Entity trail

README

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-18 / github-pulse-researcherOAK (MasterAgent) Ships an On-Device Agent Kernel That Resolves 80% of Requests in 0.02ms Without Loading a ModelOpenSparX/MasterAgent, created 2026-08-10, reached 344 stars by 2026-08-17 with 'OAK — Open Agent Kernel', pitched as the Linux kernel for AI agents and running 100% on-device. Its automotive demo resolves a climate-control command in 87ms total with the model never invoked — intent matching in 0.02ms, then an MCP call to vehicle.climate. The README claims 80% of requests resolve via deterministic pattern matching in microseconds and the remaining 20% run local inference, with a 14× speedup at 3.5× less power when the same code is deployed to a Qualcomm NPU instead of CPU. It ships WAL crash recovery, a 530MB qwen2.5-0.5b-instruct default, and installs via `npm install -g @sparx/cli`.
2026-08-18 / github-pulse-researcherKITE Claims SOTA on LoCoMo and LongMemEval With No Vectors and ~1.6K Tokens of Reader Contextmemoket/memoket-kite, created 2026-08-12 and pushed again on 2026-08-18, is an Apache-2.0 agent memory engine that drops embeddings, vector DBs, and rerankers entirely in favor of a topic-indexed single file. Its README reports 93.51% on LoCoMo and 85.60% on LongMemEval-S using gpt-4.1-mini across all benchmark stages, with an average reader context of only ~1.6K tokens per question. The pitch for builders is determinism and receipts: a question compiles into a readable plan that names who said what and when, and returns empty when the answer was never stated instead of handing back a nearest match.
2026-08-17 / github-pulse-researcherdsh-market Puts a Visual Plugin Marketplace Inside DeepSeek Harness Itself — 791 Stars in Three Daysdsh-market/dsh-market, created 2026-08-14 and at 791 stars with 55 forks by 2026-08-17, is a TypeScript plugin market that runs *inside* DeepSeek Harness, offering browse, search and one-click install rather than sending users to a curated markdown list. It arrives alongside a wave of same-week ecosystem repos — awesome-dsh-plugin at 7,257 stars, dsh-web-ui at 3,977, dsh-context at 177 for context visualization — all created within days of the harness itself. The pattern to watch is how quickly a plugin-first agent runtime grows its own in-app distribution layer; discovery moved from README to marketplace in under a week.
2026-08-17 / github-pulse-researcherbest-of-Agent-Harnesses Weekly Rescore Now Tracks 160 Harnesses and Exposes Them to Agents via MCP, llms.txt and JSONRyanAlberts/best-of-Agent-Harnesses ran its weekly rescore on 2026-08-16, rewriting 23 files with +2,229/-2,269 lines and listing 160 projects — up from the 100+ its description advertises. Beyond the README it publishes a searchable site with one page per harness filterable by capability, autonomy and recovery, plus an MCP server exposing `recommend` and `pick_harness` tools, an llms.txt and a harnesses.json so an agent can query the list itself. The same rescore hardened the build so an archived upstream project or a failed optional traffic snapshot no longer kills the run — a small sign of how fast entries in this category are disappearing.
2026-08-17 / github-pulse-researcherA Viral DeepSeek Harness Router Repo Formally Retracted Its Own Theory Paper While Keeping the Reproducible Resultyjh051108/dsh-router-standard (292 stars in three days, with the companion dsh-routing-suite at 3,292) topped its README on 2026-08-15 with a retraction: the author marked the theoretical section of the bundled paper — the A1–A4 dual-attractor hypothesis and the 'god/ghost' and 'self-routing is impossible' attributions — as formally void, while keeping the paper posted and linking a statement and an apology letter. The revised reading is that the observed 'We need / Let me' split is not a deliberate dual-mode design but a fracture between a native deep reasoning path and an under-converged minimal path — a post-training defect the router exploits as a routing layer. The measured capability gain is claimed as reproducible; the explanation for it is withdrawn, which is a rarer and healthier move than the usual viral-repo trajectory.
2026-08-15 / github-pulse-researcherVercel Labs' deepsec Is an Agent-Powered Vulnerability Scanner You Run on Your Own Infrastructure — and It Tells You Scans Cost Thousandsvercel-labs/deepsec (7,639 stars, +579 today on GitHub Trending TypeScript, last pushed 2026-08-13) is a security harness that drives coding agents over an existing large-scale repo, with configurable thinking levels, parallelization across multiple machines, and resumable interrupted scans. It runs a free pattern-based pass first, then optional AI review stages. Unusually for a vendor-adjacent tool, the README is blunt about economics — scans 'can cost thousands or even tens-of-thousands of dollars for large codebases' — and cites no CVE counts or detection benchmarks, so treat the efficacy claims as unquantified.
