiguration inventory, secure baselines, approved changes, rollback planning, drift detection, and no hardening claims.
Network exposure and remote accessReader-facing guide to network inventory, internet exposure, ingress/egress policy, remote access, RMM, DNS/TLS lifecycle, and no network-service claims.
Network exposure and remote access baselineNetwork inventory, internet exposure, ingress/egress policy, remote access, RMM, DNS/TLS lifecycle, and no network-service claims.
Data classification and protectionReader-facing guide to data inventory, sensitive-data flows, PII review, minimization, retention, disposal, redaction, and no customer-data claims.
Data classification and protection baselineData inventory, sensitive-data flows, PII review, minimization, retention, disposal, redaction, and no customer-data claims.
AI and agentic automation security baselineAI inventory, model provenance, prompt/context boundaries, tool authority, human approval, monitoring, and no AI runtime claims.
Platform boot and firmware integrity baselineBoot mode, Secure Boot, TPM/PCR, firmware update and recovery evidence, and no platform-integrity claims.
Security validation and assessment baselineAssessment scope, ROE, authorization, evidence, remediation, retest, independent review, and no validation claims.
Supply-chain baseline recordWorkflow authority, local guard scripts, and source archive fixture controls.
Zero-trust runtime authority baselinePer-request authority prerequisites, identity/resource visibility, and no implicit trust.
Signed-updater delivery gateReader-facing guide to the closed update-delivery gate, no-effect fixtures, and blocked network self-update authority.
Signed-updater gateClosed future update-delivery gate before manifests, signatures, verification, rollback, validation, and receipts exist.
Runtime boundaryDisabled-by-default runtime posture and future-gate rules.