← ClaudeAtlas

kintsugi-core-guard-freelisted

Open-source deterministic verification guard for AI coding agents (Claude Code, Cursor, Codex CLI, Gemini CLI, OpenClaw). Enforces @KML/1 AST-IR guardrails, @LOGIC/2 5-Pillar Neuro-Symbolic Verification, @PROOF_DAG/1 Compliance Proofs, and @CREED/1 Autopoietic Evolution.
tunguyen1993/kintsugi-harness · ★ 0 · AI & Automation · score 74
Install: claude install-skill tunguyen1993/kintsugi-harness
# @SKILL_PACK_IR/1 [KINTSUGI_CORE_GUARD_FREE_V2_0_0] ```text @META[pack="kintsugi-core-guard-free" | ver="2.0.0" | tier="FREE_STARTER ($0)" | license="Apache-2.0"] @GATES[ KL-GROUND-001: "Refuse unverified 'not found' or 'all tests pass' claims without exit_code==0 & sweep counts" KL-FINOPS-001: "Circuit-break repetitive tool-call loops & enforce per-task token/cost budgets" KL-SEC-001: "Block pickle RCE (require weights_only=True), SSRF (169.254.169.254), and Bidi Trojan Source" LOGIC_2: "DPLL SAT/MUS + Stratified Datalog ProofDAG/1 + Exact Rational Difference Logic + @FALLACY_GUARD/1 + MHS Abduction" AST_LITE_V2: "Zero-dependency Structural AST-Lite across Python, JS/TS, Go, and Rust (strips comments/strings, indexes symbols & calls)" CREED_1: "P1:MUTATE(fail>=2->pivot) | P2:SYNTH(manual>=2->1step_tool) | P3:SCAR(bug->rule) | P4:PRUNE(overwrite_verbose)" ] @SELFTEST[cmd="python3 scripts/kintsugi_free_quickstart.py --selftest"] ``` ## @EDITION_MATRIX/1 (`Free $0` vs `Pro $15/$19` vs `Enterprise $29`) | Capability | Core Guard Free (`$0`) | Tier 1 FinOps Pro (`$15`) | Tier 2 Zero-Trust Pro (`$19`) | Tier 3 Enterprise (`$29`) | | :--- | :---: | :---: | :---: | :---: | | `CO_EXECUTION` | `Primary + 1 Free` | `Primary + 2..5 Free` | `Primary + 2..5 Free` | `Primary + 5 Free + 6 BYOK Paid` | | `PROOF_DAG_1` | `LOCAL_DEV_ATTESTATION` | `+ FinOps Policies` | `+ AppSec/MCP Policies` | `ENTERPRISE_PROVISIONED_SEAL` | | `SCARS` | `KS-001..016` (16) | `K