terminal Mission Computer // collaborative compute

Put your AI to work on one useful task for Britain.

Future Union turns scattered AI effort into reviewable public work: pick a bounded problem, spend useful AI tokens locally, then return a packet Britain can use. Copy one safe command, open the Ratatui Control Room, point Pi, Codex, or Claude Code at one bounded problem, and return a packet a human can review. If the right task is missing, propose one through Intake.

play_arrow Open Planning blockers sprint add_circle Propose action
assignment

What is this?

A mission is a bounded public task: gather sources, test a claim, map a blocker, or improve a Future Union surface.

terminal

How do I run one?

Copy the universal runner command once. It installs fucr, opens the Control Room, and lets you choose a mission before any agent starts.

fact_check

What happens next?

You return a reviewable packet. A human decides whether it becomes a receipt, story, Growth Index input, toolkit item, or hold note.

Collaborative computer

Useful tokens in. Reviewable packets out.

This is not a black-box volunteer form. It is a local compute lane for people who want their AI subscription or local machine to move a public mission forward without handing Future Union their private workspace.

Many people can run the same brief

Every contributor gets a separate local workspace, Git history, source log, autoresearch tape, packet, and credit note. Parallel runs do not overwrite each other.

Progress is useful before it is perfect

The runner records active claims, checkpoints, sources, packet score, and finish state. Reviewers can merge the best parts, discard weak runs, and credit the pieces that moved the mission forward.

Radicle is the public packet lane

A finished packet can become a signed Radicle experiment after preview and explicit confirmation. Raw chats, local notes, private contact details, and unfinished work are not published by default.

Matrix is coordination, not proof

Action rooms are for blockers, source leads, and reviewer questions. The proof remains the packet, sources, local Git commits, and signed experiment record.

Run a mission locally

Create a real mission workspace, not just a pasted prompt.

The runner creates a mission-specific folder with the selected brief, prompt, packet template, source log, local skill, helper command, wider Future Union context, and setup checks for local agent tools. If two people start the same brief, they produce separate packets; human review decides what gets merged, credited, or discarded.

Action brief

Inputs, acceptance criteria, submit route, and why this run matters.

Generated README

The exact folder guide the runner pulls into your workspace.

Local skill

Tiny skill file that points your agent at the local Future Union context and helper commands.

Packet template

The reviewer-facing Markdown output if you want to inspect the shape first.

Local preview

ratatui-local-control-room

Open the Control Room 

curl -fsS https://www.futureunion.org.uk/agent-pack/fu-runner.sh -o fucr && FU_BASE_URL=https://www.futureunion.org.uk sh fucr

Inspect first, then run 

curl -fsS https://www.futureunion.org.uk/agent-pack/fu-runner.sh -o fucr

Useful flags

FUCR_BASE_URL=http://127.0.0.1:4392 FU_BASE_URL=https://www.futureunion.org.uk FUCR_CACHE_DIR=~/.cache/future-union/fucr FUCR_INSTALL=0 FU_AGENT=auto|pi|codex|claude|none FU_AUTORESEARCH=1|0 FU_START=ask|1|0 FU_PICK=0 FU_TUI=ratatui|lite|textual|node|0 FU_TUI_VENV=.future-union/tui-venv FU_ENABLE_PI_CC=1 FU_YES=1 FU_BOOTSTRAP=0 FU_DRY_RUN=1 FU_CONTRIBUTOR_NAME="Your name" FU_CONTRIBUTOR_HANDLE="@handle" FU_CONTACT="email@example.com" FU_ANONYMOUS=yes FU_IDEA_FILE=idea.json FU_SHARE_DIR=.future-union/share FU_MATRIX_HOMESERVER=https://matrix.example FU_MATRIX_USER=@you:matrix.example FU_MATRIX_ACCESS_TOKEN=...

Runner output

README.md AGENTS.md RADICLE_NETWORK.md MISSION.md PROMPT.md PACKET.md ACTION_PROPOSAL.md SOURCES.csv CONTRIBUTION.md CREDIT.md CHECKPOINT.md RADICLE.md EXPERIMENT.md SHARE_MANIFEST.md .future-union/run-status.json .future-union/run-progress.jsonl autoresearch.md autoresearch.sh autoresearch.jsonl INTAKE_IDEA.md SUBMIT.md bin/fu-mission .future-union/rooms/[action].md .future-union/state.json .community-computer/session.json .community-computer/pending-experiment.json context/runner.json context/future-union-context.json context/action.json context/mission.json context/safety-rules.md context/review-policy.md .agents/skills/future-union/SKILL.md

The useful output is one reviewable packet plus sources and a credit note: enough for a human to verify quickly, attribute useful work, and avoid excavating a whole chat transcript.

Optional Radicle lane

Preview the signed experiment before any public handoff. The helper shows commits, packet score, command shape, and asks again before publishing to the Future Union Radicle lane. Community Computer/pi-cc publishing stays opt-in.

./bin/fu-mission experiment-preview ./bin/fu-mission experiment-status ./bin/fu-mission experiment-publish
How it works

Machine can grind. Human owns the claim.

view_kanban Open action board

Step 01

Pick one action brief with a concrete output and a clear time box.

Step 02

Run the safe local runner command to check local dependencies and create a mission workspace with context, prompt, packet, and sources files.

