From Pushary
The Pushary newsroom
What shipped. What changed. How to keep control of your AI agents. Product updates and practical guides from the people building Pushary.
PusharyYour agent asks.
You decide.
The decision layer for AI agents.
The Pushary Manifesto
Agents will do more of the work. People will own the decisions. We are building the layer that makes both possible.
Browse articles
Latest articles · 56
Claude Code approval fatigue: what the numbers say and what to change
People click yes on almost every Claude Code prompt, and they catch less the longer a session runs. Fewer, better questions fix that. More prompts do not.
Codex stuck on Awaiting approval: what it means and how to fix it
Most stuck chats are a request you have not seen. Some are a bug where the request cannot be answered. How to tell them apart and get Codex moving again.
How long AI agents wait for people: the 2026 data
Agents work for minutes, then stop for a person. Much of the wait is a question nobody has seen yet. Every number here is public and dated.
How to roll out Codex managed settings to a team
A Codex requirements.toml example, where to deploy it, and four checks to run with a teammate before rollout.
Tools that let a team approve an AI coding agent's risky steps, compared
Claude Code, Codex and Cursor only ask the person at the keyboard. These tools can ask a lead instead. Which agents each one covers, where approvers answer, what it records and what five people pay.
Can you route a Claude Code permission prompt to your team lead before it runs?
Not with Claude Code alone. A permission prompt goes to whoever is at the terminal. Here is what a hook can do, what managed settings can't, and how a routing rule sends the step to a lead.
An AI agent broke something. How to trace what ran and whose yes let it
Git names the machine, the session names the command, and only a recorded approval names who said yes. Where to look, in order, and what each place can and can't tell you.
Claude Code managed settings for a team: a working file, and what it can't do
The one Claude Code file your team can't override: where it goes, a version you can copy, how its rules are read, and what it still can't do for a team.
Who approved it? What Claude Code, Codex, Cursor and Copilot team plans record
Each tool lets the person running it say yes. None of them, on any team plan, documents asking a second person first. Here is what each one does record.
How Pushary for Teams decides who an agent asks
Each agent asks the person running it, and a rule can send a step you pick to a lead. How that works, what the team sees afterwards, and the gaps we have not closed.
Claude Code approvals: answer from your terminal or phone
Back at your desk? Answer there. Away from it? Use your phone. Supported Claude Code approvals now stay open in both places at once.
Verify approval webhooks before your agent resumes
A callback should wake your worker, not authorize an action. Verify its signature, save it once, then check the decision against your saved operation.
The agent lost the response. Did the action happen?
Approval, a committed effect and workflow status answer different questions. Two runnable examples show what survives a lost response.
What shipped: more Codex questions reach your phone
Answer more of the questions that used to keep you at your desk. A short guide to what changed in Pushary CLI 1.9.57 through 1.9.60.
Your agent retried. Your customer should not have to approve twice.
Give the operation a stable identity, save the decision, and resume from the answer. A practical example of the part that comes after sending an approval request.
Your customer should know exactly what they are approving.
An approval button is the last step. First, name the person, show the action and make sure the answer stays attached to the work.
I got tired of rebuilding human approvals. So we built Partner APIs.
A new agent should not mean another approval system. Here is the problem behind Pushary Partner APIs, and the decisions we made while building them.
Pushary Isle: your agents' questions, in your Mac's notch
Your agent stopped to ask you something in a tab you weren't looking at. Isle puts that question in your notch, and one shortcut answers it.
Approve refunds over $500: the four things that decide whether a threshold holds
Picking the number is the easy part. Rule ordering, a missing field, a numeric string, and a retry are the four places a threshold leaks.
How to add human-in-the-loop to a Vercel AI SDK agent
The AI SDK gives your agent tools. One of those tools can be a real person. Here is the whole integration, from install to fail-closed decision.
Build vs buy: the real cost of a human-in-the-loop approval loop
The prototype takes a weekend. The system that will not silently approve a refund takes a quarter. An honest itemization, including when building is right.
Fail-closed by default: why agent approvals should block instead of fire and forget
The failure case for approvals is not exotic. It is a phone in a pocket. Fail-closed design decides in advance that the ordinary case means stop.
LangGraph human-in-the-loop: interrupt, approve, and resume from a phone
LangGraph pauses the graph and persists state. It does not notify anyone. Here are the two patterns that finish the job, with the footguns written down.
The gap between notifying a human and getting a decision back
Both start with a push message. The difference is everything after the tap, and it is the difference between an agent that announced and an agent that asked.
How to approve Claude Cowork tasks from your phone
Paste one connector link, add standing instructions, and Cowork's questions arrive as push notifications you answer from the lock screen.
An audit trail for Claude Cowork: what did your AI coworker decide while you were out?
Delegation without a record is hope. Keep every Cowork question and answer in an immutable, exportable trail, next to every other agent you run.
Claude Cowork stuck waiting for permission? Stop babysitting it
You hand Cowork a task and walk into a meeting. It hits a question at step two and waits. Here is how to get those questions on your phone instead.
