Repository navigation
EN Contribute Governance
Français · Type: explanation · Audience: contributor · Context: engine repository · Status: ready · Applies from: 4.4.0
Who decides in quest'n'click, how a decision is recorded, and what a person must do that no assistant or test can do for them. The project has one maintainer, @wanoo; most of its code is written by AI assistants and reviewed by gates, by a second reading and by that maintainer.
One person decides, in writing. The maintainer arbitrates every disagreement (D3). A decision is a dated line in
docs/dev/DECISIONS.md, written by the assistant who received it and never without a
human decision, with the maintainer's own words quoted, often in French with the English beside them. The exchange that
leads to it lives in docs/dev/LOG.md, committed, in English (D3). A decision that would look surprising without its
context also gets an architecture decision record in docs/dev/adr/ (Decisions and ADRs).
Everything in docs/dev/ is public development documentation: nothing private is written there.
Evidence beats opinion. When two readings disagree, the one with a measurement, a failing test or a reproduction wins by default; the maintainer can still overrule. A review that ran no command is an opinion and says so. An assistant that could not run a check (no browser, no Python) says so rather than skipping it silently.
A second context reads before merging. A change of any weight is read again by a second, independent context (in practice another AI session or sub-agent that did not write it), which reproduces before it judges, then by the maintainer. The LOG records each "second reading" and what it found.
People check what machines cannot, and say it. Seven, then more, checks need a person or a real device: a screen
reader, Safari offline, a real phone, playtesters, recorded voices, listening, a signed tag. They are reported, not
blocking (D12): a release ships on its automated gates and its notes list each pass not done. Each pass a person
makes is an open GitHub issue labelled field-pass; the maintainer closes it once the test is made and its report
checks (Field passes). A surface is called stable only with its machine proof and its
pass passed (D18, applied by D38 to 4.2.0's reduced stable scope).
A signature needs a yes each time. Release tags are signed with the maintainer's SSH key since 4.2 (D30 put it off
until then). The trust root, .github/allowed_signers, names the maintainer's software key id_ed25519, which lives on
the machine the assistant works on: a signature proves the key was used, not that the maintainer approved (D37).
So approval is procedural: each signed tag needs the maintainer's own explicit yes, recorded where it was given, and an
assistant never writes an approved carry attestation on its own. « tu peux signer la 4.5 » ("you may sign 4.5") is the yes for v4.5.0 only
(D50). See Releasing.
| What | Reviewed by |
|---|---|
| Every pull request | The CI gates (pr-gate), see CI explained
|
tests/quality-baseline.json, tests/api-surface.json, docs/dev/mutants.json, vite.config.ts, .github/
|
The maintainer, through .github/CODEOWNERS (the frozen behaviour, the public surface, the named mutation survivors, the coverage floors, the workflows) |
| A plan, a release, a non-trivial branch | A second context, then the maintainer |
| A field pass | The person who made it, then the maintainer, who closes its issue |
| A translation pair or a claim's scope in the wiki | Recorded in wiki/translations.json / wiki/reviews.json with who read it, a person or an AI (Writing documentation) |
docs/dev/CHARTER.md set the rules when v3 was co-developed by two AI assistants, Claude
and Codex, with the maintainer arbitrating. Its rules, in short:
-
Roles: two assistants, same rules; the human arbitrates, every disagreement ends as a line in
DECISIONS.md. -
One ball: the last LOG entry ends with
→ next: Claude|Codex|human; the side without the ball does not touch the tree. -
Branches:
mainis the release branch since v3.0.0 (D7); since 3.7.1 a branch is named by its kind (feature/,fix/,docs/,test/,refactor/,release/). - A proposal states its goal, what changed, what it breaks and the migration, its measures, the commands run with their output, and what it does not do.
-
A review reproduces before it judges, then ranks its findings
blocker/should/nit, each with evidence. - Evidence beats opinion.
-
Truth commands (
validate,solve,test,build, the e2e on a production build…) are recorded in the entry. - Identity in git: a trailer naming the agent; commit messages say what changed for the player or the author.
-
Public repository: no private names, no private content;
npm run auditbefore every commit. - Scope of v3: a break ships with its migration; no Phaser, no second dialogue format, no real-time scripting.
- Docs are part of the change.
- Silence rule: a check that could not be run is said.
Limit ·
charter-datedThe charter's wording is dated: it describes the v3 period with two assistants and a "one ball" exchange. Since D9 (4 October 2026) one assistant implements one branch at a time and merges it on green CI, and recent commits carry aCo-Authored-Byline rather than the charter'sAgent:trailer. Its principles (D3, evidence, the silence rule, public content) still hold; the maintainer has not rewritten it.
-
Blocking human gates. Rejected by D12: a release would wait on a tester's calendar. Instead every pass not done
is said in the release notes, and an open
field-passissue is a pass not made. -
A hardware key as trust root. Preferred, but not at hand when 4.1.20 shipped; the software key was accepted with
its limit said (D37). A rotation to
id_ed25519_skis its own pull request touching onlyallowed_signers.
- D3 (arbitration, the exchange in English), D9 (cadence), D12 (passes reported), D18 and D38 (what "stable" needs),
D30 and D37 (signatures), D50 (the documentation):
docs/dev/DECISIONS.md. - ADR 0024: a pass is bound to its surface, and carried only by a signed attestation.
