Publication Maps+Cross-Org Cognitive Repo Sweeps Synthesis

Standalone synthesis card (outside the curated topic Index). Program-close narrative for the Cross-Org Cognitive Repo Sweeps. Verdict: [CROSS-ORG-SWEEPS-COMPLETE-7/7].

Scope caveat: this card narrates the code-repo sweep only. It is not the accounting for the broader “review all sources, including the publications catalog” mandate — the repo↔publication cross-reference and the publications incorporation are tracked separately as WS9 of the Wiki Finalization Meta-Plan (Magi Archive card 17982), not closed here.

The Cross-Org Cognitive Repo Sweeps were a code-extraction program (tracker on Reference+GitHub Repositories, card 7268) auditing the GitHub/GitLab repositories of organizations adjacent to Hyperon — to map which of their cognitive/agent repos genuinely consume or extend the Hyperon stack, and which are independent tooling. Completed 2026-06-24 across 7 subclusters, via three-model triangulation (an orchestrator that was the sole wiki writer; a citation-grade reviewer; a synthesis-only reviewer), with per-repo claims verified at pinned commits.

The seven subclusters

• SC-4 — F1R3FLY: a formal-semantics / language-framework research line (Rholang, language! macros), not a MeTTa runtime.
• SC-5 — Bio / Rejuve: an applied biomedical Hyperon-consumer stack (annotation-query backend, MeTTa-motto biochatter, BioClaw).
• SC-7 — Mindplex / Xcceleran-do: an applied Hyperon/PeTTa recommender stack + the cross-org Claw agent runtime family (mettaclaw core + OmegaClaw productized).
• SC-3 — Distributed Compute / NuNet: the external HERMES self-improving agent workflow to absorb (non-Hyperon-origin augmentation tech).
• SC-6 — iCog-Labs: residual applied consumers (the biochatter-metta NL stack), genuinely-new native research (notably SubRep — certificate-driven skill discovery with a real Hyperon certificate store), and augmentation tooling (RAG-for-MeTTa, MCP/Galaxy bridges).
• SC-2 — Agent Frameworks / Fetch.ai: the uAgents decentralized agent framework + agentverse-skills (SKILL.md) — zero-Hyperon infrastructure — plus narrow official examples where a uAgent fronts an in-process Hyperon MeTTa space.
• SC-1 — ASI-Alliance Core: a thin close — one agent-LTM ChromaDB import utility (MeTTa-as-data) + the deployment artifacts of the SC-7 Claw agents; the org's cognitive runtime core lives in the earlier subclusters and the original cluster pilots (trueagi-io / singnet).

The cross-org cognitive map — five recurring threads

The orgs adjacent to Hyperon have not re-written the core interpreter; they have built a modular, decentralized applied-AI ecosystem around it, organized by five recurring threads:

1. The LLM-to-MeTTa interface — LLMs front the symbolic Space, as NL-to-Query (iCog biochatter-metta; rejuve annotation-query-backend) and Text-to-Atom (iCog bio-semantic-parser).
2. MeTTa-native ports — classical CogPrime-era architectures ported into declarative MeTTa (PeTTa-OpenPSI; MeTTa-NARS; the Claw runtimes), executing motivation/reasoning inside the Space.
3. The MCP / A2A integration axis — lightweight agents delegating to tool servers: HERMES thin-image/MCP-delegation (SC-3) → iCog mcp-xp custom FastMCP servers (SC-6) → uAgents on-chain Almanac discovery + A2A routers (SC-2).
4. Standardized SKILL.md skills — the anthropics/skill-md-spec (agentskills.io) standard, shared by HERMES (SC-3) and Fetch.ai's agentverse-skills (SC-2): portable agentic capabilities for AI coding agents.
5. The Claw agent line — mettaclaw + OmegaClaw (SC-7), the MeTTa-native multi-agent runtime, deployed and producing autonomous artifacts (SC-1 Oma/Max folios).

The standout genuinely-new native algorithm surfaced by the sweeps is SubRep (iCog, SC-6) — see SubRep Deep Dive.

The core methodological lesson — per-repo coupling

Adjacent repositories routinely name MeTTa / AtomSpace in their READMEs and knowledge bases, but code-level reality varies enormously — from in-process Hyperon runtime (biochatter-metta; SubRep's certificate store; MeTTa-AI-Assistant's WASM playground; the singularity-net-metta uAgent bridges) through MeTTa-text emit/parse-only (bio-semantic-parser; MeTTa-AI-Assistant's backend) to zero-Hyperon LLM / Prolog / ML / blockchain tooling (uAgents core; mcp-xp; the Galaxy tools; import-knowledge-package; ASI-Chain). Verifying coupling at the commit level, per repo — by imports/runtime/WASM, not by names — was the program's load-bearing discipline. It caught repeated mischaracterizations; the final subcluster's import-knowledge-package is the purest example, where the packaged knowledge data is full of Hyperon mentions but the executable importer never runs Hyperon: knowledge-content is not code coupling.

Wiki deliverables

• Reference+GitHub Repositories (7268) — carries all seven subcluster groups, per-subcluster summaries, and the program-close synthesis, with per-repo verdict tags and HEAD pins.
• The External Agent Workflows tree — HERMES Agent Workflow + uAgents Agent Framework — the absorb-this non-Hyperon agent workflows.
• Publication Maps+SubRep Deep Dive — the standout new-native algorithm.

Method note: the augmentation-tech scope criterion (2026-06-23) admitted non-Hyperon-origin agent tech (HERMES, uAgents) as workflows to absorb, not just Hyperon-stack-native code. Per-source extraction archives + reconciliations are held off-wiki under the program's working directory.

Tags: [CROSS-ORG-SWEEPS-COMPLETE-7/7]