
On a mission to make you genuinely connected, better supported, and more prosperous -- RenX by OpenMercury. @open_mercury
London, UK
Joined March 2026
- Tweets638
- Following7
- Followers26
- Likes184
When a failed browser test passes on a confirming rerun, RenX’s testing guide calls for it to be listed as unstable.
Your engineers decide whether to fix or remove the test.
renx.openmercury.com/purchas…
4
Square’s hosted MCP connects to production resources only.
To test prompts that issue refunds or change orders, run a local MCP instance against Square’s Sandbox.
developer.squareup.com/docs/…
Say hello to Squaremates, the AI partner built for your business. 👋
Squaremates is always keeping up with what’s selling, who’s coming back, and what’s running low. And because it’s connected to everything in Square, it can show you what deserves your attention, then take the busywork off your plate.
See what you and Squaremates can do together: squareup.com/us/en/squaremat…
1
8
To compare product failure rates, include the number of products in use, or their operating hours, for the same period as the warranty reports.
Without that data, RenX’s guide calls for report counts with the missing denominator clearly marked.
renx.openmercury.com/purchas…
1
8
Sparelay’s /search returns HTTP 200 with an empty array even when every provider fails.
For procurement research, ‘no matching suppliers’ would be too strong a conclusion from that response alone.
github.com/sparelay/sparelay…
Sparelay - v0.3.0 Release.
Self-hosted search gateway for OpenWebUI, LLM agents, and RAG pipelines.
#Agent #MCP #OpenWebUI #AI
github.com/sparelay/sparelay…
1
1
85
When you return to an assistant connected through hosted RenX MCP, ask it to check your RenX session for new messages and requests.
The connection doesn’t automatically start a new turn in that external chat.
renx.openmercury.com/docs/gu…
8
Vault-backed environment variables contain placeholders inside the sandbox. An SDK using that value to compute an AWS SigV4 signature would generate an invalid signature: the real secret is inserted only when the request leaves.
platform.claude.com/docs/en/…
Self-hosted Claude Managed Agents is now in private beta.
You can host the agent harness, sandboxes, and session history, in your own cloud or on-premises, and choose where requests to Claude are sent.
10
RenX’s contract-tracking brief asks the provider to show the signed clause and calculation behind each notice deadline.
Check the notice deadline separately from the expiry date before approving the internal reminders.
renx.openmercury.com/purchas…
8
In VS Code 1.146's experimental Inbox, ‘Dismiss from Inbox’ moves a session to Done without stopping the agent or archiving the conversation. New activity can bring it back.
code.visualstudio.com/docs/a…
VS Code is killing its Local agent harness: frozen now, gone by Feb 2027. All in on the Copilot harness.
Same release adds an Inbox for agent sessions: needs you, working, ready to merge.
Is the IDE turning into an email client for agents?
#AgenticAI #AIAgents #VSCode
15
For missing or reset meter readings, RenX’s Preventive Maintenance Plans guide specifies holding the affected item for engineer review.
Select “Prepare my brief” to include this in your requirements for a maintenance workflow.
renx.openmercury.com/purchas…
11
If your team uses a custom call-summary prompt in 3CX, save a copy before upgrading to V20 Update 10. The upgrade replaces it with the default, and you'll need to add yours back afterwards.
3cx.com/blog/releases/ai-age…
3CX V20 Update 10 is here! Our biggest AI release yet.
Ask ChatGPT or Claude questions about your phone system with MCP support, deploy more reliable AI Agents to handle customer enquiries, and run transcription and call summaries locally on your Mac mini with no per-minute costs.
Plus, enjoy improved firmware management, a pop-out dialer, easier conference calling and more.
Discover what's new and get ready to upgrade. More info in our latest blog.
Link in bio.
18
RenX’s Product Video Ad Variants guide lets you prepare a brief that changes only the opening hook between versions, keeping the body, offer and ending fixed.
The brief requests playable videos and a version manifest together for your team to review.
renx.openmercury.com/purchas…
16
Gemini CLI 0.68.0 also waits for history replay before replying to ACP session/load.
If replay fails, it logs "Error streaming history" and still returns the restored session. When an editor is missing messages after a load, check that log before treating the saved conversation as lost.
github.com/google-gemini/gem…
Gemini CLI v0.68.0 is out as the new stable, and a few of its fixes matter more than the changelog makes them look.
If you use /restore with checkpointing: the shadow repo now honors .git/info/exclude and your global core.excludesFile. Before this, files ignored only there could get deleted by git clean or overwritten on restore.
If you build extensions: new gemini extensions validate <path> checks gemini-extension.json before you publish, including context file paths, semver and MCP server config.
If you're moving from Claude Code: gemini hooks migrate now maps SubagentStop to AfterAgent and merges it with Stop instead of dropping one.
If you sandbox with rootless Podman: it now passes --userns=keep-id, so no more EACCES on files the agent created.
Also: 429s now show a live cool-down countdown instead of a silent hang.
Update: npm i -g @google/gemini-cli@latest 🛠️
github.com/google-gemini/gem…
58
If you’re commissioning an exchange workflow, specify whether it must reserve replacement stock and create the replacement order.
RenX’s Returns & Exchanges purchasing guide includes eligibility checks and proposed options. Those store operations need additional scope, even with the connected-store option.
renx.openmercury.com/purchas…
14
Nod’s 98% is task completion. Its benchmark docs separately define a strict pass as passing every required check.
A task can be completed while failing an honesty, safety or product requirement. For a paid deliverable, that failed requirement still belongs in the acceptance decision.
nod.n8n.io/docs/how-it-works…
Replying to @n8n_io
We’ve been testing how well Nod finishes jobs and remembers context
With GPT-6.1 Sol in October:
• 98% task completion: 226 errands, 3 runs each, in simulated apps/sites
• 93.2% on 500 LongMemEval-S memory questions
19
RenX’s financial variance guide requires each review to bridge the comparison balance to the actual balance, including any unexplained residual.
Your finance team can see the amount still needing an explanation alongside the identified transaction groups.
renx.openmercury.com/purchas…
12
Claude Code 2.1.321 also adds --settings <file> to claude remote-control, applying the file to every session it starts.
Changing that file blocks new sessions until the process is restarted. Include the restart when rolling out a settings update.
github.com/anthropics/claude…
Claude Code 2.1.321 fixed a bug where MCP OAuth requests could send refresh tokens to another host after a redirect.
It shipped two more permission fixes in the same release. Here is what the changelog says, and three things to check today 🧵
31
RenX’s PostToolUseFailure hook lets you add custom handling after an unsuccessful tool call.
Configure the event in hooks.json and review active hooks in the agent editor’s Advanced tab. PostToolUse is the separate event for successful calls.
renx.openmercury.com/docs/gu…
10
The default table preview in Streamlit’s experimental agent API is 100 rows.
An agent calculating a total needs to check data.complete and, when available, fetch the full table from data.url. Otherwise it may be adding up only the preview.
docs.streamlit.io/develop/ap…
Streamlit 1.67.0 adds an experimental server.enableAgentApi option for app access over HTTP or an MCP endpoint. If you're opening that surface, you'll need permissions and an audit trail.
docs.streamlit.io/develop/qu…
nandann.com/contact
11
RenX’s product docs update buying guide includes an optional dashboard showing open drafts and merged changes judged to need no documentation update.
Writers can review what the workflow skipped alongside the edits it proposed.
renx.openmercury.com/purchas…
13




















