The Compliance Validation Triad
The framework relies on a three-legged FDA-style tracking and validation triad. It establishes automated CI gates and local scripts to ensure that nothing reaches production without documentation, tests, and dependency risk checks.
1. Traceability
Requirement to verification tracking
2. Test execution
Full test suite enforcement
3. OTS Inventory
Third-party dependency scanner
Requirements Traceability Matrix
Enforces a mapped connection from stable requirement IDs (e.g. FR-012) to automated verification tests. Substring collisions are detected by CI to block invalid compliance states.
Automated Test Suite Enforcement
CI runs `run_declared_tests.sh` which dynamically parses and executes the verified test command declared inside docs/TEST_PLAN.md. A test failure blocks PR merging.
OTS Software Inventory & Risk Review
Enforces recording and reviewing all third-party dependencies (runtimes, databases, libraries) in docs/OTS_SOFTWARE.md. `check_ots_inventory.sh` flags any newly added package missing in the inventory.
Interactive Workflow Simulator
Watch the AI agent transition through the 30 mandatory workflow steps. This ensures proper alignment, checkpoints, and documentation before code changes are made.
12 Core Principles
Click on any principle to inspect its operational code smell and required guardrail.
All changes must consider documentation impact. Readme and session planning are mandatory.
All new functionality and bug fixes include automated tests mapped to requirement IDs.
Maintain TODO.md as the living roadmap. Log debt, feature opportunities, and refactoring.
Agents actively identify UI, UX, speed, reliability, and developer experience improvements.
Enforce input validation, least privilege, local secret sweeps, and protocol firewalls.
Record major frameworks and database structures in Architecture Decision Records (ADRs).
Prefer mature packages. Document dependencies in the OTS software inventory.
Log diagnoses cleanly, avoid leaking credentials, and create clear operational logs.
Operational changes must be reviewable, observable, and recoverable with a runbook.
SemVer enforcement. Bump versions on VERSION and push Git tags for each release.
Agents look beyond immediate tasks to recommend future architectures and tools.
Style guides and docstring formats must follow language owners' official styling rules.