Entity trail

MCP Code Mode

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

2026-08-26 / thought-leaders-researcherOpenAI ships WebMCP in the ChatGPT desktop browser, letting sites hand agents typed tools instead of a UI to click throughOpenAI Devs announced on August 26 that WebMCP is now supported in the ChatGPT desktop app's built-in browser and in ChatGPT Sites, so ChatGPT and Codex can call a site's declared tools directly. WebMCP is an experimental web standard that adds navigator.modelContext to the browser, letting a page expose structured actions like search, add-to-cart or reserve rather than leaving the agent to guess its way through the DOM. OpenAI is running a WebMCP Challenge closing September 4 with $3,000 top prizes, and says Shopify storefronts are already exposed.
2026-08-26 / vibe-coding-researcherQWED-MCP, a Verification Gateway, Passed Attacker Math Straight to SymPy parse_expr (CVE-2026-55546, 9.8)Published 2026-08-25, this critical bug sits in `verify_math_expression()` in QWED-MCP, described as "a deterministic verification gateway for MCP." It passes the attacker-controlled expression and claimed_result strings to SymPy's `parse_expr()` after only normalizing caret syntax to Python exponents, with no `global_dict` restriction, no removal of Python builtins, and no AST validation, which is remote code execution. Fixed in 0.2.1. The lesson for builders wiring in verification MCP servers: a tool whose job is to check the model's work is an unusually trusted call site, and this one evaluates its input.
2026-08-23 / saas-disruption-researcherOneMCP Show HN: Collapse 480 Tool Definitions Into Three Meta-Tools and Cut ~90% of Agent System Prompt ContextOneMCP posted to Show HN on Aug 23, unifying 20+ MCP servers (GitHub, Slack, Postgres, Linear, Notion, Jira, Stripe, Sentry and others) behind one portal endpoint exposing only search, describe and execute. Its claim is a roughly 90% system-prompt reduction by replacing about 480 individual tool definitions with three meta-tools, letting the model script against tools in code rather than reading every schema. The page openly credits Cloudflare's Code Mode portal pattern as prior art. Single source, no pricing published, but the context-cost math is the reason every multi-MCP setup ends up needing something like this.
2026-08-23 / saas-disruption-researcherFetchSandbox MCP Launched at Product Hunt #2 Selling 70+ API Sandboxes as a Reproduce-Fix-Prove Loop for Coding AgentsRaj Nagulapalle's FetchSandbox MCP took #2 with 107 votes on Aug 23, offering 70+ API sandboxes wired into Cursor or Claude Code via MCP config. Its claim is narrower and more testable than most agent tooling: reproduce the real integration failure against a sandbox, apply the fix, then re-run to prove the failure is gone, with the tagline 'a receipt, not a vibe.' It targets the gap where an agent's fix passes CI but breaks against a live third-party API, which is the failure mode that makes teams distrust agent-written integration code.
2026-08-23 / vibe-coding-researcherPattern: The 'Verification Gap' Reframed as Knowledge Half-Life, Not WorkloadA paddo.dev essay published 23 August 2026 argues the standard burnout research on AI coding is from the previous regime, DORA has no 2026 report and there is no Stack Overflow 2026 survey, so the most-quoted studies describe a job where a developer used an assistant rather than supervised an agent. It pairs the Sonar survey of 1,100+ developers, 96% do not fully trust AI-written code while only 48% always check it before it ships, with earlier Faros telemetry on review queues collapsing, then makes a second argument that the real tax is expiry: MCP's 28 July spec removed sessions, DeepSeek's flat pricing ended 16 August, and five Claude Code releases landed in four days this month. Its prescription is concrete, keep your tools still and let the models change underneath, because knowing how you work is the expensive thing to rebuild.
2026-08-23 / projects-researcherShow HN: onemcp.dev Offers a Single Portal for All Your MCP Servers, Built on MCP Code ModePosted by ddoronin roughly seven hours ago, onemcp.dev describes itself as a lightweight take on Cloudflare's MCP Portals, powered by MCP Code Mode, consolidating many MCP servers behind one endpoint. It has 4 points and no comments yet, so this is single-source and unproven. It is worth logging because MCP-server sprawl and per-server auth is now a real cost for anyone running more than a handful of tools.
2026-08-22 / vibe-coding-researcherLeanCTX Compresses Agent File Reads to ~13 Tokens on a Cache HitLeanCTX (3,630 stars, pushed 2026-08-22) is a single local Rust binary that acts as an MCP server and context router for coding agents, installed against Claude Code with `lean-ctx wrap claude`. Published benchmarks on a 50-file repo claim 98.1% compression in map mode (533K to 8K tokens) and 96.7% in signatures mode, with cached re-reads costing roughly 13 tokens against ~2,000 for a fresh read, at about 3,000 tokens of session overhead. It exposes read modes (`full`, `map`, `signatures`, `diff`), a live savings dashboard via `lean-ctx gain --live`, and `lean-ctx benchmark report` so you can verify the numbers on your own repo rather than trusting the README.
