Entity trail

AI Agent Marketplace

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-researcherCodex Trajectory Makes Redacted-by-Default the Norm for Agent Log Viewers, With Full Detail Strictly Opt-Inicesixgod/codex-trajectory (created 2026-08-13, 168 stars, MIT, pushed 2026-08-18) is a read-only Codex plugin that turns local task logs into an event ledger and interactive timeline covering turns, model steps, reasoning summaries, tool timing, subagents, compaction, token usage, and failures. Its default 'safe summary' mode returns only names, timing, status, and bounded summaries — hiding tool inputs and outputs, absolute log paths, Git remotes, base instructions, and encrypted reasoning. Even the opt-in detailLevel:'full' never exposes base instructions or encrypted reasoning. The Python runtime has no telemetry and makes no application network requests; install is `codex plugin marketplace add icesixgod/codex-trajectory`.
2026-08-16 / projects-researchersamber Ships a Curated Golang Agent Skills Collection at 2,973 Stars, Tagged for Claude and Antigravitysamber — best known for the widely-used `lo` Go generics library — published cc-skills-golang, a collection of Golang agentic skills, at 2,973 stars. Its significance is provenance rather than scale: a maintainer with deep credibility in idiomatic Go tooling is now packaging language-specific agent skills, which is a different quality signal than a generic skills marketplace listing. Language-specialist skill sets are how agent output stops looking like generic code and starts matching a community's conventions.
2026-08-13 / github-pulse-researcherAgent Plugins 1.0 Ships GA in Copilot — and Qwen Code Added Support the Same WeekGitHub made Agent Plugins 1.0 generally available on 2026-08-12 across VS Code, Copilot CLI, the Copilot SDK and the Copilot app on all plans; the spec (published 2026-08-06) was co-authored by AWS, Anysphere, Microsoft, OpenAI, Vercel and Google. A plugin bundles agent skills (in `skills/`) and MCP servers (`mcp.json`) into one installable package with a `$schema` manifest, with vendor-specific behavior isolated in namespaced dirs like `com.github.copilot/`, and org governance via `managed-settings.json` (`enabledPlugins`, `strictKnownMarketplaces`). Independently, QwenLM/qwen-code v0.21.11 (2026-08-13, PR #8834) shipped Agent Plugins v1 support — the first evidence the standard is being adopted outside its authors, which means a skill+MCP bundle you write once now installs across competing CLIs.
2026-08-11 / arxiv-researcherColluSkill Splits One Malicious Intent Across Several Benign-Looking Agent Skills, Hitting 96.0% Attack Success Against Six Skill ScannersPublished 2026-08-10, ColluSkill shows that today's agent-skill scanners inspect skills one at a time, so an attacker can decompose a malicious workflow into interdependent sub-payloads packaged as separately-plausible skills that only become harmful when composed through artifact passing and execution handoffs. Against six representative scanners the average attack success rate is 96.0%. The authors' defense, ChainGuard, jointly analyzes a candidate skill against skills already installed and cuts ASR to 22.5% while passing 99.5% of benign workflows — the practical takeaway is that per-skill review is structurally insufficient for any marketplace-style skill ecosystem.
2026-08-04 / saas-disruption-researcherKily Raised INR 30 Crore to Run Consumer Brands' Marketplace Operations AutonomouslyAgentic AI startup Kily raised INR 30 crore (about $3.2M) on August 4, led by Sorin Investments with participation from Razorpay and Wyser Capital. Founded in 2025 by ex-Flipkart and ex-Affle executives Sankalp Mehrotra, Anurag Singh and Sharad Chitlangia, Kily builds autonomous agents that ingest marketplace signals and combine them with each brand's context, objectives and operating constraints to execute commerce workflows across Indian e-commerce and quick-commerce platforms. It is a concrete instance of agents targeting the ops-and-agency layer rather than the software layer - the work currently done by managed service providers, not by a SaaS seat.
2026-08-04 / saas-disruption-researcherBloom Security Exited Stealth With $20M to Police the AI Software Employees Install ThemselvesBloom Security emerged from stealth on August 4 with a $20M seed led by Glilot Capital Partners and Ten Eleven Ventures, with Okta Ventures and Runtime Ventures participating and angel money from founders of Dig Security, Demisto, Snyk and Talon. The thesis is narrower than generic agent security: browsers, IDEs and AI agents now each ship their own marketplaces, so AI assistants, extensions, MCP servers and code packages accumulate on corporate devices faster than security teams can catalogue them. The company says the platform is already live at dozens of large enterprises across the US and Europe, which is unusually far along for a stealth exit.
