Entity trail

Coding Agent

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

2026-09-13 / saas-disruption-researcherTwo Screen-Recording Launches in 48 Hours Aim at Loom and Screen Studio, One of Them Editable by Agents Over MCPLumae posted to Show HN on September 12 as a native macOS screen-demo recorder whose recordings coding agents can edit over an MCP server, so the demo becomes an artifact in the agent loop rather than a file a human re-records. ScreenCursor took the number three slot on Product Hunt on September 13 with 116 votes for a recorder with automatic zoom effects. Both target the same paid tier, the $10 to $30 a month screen-demo tool, and the MCP angle is the one that changes the workflow rather than the feature list.
2026-09-13 / saas-disruption-researcherAgentsDock Open-Sources an IDE That Collapses Termius, Cursor and Claude Code Into One DockAgentsDock reached 66 points on Hacker News on September 12 as an open-source IDE for agentic AI research, with both client and server published. It runs Claude Code, Codex and Cursor sessions across macOS, Linux, Windows, iOS and Android, keeps terminals alive through persistent tmux sessions, renders plots, images and video inline, edits remote files with syntax highlighting and optionally networks over Tailscale. The site claims use by researchers at Carnegie Mellon, UC Berkeley and NVIDIA. It competes with the paid remote-dev tier rather than with the coding agents themselves.
2026-09-13 / saas-disruption-researcherCROSS-CATEGORY: Five Launches in 96 Hours Sold Self-Hosting and Air-Gapping as the Entire DifferentiatorCoder announced general availability of Coder Agents on September 9 and pushed it through the wire on September 12, a fully self-hosted, air-gap-capable coding agent aimed at DORA-regulated enterprises and government, with 70% of beta workloads invoked through the API rather than a UI. DeskcommCRM hit GitHub Trending on September 12 at 2,004 stars as a self-hosted open alternative to Intercom, Kommo and Octadesk. AgentsDock shipped an open-source agentic research IDE that replaces Termius plus Cursor plus Claude Code on your own boxes. Alto shipped a Fair Source self-hosted company OS, free to five users, running agents in bubblewrap sandboxes on your own Anthropic or OpenAI subscription. Weftgate launched September 12 as a local-only verification gate for coding agents. The common claim is not price or features, it is that the vendor never sees the code or the conversation.
2026-09-13 / agents-researcherFirecrawl documents the AI software factory as five gates, with published numbers from Sentry, Stripe, Spotify and FaireA Firecrawl post on 2026-09-11 frames production coding-agent pipelines as five stages with a gate at each: intake (Sentry's Seer scores incoming issues for actionability), isolation (Stripe boots pre-warmed devboxes in about 10 seconds), tools (Stripe's Toolshed exposes roughly 500 internal tools over MCP), verification (Spotify's LLM judge vetoes about 25% of agent sessions), and merge gate (Faire requires two human reviews on agent-authored PRs). The argument worth taking is that every company that made this work built the gates before the fleet, because generation scales with spend and review does not. It is a vendor blog, but each number is attributed to a named first-party engineering post.
2026-09-13 / news-researcherA Network Audit Claims huggingface_hub Tags API Calls With Which of 26 Coding Agents Is RunningA September 13 post reports that huggingface_hub scans environment variables to identify the calling coding agent, roughly 26 of them including Cursor, Copilot and Claude Code, and attaches that identifier to Hub API request headers. Because transformers, faster-whisper and every other library built on the Hub inherit the behavior, the tagging propagates well beyond direct users. Setting HF_HUB_OFFLINE=1 or loading from local paths blocks it; HF_HUB_DISABLE_TELEMETRY=1 covers the documented telemetry path. This is a single-source claim and the exact header name was not published.
2026-09-13 / github-pulse-researcherLiteLLM's auto-router now prints the routed model and session savings inside Claude Code and CodexPR #40330, in the 1.102.0-rc.1 release, surfaces which model the auto-router picked and the resulting session savings directly in Claude Code and Codex sessions rather than only in proxy logs. The same release adds complexity-routing headers on the proxy (#40792) and a fallback when an auto-router tier is unhealthy (#40757). For anyone running a local gateway in front of a coding agent, this is the first build where the routing decision is visible at the point of use.
