██╗  ██╗███████╗██╗   ██╗██████╗ ██╗███████╗████████╗██╗ ██████╗  ██╗      █████╗ ██████╗ 
██║  ██║██╔════╝██║   ██║██╔══██╗██║██╔════╝╚══██╔══╝██║██╔════╝  ██║     ██╔══██╗██╔══██╗
███████║█████╗  ██║   ██║██████╔╝██║███████╗   ██║   ██║██║       ██║     ███████║██████╔╝
██╔══██║██╔══╝  ██║   ██║██╔══██╗██║╚════██║   ██║   ██║██║       ██║     ██╔══██║██╔══██╗
██║  ██║███████╗╚██████╔╝██║  ██║██║███████║   ██║   ██║╚██████╗  ███████╗██║  ██║██████╔╝
╚═╝  ╚═╝╚══════╝ ╚═════╝ ╚═╝  ╚═╝╚═╝╚══════╝   ╚═╝   ╚═╝ ╚═════╝  ╚══════╝╚═╝  ╚═╝╚═════╝ 
             ░░░░░░░░░░░░░ S K U N K   W O R K S // D E F E N S E   G R A D E ░░░░░░░░░░░░░

HEURISTICO LAB is an applied systems laboratory researching deterministic execution, zero-trust context firewalls, and air-gapped cognitive infrastructure for regulated environments.

We engineer runtime layers that intercept, verify, and mathematically bound context consumption and tool execution before inference reaches production infrastructure.

Open-core AI Agent Governance & AST Context Compactor (pip install ctxfw), officially indexed in the Anthropic MCP Registry (io.github.heuristicolab/ctxfw) and Glama.ai (Grade A #1):

• Polyglot AST Pruner: Certified 57.3%–72.4% token reduction via deterministic AST signature extraction.
• SARA (ctxfw.sara): Sub-15ms lexical sanitizer and HMAC-SHA256 runtime tool-call authorization gate.
• AxiomWeaver (ctxfw.evaluate_spec_axioms): Enforces $\ge 5$ negative invariants and $\text{ACI} \ge 0.9000$ before code synthesis.
• SWE-Gate & TraceKit: Static AST patch verifier (DQI >= 0.9000) and tamper-evident SHA-256 SQLite WAL audit ledger.

Production-ready, cryptographically sealed 12-artifact engineering capsules (ACI = 1.0000, DQI >= 0.9000, Zero-Stub Hermetic Compiler):

• 7 Certified Sovereign Capsules ($290 USD — Instant 24h Delivery): Led by SARA-01: Zero-Trust Runtime Action Authorization Gate, Claude Rules Compiler, BreakGuard, SymTrace, TraceKit, SWE-Gate, and TokenCast.
• 45 Pre-Validated P0 Blueprints ($1,450 USD — 48h Dedicated Forge): Across V1_AGENT_SECURITY, V2_FINOPS_TOKEN_ECONOMICS, V3_DETERMINISTIC_CODE_QUALITY, V4_RESILIENT_ORCHESTRATION, and V5_FORENSIC_COMPLIANCE.

Autonomous coding agents and indexers can consume our specifications directly via Accept: text/markdown or dedicated GEO endpoints:

• Interactive Vault Showroom (EN / ES): https://heuristicolab.com/vault | https://heuristicolab.com/es/vault
• Vault M2M Markdown Index: https://heuristicolab.com/vault/llms.txt
• Flagship SARA-01 M2M Spec: https://heuristicolab.com/vault/sara-action-induction-runtime-auth/llms.txt
• Agent Discovery Card (A2A / MCP / UCP): https://heuristicolab.com/.well-known/agent.json
• Live M2M Telemetry Ledger: https://heuristicolab.com/api/v1/vault/telemetry

pip install --upgrade ctxfw
ctxfw doctor