docs(golem): add GitHub repository metadata with aligned description and topics

Collects all enchanter-ai repo descriptions as format reference, then
derives the golem description following the two-sentence hook/payload
pattern used across the ecosystem.

https://claude.ai/code/session_01XGVygYDP4Ys9EbZ9RHwdRZ

Author: claude

golem-github-metadata.md

@@ -0,0 +1,42 @@
+# golem — GitHub repository metadata
+
+## Description
+
+> Kill the silent zero. Compile-once expression engine that catches every undefined variable at parse time and hardwires deterministic cost gates and routing policies into AI agent pipelines.
+
+**Character count:** 190
+
+---
+
+## Topics
+
+```
+ai-agent, anthropic, deterministic, enchanter-ai, expression-evaluation, go, guardrails, llm, mcp, policy-engine, python, safety
+```
+
+---
+
+## Format reference
+
+All enchanter-ai descriptions follow two sentences:
+
+1. **Hook** — imperative or crisp noun phrase (≤ 6 words)
+2. **Payload** — one sentence, concrete mechanism + outcome, strong action verbs
+
+| Repo | Description |
+|------|-------------|
+| emu | Stop burning API tokens. 9 algorithmic engines for real-time context optimization, infinite-loop detection, and smart prompt compression. |
+| crow | Bayesian change-trust layer. Autonomously scores every AI code edit in real-time, instantly flagging the exact files that demand human review. |
+| lich | Ruthless, algorithm-backed code review. Tests AI-generated code in an isolated sandbox to mathematically prove it works before merging. |
+| wixie | Self-healing prompt engineering. Autonomously test, converge, and translate prompts across 225+ AI models. |
+| beholder | The foundational SDK for algorithmic AI orchestration. The central nervous system powering the Enchanter ecosystem with native MCP support. |
+| mimir | Verifiable provenance for MCP tool-call results — signed envelopes + σ-bound quality scoring + on-chain anchoring with EigenLayer slashing. Live on Sepolia. |
+| sylph | Zero-touch Git orchestration. AI autonomously manages parallel branches, segments tasks, writes Conventional Commits, and opens clean PRs. |
+| djinn | Cure AI memory loss. Pins session intent, actively prevents context drift, and keeps long-running agents hyper-focused on the original goal. |
+| hydra | Real-time AI threat interception. Blocks CVE-mapped attacks, poisoned configs, and destructive terminal commands before execution. |
+| vis | Conduct-as-code for AI agents. Refuses drift, sycophancy, and silent failure. |
+| gorgon | Static codebase intelligence. Maps structural hotspots and dependency cycles to guide AI agents safely through complex enterprise architectures. |
+| pech | FinOps for autonomous agents. Track per-tier token burn, forecast budgets via exponential smoothing, and trigger graceful AI degradation. |
+| naga | Source-as-spec architecture enforcer. Uses deep AST fingerprinting to ensure AI-generated code perfectly mirrors your internal coding standards. |
+| schematic | Bootstrap production-ready AI agents instantly. The official template injecting shared behavioral contracts and strict execution engines into new plugins. |
+| **golem** | **Kill the silent zero. Compile-once expression engine that catches every undefined variable at parse time and hardwires deterministic cost gates and routing policies into AI agent pipelines.** |