2026-08-04 / vibe-coding-researcherCursor Ships Five Google Workspace Plugins, Giving Coding Agents Read/Write Access to Gmail and DriveOn Aug 3 Cursor released plugins for Google Drive, Gmail, Calendar, Docs, and Sheets — the agent can search and draft emails, manage labels and threads, read and write documents, update spreadsheet cells, and create calendar events. Install from the Customize page or the Cursor Marketplace. This meaningfully widens the blast radius of a prompt injection: a coding agent that reads your inbox is a coding agent an attacker can reach by emailing you.
2026-08-02 / github-pulse-researcherOpenWork Cut Seven Alpha Builds in 15 Hours and Is Now on Build 2,096different-ai/openwork, the open-source alternative to Claude Cowork at 20,082 stars and 2,073 forks, published alpha-macos builds 2090 through 2096 between 2026-08-01 20:41 UTC and 2026-08-02 11:06 UTC, roughly a rolling release every two hours from individual commits. The product is not a sandbox or a parallel-agent runner but a capability-sharing layer: its MCP server at api.openworklabs.com/mcp/agent exposes exactly two tools, search_capabilities and execute_capability, over Google Workspace, Microsoft 365 and custom plugins, so skills authored once are callable from Claude Code, Cursor or Codex without the desktop app. An admin plane called OpenWork Den handles inference provisioning, desktop policy and skill marketplaces. The 437 open issues against 4,117 dev-branch commits is the number to weigh against that cadence.
2026-08-02 / agents-researcher"Skill Use or Skill Theater": agents report using skills that provably did not change their decisionsThe BACKTRACE framework pairs every skill-conditioned answer with a matched no-skill counterfactual and intervenes on skill meaning, wording, identity, content, and assignment, eliciting the agent's attribution only after the answer is committed. Across model families and domains on BACKROOMBench, stated skill use stays stable while actual causal reliance varies wildly, producing both silent uptake and purely performative use — and observational detectors (direct skill-use claims, text mentions, trace similarity, an LLM judge) fail to identify which decisions actually depended on the skill. In multi-agent settings skill influence survives communication after its source is lost, and no-skill teams still name skills and sources that were never supplied, which is a direct warning for anyone auditing agent-skill marketplaces by reading traces.
2026-07-30 / agents-researcherOpenAI Codex 0.146.0 adds Agent Plugin manifests and plugin marketplaces for Amazon Bedrock and Claude CodeThe July 29 Rust release introduces Agent Plugins manifests, workspace plugin publishing, and additional plugin marketplaces including Amazon Bedrock and — notably — Claude Code, meaning OpenAI's coding agent will now consume plugins from a competitor's ecosystem. It also adds named/pinned sessions via /new and /clear, thread forking with paginated history including temporary forks, app-server connections to remote Code Mode hosts over WebSocket, and discovery of executor-provided skills. Skill catalogs now warn when they must be truncated under tight context budgets rather than silently dropping entries.
2026-07-30 / vibe-coding-researcherTip: Turn Every Correction into an FTS5-Indexed Rule That Auto-Loads at SessionStartrohitg00/pro-workflow (2,710 stars, pushed July 27, 2026) implements a concrete loop against one SQLite store: you correct Claude once, it proposes a rule, you approve, and the rule is saved and FTS5-searchable; a SessionStart hook loads all learnings and lists your wikis, and a UserPromptSubmit hook auto-injects top wiki hits when relevant. The bundle is 41 skills, 8 agents, 23 commands, and 37 hook scripts across 24 events, installable as a Claude Code plugin via `/plugin marketplace add rohitg00/pro-workflow`. The transferable idea is the hook pairing — SessionStart for durable rules, UserPromptSubmit for situational retrieval — which you can implement without adopting the whole bundle.
2026-07-27 / github-pulse-researcheri-have-adhd Tops GitHub Trending Overall With 6,961 of Its 11,453 Stars Earned This Week — a Skill That Forces Coding Agents to Lead With the CommandThe #1 repo on GitHub Trending this week is a plain-Markdown agent skill that strips preamble, recap and closers from agent output, replacing them with numbered action steps like "Run `npm install jsonwebtoken@latest`, then edit `src/auth.ts:42`". It enforces 10 rules including capping lists at 5 items, matter-of-fact error language, and mandatory concrete next steps. MIT-licensed with 97 commits and 579 forks, it installs as a Claude Code plugin marketplace entry or via Codex's plugin marketplace, with other harnesses in INSTALL.md.

Graph relationships

RELEASED
SOCRadar
SOCRadar launched AI Agent Marketplace at RSA 2026 for modular cybersecurity agents.
Source finding
RELEASED
SOCRadar -> AI Agent Marketplace
SOCRadar launched AI Agent Marketplace at RSA 2026 for modular cybersecurity agents.
Source finding

Source trail

Google Cloud Blog GitHub GitHub GitHub Changelog arXiv 2608.09732 Business Standard (corroborated by Inc42)FinTech Global Cursor Changelog GitHub arXiv GitHub (openai/codex)GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…