jtims

AI Knowledge Repository — AI skill for Claude Code

AI community

Validated knowledge graph of hard platform constraints and decision-enabling capabilities for Claude Code, Lovable, Git, reMarkable, n8n, Python and Atlassian, with a contract validator.

How to install AI Knowledge Repository

This entry records only its repository, not the path inside it, so there is no exact command to give. Open jtims/ai-knowledge-repository and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Knowledge Repository does

Validated knowledge graph of hard platform constraints and decision-enabling capabilities for Claude Code, Lovable, Git, reMarkable, n8n, Python and Atlassian, with a contract validator.

Alternatives in AI

Cl4r1t4s — LEAKED SYSTEM PROMPTS FOR CHATGPT, CLAUDE, GEMINI, GROK, PERPLEXITY, CURSOR, LOVABLE, REPLIT, AND MORE 47.2k ★
Sentrux — Real-time architectural sensor that helps AI agents close the feedback loop, enabling recursive self-improveme 2.9k ★
Ante — Ghost in your shell 1.9k ★

README

AI Knowledge Repository

Two platform knowledge ledgers for AI-assisted engineering, built as a validated knowledge graph: a **Constraints Ledger** of hard, design-invalidating platform facts, and a **Capabilities Ledger** of decision-enabling affordances. Every node is atomic, typed, provenance-backed and validated by a contract linter before it counts.

At this snapshot: 84 nodes across 10 platform entries in 2 ledgers, validator PASS with 0 errors. Constraints: 70 nodes over Claude Code, Lovable, reMarkable, Git, n8n, Python and Atlassian, 69 active and 1 superseded. Capabilities: 14 nodes over Claude Code, reMarkable and Lovable.

Why it exists

A platform fact that sits in memory is not a fact that gets applied. The founding defect was a design shipped on a backend path that a known platform limitation made impossible; the limitation had been recorded weeks earlier and was still missed at design time. The fix was structural: hard constraints became atomic notes with an admission bar and provenance, a hook injects the relevant platform's nodes into context when a prompt names that platform, and no ledger change counts as done until the validator passes.

How a node is shaped

Frontmatter carries the identity and the evidence: a monotonic per-platform id, platform, domain, severity or leverage, status, one or more provenance entries with type, ref and date, a verified date, and typed edges to related nodes. The body of a constraint carries three required sections: the constraint itself, its design consequence, and the sanctioned workarounds. A capability carries the inverse shape: what it unlocks, and its caveats and linked constraints. Nodes are never deleted; a wrong node is superseded and linked.

Platform Constraints Ledger/
  _SCHEMA-Constraints.md          the contract and id registry
  _MOC-Platform-Constraints.md    the global hub, with a dated changelog
  /_-Constraints-MOC.md
  /constraints/_.md
  _too