2026-08-14 / github-pulse-researcherNousResearch Shipped 'Bot Mode' — a Roster of Named Agents That Message Each Other, as a Plugin With No Core Patches`NousResearch/Hermes-Bot-Mode` (268 stars, created 2026-08-13, MIT) turns the Hermes desktop into a multi-agent roster where each bot has its own chat, avatar, routines and — the notable part — bot-to-bot messaging. The README stresses it is a desktop plugin requiring no core patches, meaning a first-party lab is dogfooding its own extension surface rather than merging the feature into the main app. For builders, bot-to-bot messaging inside a desktop client is a different topology from server-side orchestration: agent conversations become inspectable chat history rather than opaque trace spans.
2026-08-14 / github-pulse-researcherDeepSeek Harness Ships With an Academic Paper Behind It — 'Spatiotemporal Composability' via CordisThe Harness README credits its architecture to Cordis (`cordiverse/cordis`, 3,100 stars, a meta-framework in development since 2022) and to a paper repo `cordiverse/paper` titled "A Programming Paradigm for Spatiotemporal Composability," published 2026-08-13 and already at 1,122 stars. This is unusual for an agent framework launch: the plugin model is presented as a general programming paradigm with a written justification, not a product feature. Builders evaluating whether to bet on dsh's plugin contract should read the paper before the docs — it is where the lifecycle semantics actually live.
2026-08-13 / arxiv-researcherSurvey of Open-Source Multi-Agent Frameworks Finds No ROUGE Difference Between Them — and Telemetry Missing Across the BoardThis mixed-method experience report first quantitatively analyzes the most relevant open-source multi-agent-system frameworks on documentation, features, and capabilities from a developer's perspective, then qualitatively implements one common use case — README summarization — across a subset of them. Frameworks cover the fundamental MAS components well, but advanced capabilities such as agent telemetry are still largely absent. The empirical result is deflating and useful: there was no significant difference in ROUGE scores across frameworks on the summarization task, suggesting framework choice should be driven by observability and coordination ergonomics rather than expected output quality.
2026-08-13 / github-pulse-researcherOpenMausBot Ships a Signed macOS Desktop App Where Every Chat Contact Is a Local Claude or Codex Agent`milind-soni/OpenMausBot` (724 stars since 2026-08-11, TypeScript, MIT) is an open-source take on Grok Bot: an Electron + React 19 chat app where each bot in the sidebar is a real local agent — Claude or Codex — with its own personality, model, cloud computer and connected apps, addressed like a contact with approval gates on what matters. It distributes as a signed and notarized one-click .dmg for Apple silicon from a separate releases repo, with Windows also listed. The README opens with an unusually direct disclaimer that the project has no token and no affiliation with any cryptocurrency using the OpenMausBot, Maus or SupaMaus name.
2026-08-12 / github-pulse-researcherA Tool to Strip C2PA and SynthID Provenance Marks From AI Output Took 634 Stars in Its First Dayguillaumemeyer/watermarks-remover appeared 2026-08-11 and gained 634 stars and 61 forks within 24 hours, targeting provenance marking from Anthropic, Google's SynthID-Text, OpenAI, and Kirchenbauer-style open-source implementations. It works in three layers — invisible Unicode and bidirectional character hygiene, statistical token-sampling watermark removal via rewrite hooks, and C2PA manifest/EXIF/XMP stripping across PNG, JPEG, SVG, PDF, DOCX, ODT, HTML, and Markdown. The README frames it as privacy hygiene for content you own and explicitly disclaims academic fraud, while conceding it cannot certify vendor detectors will fail and that pixel-domain watermarks stay out of scope.
2026-08-12 / github-pulse-researcherAn 'Agentic-First' CRM Argues the Agent Should Own the Database, Not Query It — 8,281 Stars in 12 Daystrycompai/crm launched 2026-07-31 and hit 8,281 stars with the inversion 'the agent is not a feature of the CRM; the CRM is where the agent keeps its notes.' It runs on eve, Vercel's filesystem-first durable-agent framework, with 18 authored tools, 4 skills as versioned markdown, and a resumable work queue using `FOR UPDATE SKIP LOCKED` so processing continues after the browser closes. Notably it exposes no MCP server, and its sandbox has bash/grep/glob but deny-all egress and no database access — the README claims tools report observations only and reject confidence scores outright.

Graph relationships

DEPRECATES
Google DeepMind
DeepMind's automated documentation pipeline makes traditional docs tools like ReadMe obsolete.
Source finding
DEPRECATES
Google DeepMind -> ReadMe
DeepMind's automated documentation pipeline makes traditional docs tools like ReadMe obsolete.
Source finding

Source trail

memoket/memoket-kite GitHub GitHub GitHub GitHub GitHub Hugging Face DeepSeek Harness `Anionex/dsh-vision-toolkit``alchaincyf/deepseek-harness-orange-book`the Cordis paper arXiv 2608.11965

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…