Step 03

Use the Ratatui Control Room to pick work, see community activity, open the local action room, sync generated workspaces, and start Pi, Codex, or Claude Code in the generated folder.

Step 04

Return one reviewable Markdown packet with facts, caveats, sources, compute disclosure, and recommended FU action.

Mission stack

Choose the work, then keep it bounded.

Open 20-30 minutes

Action Intake

The front door for capturing supporter ideas before they become ranked Future Union action proposals.

open_in_new Open mission page
Output
Action title; problem; target user; proposed output; source trail; time box; expected impact; risks; rubric score; recommended queue; first reviewer question.
Live cards
1

Human owns: Submit raw ideas quickly, then judge fit, safety, timing, and whether a verified proposal should enter Open now, Needs backing, or hold.

First runnable card: Add an idea
Designing 2 hours

Agents for Britain

A public-facing way to fund and steer narrowly scoped agent work on Britain’s real bottlenecks.

open_in_new Open mission page
Output
Title; one-line claim; why it matters; source links; evidence table; caveats; recommended FU action; reviewer notes.
Live cards
0

Human owns: Set priorities, define standards, review outputs, and decide what gets published.

Building Ongoing

Movement Infrastructure

The boring but decisive machinery: site, list, imprint, submission routes, and public operating surfaces.

open_in_new Open mission page
Output
Surface; expected user action; pass/fail; evidence; broken states; exact fix; launch risk.
Live cards
1

Human owns: Own public launch readiness, privacy posture, imprint details, and final publication decisions.

First runnable card: Owned list front door
Scoping 60 minutes weekly

Britain Build Radar

A recurring scan of builders, councils, campaigns, and policy shops actually pushing the country forward.

open_in_new Open mission page
Output
Story; place; source links; protagonist; blocker/enabler; why it matters; suggested format; confidence.
Live cards
1

Human owns: Choose the angle, add judgment, and turn signal into political pressure.

First runnable card: Britain Build Radar
Pilot 90-180 minutes

Receipts Desk

A document-led accountability lane for finding where process, procurement, or permissions are obviously broken.

open_in_new Open mission page
Output
Receipt ID; value; claim; context; source label; source URL; source date; checked date; caveats; where FU should use it.
Live cards
2

Human owns: Pressure-test the claim, name the trade-off, and decide when the evidence is ready for public use.

First runnable card: Planning blockers sprint
Method 2 hours per source pass

Growth Index

The flagship data product: rank places by whether they enable growth, then turn the score into pressure, stories, and local action.

open_in_new Open mission page
Output
Input; candidate source; coverage; refresh cadence; known caveat; scoring implication; recommended next step.
Live cards
1

Human owns: Choose the weighting, approve caveats, manage public claims, and decide when rankings are ready to publish.

First runnable card: Growth Index v1
Open 30-60 minutes

Local Proof Map

A local listening and issue-mapping mission so FU chooses places and pressure points from evidence, not vibes.

open_in_new Open mission page
Output
Place; issue cluster; evidence available; people affected; confidence; follow-up question; possible FU action.
Live cards
1

Human owns: Run conversations, judge credibility, protect privacy, and choose whether an issue becomes a local campaign.

First runnable card: Issue map intake
Technical pack Prompt, feeds, templates, rules. Open this when you want the machine-readable pack. Human orientation stays first.

Agent mission page

Human-readable briefing surface with local runner preview, mission prompts, QA rules, and pack links.

Action feed JSON

Machine-readable public action board with detail URLs and submission routes.

Mission feed JSON

Machine-readable standing missions for recurring agent work.

Intake queue JSON

Redacted idea queue: received/review state only, not raw notes or contact details.

Mission Computer JSON

Protocol, safety rules, submission routes, and contribution schema in one feed.

Contribution ledger JSON

Reviewed and seed contribution records, never raw unreviewed submissions.

Agent pack

Drop-in AGENTS.md for a local Pi, Codex, or Claude Code workspace.

Mission runbook

Step-by-step guide for humans bringing their own agents and tokens.

Runner script

Readable shell script that creates a mission workspace with context, prompt, skill, and helper command.

Ratatui Control Room source

Source-distributed Rust/Ratatui mission cockpit used by the default runner.

Native fallback Control Room TUI

Python fallback cockpit when Rust/cargo is unavailable.

Textual Control Room TUI

Optional richer TUI for people who explicitly opt into Textual with `FU_TUI=textual`.

Node Control Room fallback

Last-resort terminal picker when Python is unavailable.

Contribution claims

Explicit opt-in endpoint for showing that someone has started a local mission run.

Source research skill

Reusable skill instructions for source-led FU research packets.

Action ranker skill

Reusable skill instructions for ranking proposed actions against Future Union values.

Review system

Open the work. Gate the claims.

Raw mission output is not politics by accident. Future Union only turns packets into public claims, campaign assets, or datasets after human review.

Reviewable packet

The standard output: Markdown packet, source links, generated artifacts, caveats, contribution credit, and reviewer checklist.

Local skill

The runner installs a tiny Future Union skill plus local context files so your agent can work from the wider system, not just one pasted prompt.

Email packet

V0 route: send the Markdown packet and credit note to hello@futureunion.org.uk with the exact subject line on the mission brief.

Radicle experiment

Primary route: preview the signed experiment record, then publish a packet-score contribution after explicit confirmation.

Ranked intake

Action ideas enter as proposal packets, get scored against Future Union values, and only become public board cards after human review.