Stop making your agent ask permission for everything. Or nothing.
Reads run free. The irreversible slice is small, and it is the only thing that should be allowed to interrupt you.
Pushary, featured on AI Plaza
AI Plaza interviewed our founder about the gap Pushary closes, the moment an agent finishes and nobody is watching.
What an AI agent audit log should capture for teams and compliance
The fields a coding-agent audit record needs to be worth keeping, and the honest line on what GDPR-aligned and self-assessed actually means.
What shipped: more agents, and agents that ask before risky steps
Native hooks for Gemini CLI and Codex, plus a setup change that makes both agents ask before risky steps instead of only being gated.
Connect Claude Desktop with one URL: notify and ask, nothing else
Paste one URL to let Claude Desktop reach your phone. Connector-scoped revocable key, notify and ask only, no enforced gating.
How permission policies learn your defaults
Match rules on what the agent is about to run, not just the tool name, and let your real approve and deny history write the rules for you.
Who is accountable when an AI agent makes a mistake?
An agent has no accountability of its own. The human who ran it owns the outcome, which is why a record of who approved what matters.
How to run multiple AI agents at once without losing track
The workflow and the board for running concurrent agent sessions without losing track of which one needs you.
AI agent control glossary: HITL, MCP, permission gates, kill switch, audit trail
Each term gets a one-line definition first, then context and a short example. Built so you can lift any single entry.
Pushary for Cursor: approve the agent's shell commands from your phone
A Cursor plugin that gates shell commands behind a phone approval, fails closed, and survives the Windows stdin double-fire bug.
What shipped: reliable delivery across browser, mobile, and the iOS home screen
We fixed silent push failures on agent sites, put Web Push, Mobile, and Slack in one Connections tab, and gave iOS home-screen users a pending-questions inbox.
Vibe coding safety: letting AI run code without babysitting it
The real risk in vibe coding is the commands an agent runs, not just the code it writes. Scanners miss that. Human approval on actions closes it.
How to run an AI agent overnight and stay in control from your phone
Let the agent work while you sleep, but keep the gates. Risky steps route to your phone for a yes or no, and a kill switch ends a bad run.
The Pushary mobile app: answer approvals and send commands from your phone
The phone is now two-way: approve an agent action from the lock screen, then queue an instruction back to the running session.
How to stop an AI agent from running up your token bill
A max-budget flag warns you. A budget that kills the session and an approval gate stop the bill from growing while you are away.
Deny with a reason, and see the risk on every approval
Denying an agent action now sends a reason back, and approval cards flag the asks that carry real risk.
One board for every running agent
Delivery modes, presence-aware routing, and a Fleet board that groups concurrent agent sessions into lifecycle columns.
What --dangerously-skip-permissions does, and a safer way to run unattended
Let routine work continue. Choose which commands need your answer, and know what an approval hook cannot protect.
What permissions should an AI agent have? A starter least-privilege policy
Reads auto-approve, writes and pushes and spends ask, secrets and force-pushes deny. A concrete starter policy you can copy.
Stop runaway spend: kill an agent session at a daily budget
Cost guard reads an agent's spend from its transcript and kills the session when it passes a daily budget you set.
What shipped: approve from Slack, REST, n8n, or Zapier
Four new ways to put a human in the loop: Slack Block Kit, a REST ask endpoint, an n8n node, and a Zapier integration.
See exactly what your agents did: receipts, a what-changed view, and exports
Every agent session gets a receipt: structured tool actions, what changed, who answered where, and an export when you need it.
The four levels of AI agent oversight: notify, approve, policy, audit
Oversight comes in four levels. Notify is parity now. Approve catches the moment. Policy decides without you, including on the amount. Audit proves what happened.
Claude Code hooks explained: PreToolUse, PostToolUse, and Stop
All 33 hook events in one table, the JSON each one receives, the exact output fields, and how PreToolUse blocks a tool call before it runs.
Is it safe to let an AI agent run unattended?
It depends on what you put around it. Unattended is safe when reads run free, risky actions gate to your phone, and a kill switch and audit trail are in place.
Allowlist vs denylist for AI agent permissions, and why prefix denylists leak
A denylist tries to name every dangerous command and always misses one. Pair a read-only safe floor with an allowlist instead.
What is MCP, and how agents use it to notify you and ask for approval
The Model Context Protocol lets agents call external tools. Two of them can ping you and ask for a yes or no before the agent acts.
Human in the loop for AI agents, explained
The agent works on its own and stops only for the few decisions a person should see. Here is what that means and where the checkpoints sit.
Can an AI agent delete your files or drop your database?
Yes, an agent with shell or database access can delete your files, and it has happened in production. Here is how to gate it.
Let your agents work. Stay in control.
Get the questions that need you on your phone or Mac. Approve, reject, or answer so your agent knows what to do next.
Press and contact
Pushary is built by RalphNex OÜ. For interviews, product questions, or corrections, reach us through our business contact.
Contact PusharyFollow what ships
Release notes record product changes. Guides explain how the product works. Articles show their author and publication date, with a separate date for updates.