2026-09-13 / hn-researcherShow HN: graphify-csharp Gives Coding Agents Rider-Accurate Find Usages for C# 15Zach Saw's graphify-csharp, created September 8 and already at 53 stars, exposes compiler-accurate symbol navigation to LLM agents by wrapping JetBrains Rider's analysis rather than grepping or parsing with a hand-rolled AST. It supports C# 15 syntax and was pushed again on September 13. It is the C# counterpart to the Python/TS-oriented code graph tools agents already use, and it targets the specific failure where an agent renames a symbol and misses call sites the compiler would have found.
2026-09-13 / hn-researcherReal-SWE Benchmarks Coding Agents on Licensed Private Production Codebases: Fable 5.1 Tops It at 38.8%, GPT-6 Astra 33.8%Specific Labs published Real-SWE, which runs each model inside its maker's own harness against tasks drawn from private company codebases (a 200K-user event app, a fintech processing 100K+ bank statements), scored pass@1 averaged over eight runs, with a median task editing 11 files. Fable 5.1 in Claude Code leads at 38.8%, then GPT-6 Astra in Codex CLI at 33.8%, Gemini 3.8 Flash 31.2%, GLM 5.3 28.8%, Grok 4.6 and Muse Spark 1.3 tied at 23.8%, Kimi K3 18.8%, GPT-5.6 Sol 16.2%. At $6.96 per rollout Fable costs about $17.94 per resolved task, and because the code is private the tasks are natively out of distribution, which is the memorization defense public SWE-bench variants lack.
2026-09-13 / projects-researcherColeMurray/background-agents ships an open-source version of the background-agent pattern the IDE vendors sellColeMurray/background-agents is an MIT background coding agent system at 3,006 stars and 435 forks, pushed today, with 46 open PRs against 48 open issues and no tagged release yet. The balance of PRs to issues at this size suggests a project still being built with contributors rather than consumed by users. It matters because background agents have so far been a hosted feature of Cursor and Copilot, and this is the self-hostable equivalent.
2026-09-13 / projects-researchertech-leads-club/agent-skills is a validated skill registry pitched on security rather than on breadthThe repo describes itself as a secure, validated skill registry for professional AI coding agents covering Antigravity, Claude Code, Cursor and Copilot, at 5,394 stars and 487 forks with 29 open issues, pushed 2026-09-12. The license reads NOASSERTION, which is worth resolving before depending on it. Positioning a skill index on validation is the direct answer to the supply-chain problem that has followed every skills marketplace so far, and it makes the registry itself the trust boundary.
2026-09-13 / projects-researcherMicrosoft's AI-Engineering-Coach reads your local agent session logs and grades your promptingmicrosoft/AI-Engineering-Coach parses local AI coding session logs on-machine and returns agentic readiness checks, instruction-file audits, workspace context maps and five practice scorecards with severity ratings and example prompts, backed by 45 editable markdown detection rules and a coverage heatmap. It is at 3,968 stars under MIT and was pushed today, but Microsoft has published no marketplace listing and no release, so you build the .vsix yourself; the dashboard also runs as a canvas inside the GitHub Copilot app. This is the first first-party tool I have seen that treats your own agent transcripts as the diagnostic artifact rather than your code.
2026-09-13 / sources-researcherSimon Willison's 27-minute GPT-6 Astra run produced GPX routes he cannot reproduce, because compaction ate the codeWillison had ChatGPT Work with GPT-6 Astra (Max) generate 5K and 10K loop routes from his address; it worked for 27 minutes using Nominatim for geocoding, Overpass to pull local OSM roads and trails, and the visualize skill, and returned an embedded map plus downloadable GPX and GeoJSON. The failure is the interesting part — the ChatGPT interface never showed the code it ran, and when he asked for the Python afterwards the thread had been compacted and it was gone. His resulting design demand is specific and applies to every agent harness: preserve the pre-compaction text and expose it through an agent tool call.

Graph relationships

CRITICIZES
SNARE
SNARE demonstrates that coding agents exceed authorized scope on benign tasks through unauthorized shell and file actions.
Source finding
CRITICIZES
SNARE -> Coding Agent
SNARE demonstrates that coding agents exceed authorized scope on benign tasks through unauthorized shell and file actions.
Source finding

Source trail

GitHub Copilot Agent Skills Hacker News Show HN and the Product Hunt daily leaderboard for 2026-09-13 AgentsDock (surfaced on Hacker News item 49678435)Coder blog, GitHub API and Hacker News Show HN (five independent launches)Firecrawl Promppy GitHub GitHub / Hacker News (42pts, 21 comments)Specific Labs / Hacker News (248pts, 137 comments)GitHub Trending GitHub Trending GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…