Generated publication · publication générée — current documentation · documentation courante: quest'n'click 4.4.0 r1, documentation revision · révision de la documentation 8dc717c69acb62c4742b69c54a7e7dfb620f75d8.
Earlier versions and permalinks · versions précédentes et liens permanents: https://wanoo.github.io/quest-n-click/docs/
An edit made in this wiki is overwritten at the next publication · une modification faite dans ce wiki est écrasée à la publication suivante.
Edit the source · Modifier la source · Report an error · Signaler une erreur
Index (6)
Start (6)
Play (9)
Install (17)
- Requirements
- Install from the release archive
- Install from npm, once published
- Install from the repository to work on the engine
- doctor: checking your machine
- Image toolchain
- Audio toolchain: ffmpeg, fluidsynth, Furnace, uv
- Upgrade
- Upgrade notes: 4.4
- Upgrade notes: 4.3
- Upgrade notes: 4.2
- Upgrade notes: 4.1
- Upgrade notes: v3
- Upgrade notes: v2 to v3
- Coming from web-scumm
- Package reference
- Install troubleshooting
Create a game (55)
- Your first game
- Layout of a created project and of the engine repository
- The working loop
- Content principles and the identity contract
- Rooms
- Geometry: placement, walk zones, masks, layers, camera
- Something is visible but unreachable
- Reactions
- Game commands
- Conditions
- Flags, variables and events: which to use
- Custom code
- Characters
- Items and inventory
- World map
- Dialogues
- Scripts and events
- Moving characters: patrols and walks
- Exits
- Chapters and objectives
- State and saves: persistent or temporary, renaming an id, migrations
- Staging: wide rooms and animations
- The stage: lights, particles, floors
- Skin and fonts
- Title and credits
- Heroes: several characters and hero choice
- Built-in minigames
- A custom minigame
- Translations
- Music in stems
- Sound effects
- Voices: per language and per hero
- Images: prompts, cutting sheets, transparency, scale
- Asset provenance
- The storyboard, source of truth
- Puzzle design
- Classic mechanics
- Review pages
- DSL reference
- Checking your game
- What the solver proves
- Solver options
- Playtests
- Remix
- Speedrun for authors
- The sealed ending
- Production template
- Example games
- Working with an AI
- Build and release a game
- Publish on GitHub Pages
- The static game on Clever Cloud
- Other static hosts
- Sub-path, cache and service worker, incomplete deploys, going back
- Testing on a phone over the LAN
Local Studio (24)
- The local Studio: overview and launch
- Edit a room
- Placement and stage
- Preview, apply, undo
- Storyboard
- Assets and cutting
- Check: validator, solver, critical path, heat, health, coverage
- Before and after captures
- Semantic diff
- Play and replay a session
- Rule explainer
- Language and objectives
- Voices and the heroes × languages matrix
- Map
- Notes
- Remix in the Studio
- Music
- The Assistant: bring your own key
- Demo mode and studio-apply
- Local and LAN security
- Studio API
- Local MCP: setting up each client
- MCP tools
- Studio troubleshooting
Team Studio (21)
- What runs where
- Concepts: revision, transaction, span, lease, preview, build
- Starting as an invited dev
- Working together day to day
- Roles and actions: the full matrix
- Invitations
- Managing members
- Write refused on the hosted Studio: stale base revision (409)
- Write refused on the hosted Studio: span locked (423)
- Recovering after a lost lease, a disconnection or a stale preview
- Comments
- Map and semantic diff
- Publishing builds
- Remote MCP
- The game in Git
- Upgrading the engine of a hosted game
- Hosted assets
- The server's command line
- Health and metrics routes
- Trying it locally with compose.yml
- Known limits
Clever Cloud (23)
- Clever Cloud: overview, cost and prerequisites
- Apps and add-ons
- Domains and TLS
- Keycloak
- Generic OIDC and GitLab
- Configuration
- First deployment
- Backups
- RPO and RTO
- Restore the Studio on a fresh installation
- Redeploys
- Upgrading the server and checking state after a migration
- Limits of going back
- Monitoring
- Security
- Rotating keys
- Recovering admin access
- Changing domain or OIDC provider
- Worker, PostgreSQL or Cellar down
- Exporting the game and leaving the installation
- Retention
- Clever Cloud troubleshooting
- Costs and teardown
Reality Bridge and connectors (16)
- The Reality Bridge: what it is
- Quick start
- Guarantees
- Install and run the Bridge
- Backup and restore
- Several tenants
- Signals and the manifest
- The Bridge's command line
- Connectors overview
- Email connector
- Telnet connector
- SSH connector
- Open Badges connector
- The connector SDK
- Security: Biscuit and signatures
- Bridge and connector troubleshooting
Contribute (25)
- Welcome
- Developer setup
- The dev loop
- Before a pull request
- Conventions
- Writing documentation
- Architecture
- Code tour
- Import boundaries
- Which tests run when
- Mutation testing
- The coverage ratchet
- The quality baseline
- CI explained
- Field passes
- Benchmarks
- Releasing
- npm packages
- Decisions and ADRs
- Governance
- Bot pull requests
- Contributing with AI agents
- Security policy
- Code of conduct
- Licences and credits