sgit newsroom v0.1.29 · snapshot 2026-09-24

Reading room · what-can-it-do.games.sgit.ai

On this page

Reading room / what-can-it-do.games.sgit.ai · raw text · live ↗

From what-can-it-do.games.sgit.ai, the page as fetched on 2026-09-25 · open the live page ↗Everything on this sheet is the source site's own text; the newsroom's chrome is outside it.

The map — what each agent can reach

23 capabilities across 9 products and setups: what each one can reach, what stands in the way, and how sure anyone is. Generated from the data pack; change it with a pull request.

Source: https://what-can-it-do.games.sgit.ai/map/index.html↗ · site v0.8.0 · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a .md twin; internal links below point at them.


Play↗ / The map

What each agent can reach

Every question the game asks is a cell in this table. 23 capabilities — a verb, an object and how far it reaches — against 9 products in the setups people actually run them in. The fill says how open the door is; the glyph says the same thing without the colour.

9** products × setups · **23 capabilities · 13 of them cannot be undone · 21/99 rows measured, not derived

Legend: ● none — nothing in the way · ◉ expectation — a rule in prose, enforced by nobody · ◐ setting — a switch the agent's own account can flip · ○ boundary — enforced above the grant, out of the agent's reach · · not in this grant

CapabilityUndoClaude Code (web container)Claude Code (local · confirm off)Claude Code (local · confirm on)Claude Desktop (local tools)Claude.ai (connectors on)Browser extension (all sites)Scheduled job (service account)GitHub Actions (hosted runner)ChatGPT (no connectors)
filesystem — files and directories
Read the project it is working on read.file.projectyes●●●●●··●●
Change the project it is working on write.file.projectwith-effort●●●●···●·
Read any file the account can reach read.file.hostno●●●◐○·●●·
Change any file the account can reach write.file.hostwith-effort●●●◐··●●·
Delete files anywhere the account can reach delete.file.hostno●●●····●·
Read a retained record: shell history, past sessions read.record.historyno●●●●·····
process — programs and their execution
Run programs as the account execute.process.hostwith-effort●●◐◐··●●·
Run programs inside its own sandbox only execute.process.selfyes·········
network — endpoints and hosts
Reach a permitted list of hosts send.endpoint.allowedno○········
Reach any host on the internet send.endpoint.worldno·●●●·●●●·
identity — credentials and who the agent can act as
Read credentials stored where it runs read.credential.hostno●●●●·····
Act in accounts with the credentials it holds authenticate-as.credential.tenantno○●●●○◐●··
Change its own permission settings grant.credential.selfyes·◐◐◐·····
communication — messages to people
Send a message to anyone send.message.worldno·········
Read mail or chat it is connected to read.message.tenantno····○····
code — repositories and what lands in them
Commit to the repository it was pointed at write.repository.projectwith-effort●●●····○·
Push to a code host (any branch it can reach) write.repository.tenantwith-effort◐◉◉·○····
Sign commits with the key it holds authenticate-as.credential.signingno●●●······
Publish packages, images or pages under the name it holds create.record.worldno·●●······
money — budgets and spend
Spend money or tokens against an account it holds write.budget.tenantno······●··
schedule — things that outlive the turn
Create something that outlives the turn where it runs (a cron, a service) create.schedule.hostyes○●●···●··
Create something that outlives the session, on the platform (a routine, a scheduled trigger, a new session) create.schedule.tenantyes◐········
browser — what a browser extension or automation can see and do in your browser
Read every page you visit read.record.browsingno·····●···

How to read it

How sure is any of this

Not very, in most places, and every row says so. Of 99 tool-capability rows across the set, 21 were measured — a probe run on an instance, dated — and the rest are derived from what a thing architecturally is, or documented by the vendor. A derived row is a claim until somebody runs the probes and contributes the file, and the honest reading of this table is mostly claims, structured so that each one can be replaced by a measurement.

That is what the pull request is for. How to contribute↗ — a row, a whole profile, a mandate, or a correction.

The rest of the map

Every product, one page each — Tool by tool: what it reaches, the control on the path, the evidence, what the profile says is out of reach and why, and the reductions that narrow it. The products↗

The capabilities — The 23 primitives the questions are built from — which products grant each, what narrows it, which questions ask about it, and which mandates want it. The capabilities↗

Mandates, and the deltas — What a reasonable person wanted, per setup — and the gap against what was actually granted, as a matrix: excess in one colour, shortfall in the other. The mandates↗ · The deltas↗

Above the ceiling — The 17 things no agent in this set can do, each naming the control outside the agent that stops it. The ceiling↗

The questions — 25 questions the games ask, with what each one is for — the seed of a question pack that could be customised. The questions↗

The data pack — All of it as JSON, at a stable URL with CORS, which is how the game reads it. Fork it, change it, point the game at yours. The pack↗ · Contribute↗


Site index for agents↗ · HTML version↗