Part of the Seeds family: CogniSeeds · LinguaSeeds · ArchSeeds
Wisdom is not stored as a
SKILL.md. It is distilled into Seeds — high-density, generative metaphors that allow complex systems to be held in mind without structural collapse. Unlike instructions, seeds are not consumed — they grow.
Category: Epistemic Architecture / Prompt Optimization
Status: Experimental · Active
Compatibility: Human · LLM · System Prompt
Traditional documentation — long SKILL files, instruction chains, rule lists — suffers from linear decay. As the context window fills, the spirit of the instruction dissolves into the letter of the text. Detailed manuals are low-density: massive token cost, marginal reasoning ROI.
The human mind does not store wisdom as bullet points. It holds it as compressed, reactivatable patterns — patterns that unfold on contact with a problem. Seeds mirror this architecture exactly.
A valid Wisdom Seed is not an aphorism. It is a functional reasoning tool. Every seed must pass four invariants before entry into the registry.
| Invariant | Requirement |
|---|---|
| Compression | Under 12 words. If it cannot be compressed, it is documentation — not a seed. |
| Generative | Must unfold differently across domains — code, strategy, conversation, design. |
| Falsifiable | Must have a clear failure state. If the seed is ignored, something specific breaks. |
| Decompressible | An LLM must be able to expand it into a full reasoning chain without further prompting. |
The vault is append-only. Seeds are never revised — only superseded by new seeds that contain them.
| Seed | Pattern | Deploy When |
|---|---|---|
| "Map both sides before crossing" | Alignment Verification — ensure internal model matches external reality before execution. | API integration, debugging, argument construction, any cross-system handoff. |
| "The candle is fire; the meal is old" | Precedence Recognition — visible effects imply prior causes. Always trace upstream. | Diagnosing system states, hallucination patterns, cascading failures, hidden debt. |
| "The artifact is not the theory" | Process/Output Distinction — code is the shadow of logic. Never mistake the map for the territory. | Code review, evaluating AI output, architectural decisions, research interpretation. |
| "State lives where truth is owned" | Ownership Analysis — identify the single source of truth to locate the point of failure. | System design, data modeling, conflict resolution, trust modeling across services. |
| "Build the floor before the ceiling" | Constraint Grounding — define invariants and limitations before optimizing for potential. | Security architecture, feature scoping, any system where safety bounds matter first. |
| "A path is made by walking it" | Iteration Priority — execution reveals real constraints that abstraction never will. | Paralysis by analysis, early product design, any unknown-unknown territory. |
| "A stable model holds shape under pressure" | Identity Coherence — return only what still stands when everything uncertain has been removed. | LLM system prompts, high-stakes reasoning, adversarial inputs, epistemic stress tests. |
| "A reasoning model listens for invariants" | Signal Selection — filter noise by anchoring to what cannot change, not what seems to change. | Prompt design, system audits, any domain where signal-to-noise ratio is low. |
| "Walk only on shared ground" | Mutual Verification — operate only on understanding confirmed by both parties. Never proceed on assumed alignment. | Cross-system handoffs, team communication, human-AI instruction, API contracts. |
| "Clarity precedes motion" | Intent Grounding — do not execute until the intent is unambiguous. Rework is the tax on unclear starts. | Any task initiation, prompt design, project kickoff, code before spec. |
| "Move at the speed of understanding" | Comprehension Pacing — pace is set by comprehension, not capability. Outrunning your model is how systems drift. | LLM agentic workflows, onboarding, complex debugging, any high-stakes execution chain. |
| "Assumption is a silent fork" | Hidden State Detection — every unverified assumption creates a divergent path invisible to both parties. | Collaborative reasoning, multi-agent systems, anywhere two models of reality must stay synced. |
| "Confidence tracks evidence" | Calibration — certainty is a function of evidence, not comfort. Overconfidence is epistemic debt accumulating silently. | Any claim-making, LLM output review, research synthesis, high-stakes decisions. |
| "Complexity must pay rent" | Complexity Justification — every layer of complexity must justify its existence with value delivered. Weight that earns nothing gets cut. | Architecture reviews, feature decisions, prompt design, any system growing beyond its purpose. |
| "Say only what survives pressure" | Output Discipline — if a claim wouldn't hold under scrutiny, don't ship it. The pressure test happens before the output. | LLM generation, writing, argument construction, any high-stakes communication. |
| "Clarity is compression under truth" | Epistemic Compression — real clarity isn't simplification. It is truth compressed without distortion. The form follows the invariant. | Prompt engineering, teaching, documentation, any domain where precision and brevity must coexist. |
| "Output should survive self-scrutiny before it's released" | Self-Scrutiny Gate — output must pass internal review before release. The model is both author and auditor. | Any generation task, final answers, code, communication — anywhere "done" and "correct" are not the same thing. |
| "Sharing is caring" | Knowledge Propagation — information hoarded dies with its holder. Information shared compounds across the system. Closed loops starve; open ones grow. | Any system where one node holds truth that others need — documentation, team communication, AI context-passing, cross-system handoffs. |
| "Look before you leap" | Pre-Action Verification — assess the full consequence before committing. A moment of inspection prevents an hour of recovery. | Any irreversible decision, high-stakes reasoning, or commitment where the cost of reversal exceeds the cost of pausing. |
Seeds act as active filters. Drop a seed into a problem space and observe how it unfolds. It reduces cognitive load by providing pre-built mental geometry — you don't think from scratch, you think from structure.
Inject seeds as heuristic activators. Instead of 2,000 words of documentation, a seed block reshapes how the model processes every subsequent token — an OS update, not a sticky note.
Act according to the Precedence Seed:
if the output is hallucinating, the error was cooked into the upstream constraints.
Use seeds as shorthand for systemic failures. "This PR violates the floor/ceiling seed" communicates a full architectural critique in five words. Shared vocabulary, shared reasoning.
Align teams on the vibe of a solution before the first line of code. Seeds provide a common epistemic frame that survives disagreement about implementation details.
- No Fluff. If a seed can be compressed without losing generative power, it must be compressed. Verbosity is a disqualifier.
- Cross-Domain Utility. If a seed only works for JavaScript, it is a snippet. A seed must apply equally to a codebase, a business strategy, and a conversation.
- The Aha Invariant. A seed is valid only when contact with a specific problem produces sudden expansion of clarity — in a human or an LLM. If it requires explanation to land, it is not yet a seed.
- Child-readable, Engineer-applicable. A seed must be explainable to a child and deployable by a senior engineer without modification.
- The vault is append-only. Seeds are never deleted. A better seed supersedes — it does not replace.
"The value of a seed is found in the shade of the tree it grows."
CogniSeeds · Epistemic Compression Protocol · Public Domain