Leverage the best available tool; don't reinvent substrate
State: open/published
Summary
Section titled “Summary”- Leverage external tools and agents built by parties with more resources; do not build parallel implementations.
- SDLC’s differentiation is the planning model (entities, contracts, relationships) and the zero-infrastructure coordination substrate. The coding agent, the review surface, and the version store stay leverage points.
- Prefer adapter-shaped interfaces; today’s prose skill becoming tomorrow’s thin adapter to an external engine is fine.
- How much abstraction to build around a leveraged tool is P-0008-harness-agnostic-substrate’s rule; this principle only says leverage first.
Statement
Section titled “Statement”When external tools, frameworks, or agents are being built by parties with far more resources, SDLC leverages them (through strategy patterns, adapter layers, or direct integration) rather than building parallel implementations. This holds even when leveraging means today’s prose skill becomes tomorrow’s thin adapter to someone else’s workflow engine. Building is in scope only when a named property no external tool supplies makes leveraging impossible; “nobody’s is quite right” does not qualify.
Agentic coding tooling is being built at scale by Anthropic, OpenAI, GitHub, and many others. A solo project that builds parallel orchestrators, workflow executors, provider abstractions, or coding agent harnesses loses on quality and cadence, and the work does not compound because it has to chase what the leaders ship.
The payoff of leveraging is asymmetric: when an external tool reaches a higher progression stage for a workflow than SDLC’s own implementation, swapping it in is a net win. The cost is the discipline of adapter-shaped interfaces instead of hard-coded integrations, which P-0003-pragmatic-architecture already demands.
The carve-out: the coordination substrate. D-VSLI-distributed-work-runner-architecture
builds a scheduler and work-runner engine rather than adopting one. That is this principle applied.
No external engine supplies durable coordination with no server, no broker, and no database, over
git refs and markdown (github-ref-leases/README); every candidate in D-VSLI’s prior-art table
needs a hosted control plane or a database as system of record. Where prior art does supply the
semantics (Temporal’s event-sourced runs, Orleans’ virtual actors, OTP supervision, Kubernetes
reconcile loops, Airflow’s trigger vocabulary), it is adopted as semantics over the substrate.
The rule this yields: leverage the semantics, build only the layer no one ships, and name the missing property before claiming the carve-out.
How it applies
Section titled “How it applies”- Claude Code is the harness today because it is the best available, and it is swappable because the substrate does not depend on it (P-0008-harness-agnostic-substrate).
- Strategy patterns are the integration shape for orchestrators and workflow executors. The same
task lifecycle can be driven by SDLC’s own
task-workskill or by an external orchestrator with equivalent primitives, because the task’s state, lease, and PR-open contract are all externally observable. - Do not build what is not differentiating. Coding agents, IDE integration, test runners, and CI are leverage points, not build points.
Examples
Section titled “Examples”- Claude Code as today’s harness rather than a harness built from scratch.
- GitHub PRs as the review surface rather than a custom review tool.
- Obsidian as the editing and rendering UI, using its transclusion, link graph, and Bases features.
- git as the version store; no custom revision system or database-backed history.
- Astro Starlight for the docs site (
solutions/ontological/site-template/).
Implications
Section titled “Implications”- The deterministic workflow engine is built here rather than adopted (D-VSLI-distributed-work-runner-architecture) because no third-party engine runs on the zero-infrastructure substrate; see the carve-out under Why. The agent driver, the coding agent, the review surface, and the version store remain leverage points.
- P-0001-prefer-deterministic-over-llm’s Stage 4 implementations may be external: the LLM-as-tool role can be filled by SDLC’s own scripts or by an external agent platform.
- When a skill could be a thin adapter to an external capability or a full re-implementation of it, write the adapter. It is shorter, easier to swap, and improves when the tool improves.
Leveraging Claude Code today is consistent with the principle because the substrate does not depend on Claude. A leverage decision that would couple the substrate to one provider is the wrong shape.