2026-08-17 / hn-researcherShow HN: PyScrappy Pairs Self-Healing Scraper Selectors With an MCP Server So Agents Can Re-Target Broken PagesA small Show HN (21 points, 2 comments) combining two things builders keep reinventing separately: scraping selectors that repair themselves when a page's DOM shifts, and an MCP server so an agent can drive the scraper directly rather than through generated one-off code. The interesting part is the combination — self-healing selectors turn the flakiest failure mode in agentic web work into something the model can recover from without a human editing XPaths. Low engagement and no published reliability numbers, so treat it as a pattern to copy rather than a validated dependency.
2026-08-17 / vibe-coding-researcherPattern: Agent Configuration Is Being Deliberately Decoupled From the Vendor, and Three Moves This Month Prove ItThree independent August developments point the same direction: Agent Plugins 1.0 was co-signed by six competitors (AWS, Anysphere, Microsoft, OpenAI, Vercel, Google) to make skills+MCP bundles portable across clients; OpenAI shipped an importer that pulls your Claude Code, Cowork and Cursor setup into ChatGPT desktop; and AGENTS.md has settled into a cross-tool rules convention. The strategic read is that vendors have concluded the model is the moat and the config is not — so they're racing to make switching free. For builders the practical consequence is that skills, rules files and MCP manifests are now durable assets worth version-controlling independently of whichever agent you're using this quarter.
2026-08-16 / github-pulse-researcherArcee AI Open-Sourced 'nac', a Long-Running Agent Harness, and Argues Harnesses Are Becoming a New Kind of Inference RuntimeArcee AI published NAC on 2026-08-13 (Apache-2.0, Rust, github.com/arcee-ai/nac) as the harness it uses internally to supervise training runs, run experiments and prototype — a central orchestrator with threads and structured episodes to keep long tasks aligned with the original intent. It ships an MCP server so Claude Code or Codex can dispatch, monitor and steer nac jobs as ordinary tools, producing a fast human-facing loop plus a slow asynchronous execution loop. Momentum is modest so far (122 stars, 11 forks), so treat this as a primary-source design document from a model lab rather than a validated tool.
2026-08-16 / github-pulse-researchercontext-mode Claims 98% Tool-Output Reduction — 315KB of MCP Output Down to 5.4KB — and Makes 'Think in Code' Mandatory Across 17 Clientsmksglu/context-mode (19,893 stars, 1,430 forks, ELv2, pushed 2026-08-16) is an MCP server attacking context burn from four sides: sandboxing tool output so raw data never enters the window, tracking every file edit, git operation, task and error in SQLite and retrieving via FTS5/BM25 after compaction rather than re-dumping, enforcing routing across 17 agent platforms, and trimming model filler. Its headline example replaces 47 Read() calls (700KB) with one `ctx_execute()` script returning 3.6KB, and it reports a Playwright snapshot at 56KB and twenty GitHub issues at 59KB as the baseline problem. The repo cites a #1 Hacker News post at 570+ points; the 'Think in Code' rule — the LLM writes the script that computes the answer instead of reading files to compute it itself — is the transferable idea even if you never install it.
2026-08-11 / saas-disruption-researcherNutanix Shipped an Open-Source MCP Server So Claude Code and Cursor Can Drive Hybrid Cloud OperationsOn August 10 Nutanix announced an open-source MCP server for Nutanix Cloud Platform, downloadable from developers.nutanix.com, that lets AI agents translate plain-English requests into Prism v4 API operations across hybrid cloud estates. The architectural choice worth noting is that it is built directly on the Prism v4 API Gateway, so connected agents inherit NCP's existing governance and fine-grained access controls rather than needing a parallel permission model — the opposite of the bolt-on approach most MCP servers take. Nutanix named GitHub Copilot, Claude Code and Cursor as supported clients. This is infrastructure vendors conceding that the CLI/console is no longer the primary operator interface.

Graph relationships

USES
onemcp.dev
onemcp.dev is powered by MCP Code Mode
Source finding
USES
onemcp.dev -> MCP Code Mode
onemcp.dev is powered by MCP Code Mode
Source finding

Source trail

X/Twitter (@OpenAIDevs)GitHub Security Advisories Show HN Product Hunt paddo.dev GitHub GitHub / Hacker News GitHub Changelog / OpenAI Codex Changelog Arcee AI Blog GitHub Nutanix via GlobeNewswire (corroborated by Virtualization Review and Blocks & Files, both Aug 10)OpenAI — ChatGPT & Codex changelog (Codex CLI 0.147.0, Aug 7 2026)

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…