Release-Evidence Folder Taxonomy for Solo Indie Teams - Q3 2026 Cert Intake Without Scrambling (2026)

Your cert rejection email references artifact folder three. Your teammate replies, “Which Drive link?” Partner intake asks for reproducible capture and you send a Loom link from February. Q3 2026 console and partner windows do not wait for you to invent a filing system under pressure.

This Process & Workflow guide gives solo and two-person teams a single release-evidence/ taxonomy—seven subfolders, one README map, and a ninety-minute bootstrap you can run in May 2026 before August intake ramps.

What you will have after ninety minutes

A reviewer—or your future self—can open the repo, read release-evidence/README.md, and answer which build, which bullets, and which file without opening Discord. That is the whole win. Everything below is scaffolding for that outcome.

Who this is for

• 1–3 person teams with no producer writing spec folders
• Studios approaching first console submission or first publisher diligence in H2 2026
• Teams already running a 30-minute operating review but storing proof in random chat threads

Not for: fifty-person pipelines with dedicated compliance staff—they already have CM tools.

Why this matters now (May 2026)

Three pressures converge this quarter:

1. Q3 2026 partner cert intake opens late August for many late-2026 launches—packets submitted without folder discipline get “could not reproduce” bounce-backs per shared 2026 forum patterns.
2. Reproducible-capture expectations hardened across Quest, Deck Verified refresh intake, and indie console programs—reviewers want mapped artifacts, not vibes.
3. Stack rationalization (one engine, one storefront) concentrates risk on fewer SKUs—your evidence folder becomes the institutional memory when headcount does not.

Direct answer (TL;DR)

1. Create release-evidence/ at repo root (not inside Assets/).
2. Add seven subfolders (below) plus README.md mapping rejection bullets → paths.
3. Tag each submission as release-evidence/<tag>/ (example: 2026-08-rc1).
4. Never delete old tags—append new tags; reviewers compare builds.
5. Bootstrap in 90 minutes once; maintain in 15 minutes weekly via operating review Block 5.
6. Pair with Quest resubmission playbook when console is in scope.

The root layout (copy this tree)

release-evidence/
  README.md                 # master map + version table
  _templates/               # empty stubs new tags copy
  ai-disclosure/            # generative vs assistive inventory
  build-and-binary/         # manifests, hashes, depot notes
  cert-and-platform/        # rejection PDFs, resubmission maps
  pricing-and-store/        # anchor tests, regional tables
  privacy-and-legal/        # policies, Data Safety exports
  qa-and-repro/             # replay hooks, input captures, soaks
  marketing-and-demo/       # demo labels, trailer frame proofs

Rule: If a file does not fit one folder, you either need a new tag (submission) or you are hoarding unrelated PDFs.

Treat release-evidence/ like a save file for institutional memory: you do not edit the save from three apps at once—you load one tag, update mapped rows, save, and exit. Studios that let every teammate create ad-hoc top-level folders (“misc”, “old”, “final2”) reliably lose cert cycles to confusion, not code quality.

README.md master map (template)

# Release evidence index

| Tag | Date | Build ID | Platforms | Owner |
|-----|------|----------|-----------|-------|
| 2026-08-rc1 | 2026-08-12 | 1.0.4.512 | Win + Quest | Alex |

## Bullet → artifact map (2026-08-rc1)

| Reviewer / partner bullet | Path |
|---------------------------|------|
| Save corruption on force-quit | qa-and-repro/2026-08-rc1/input-capture-save-quit.zip |
| AI disclosure incomplete | ai-disclosure/2026-08-rc1/touchpoint-inventory.md |
| Regional price mismatch | pricing-and-store/2026-08-rc1/regional-table-v3.csv |

## Freshness
- Last operating-review check: 2026-08-09
- Next tag planned: 2026-08-rc2 (scope: localization overflow only)

Reviewers read this file first. If they cannot map a bullet in five minutes, the taxonomy failed—not the reviewer. Your job is to make row three obvious, not to impress anyone with folder depth or clever codenames.

Folder 1 — ai-disclosure/

Purpose: One honest packet for Steam, Play, App Store, and partner annex questions.

Contents per tag:

• touchpoint-inventory.md (store, demo, trailer, Discord bot, etc.)
• generative-vs-assistive.md (what models touched which assets)
• Screenshots of disclosure UI where required

Pairs with: AI disclosure intake checklist.

Anti-pattern: Dumping ChatGPT logs without classification—reviewers do not grade poetry.

Folder 2 — build-and-binary/

Purpose: Prove which bits shipped.

Contents per tag:

• build-manifest.json or engine export equivalent
• depot-parity-notes.md (demo vs main branch)
• Hash list for Windows/Mac/Linux artifacts

Pairs with: manifest diff routines and demo depot parity passes.

Anti-pattern: “It is the latest build” without ID—useless in cert disputes.

Folder 3 — cert-and-platform/

Purpose: Console and partner submission cycles.

Contents per tag:

• Redacted rejection PDFs or bullet lists
• resubmission-map.md (one row per bullet → evidence path)
• Scope-cut document when applicable

Pairs with: 21-day Quest resubmission playbook and milestone payment checklists.

Anti-pattern: Patching week one without updating the map—second rejection guaranteed.

Folder 4 — pricing-and-store/

Purpose: Store economics evidence—not finance spreadsheets for investors.

Contents per tag:

• Locked US anchor doc from price-anchor worksheet
• Regional CSV version
• Screenshot of store strings during A/B weeks

Pairs with: Regional pricing guide.

Folder 5 — privacy-and-legal/

Purpose: Policies and platform compliance exports.

Contents:

• Privacy policy version hash or URL snapshot date
• Google Play Data Safety JSON export
• Apple privacy nutrition labels screenshot

Pairs with: 14 free privacy templates.

Folder 6 — qa-and-repro/

Purpose: Reproduce the bug on reviewer hardware.

Contents per tag:

• Input recordings / replay hook exports
• soak-test-log.md (72-hour notes)
• Soft-lock repro steps with save snapshots

Pairs with: Deterministic soft-lock replay hooks and screenshot-first bug intake.

Anti-pattern: Video-only repro with no save file—reviewers hate it in 2026.

Folder 7 — marketing-and-demo/

Purpose: Prove store promises match build truth.

Contents per tag:

• Demo label screenshot (prototype vs vertical slice honesty)
• Trailer frame grabs with feature callouts circled
• Capsule vs in-game comparison stills

Pairs with: Demo is not a vertical slice and capsule iteration calendar.

_templates/ — stop reinventing folders

Copy _templates/tag-stub/ when opening a new tag:

_templates/tag-stub/
  README.md
  ai-disclosure/.gitkeep
  build-and-binary/.gitkeep
  cert-and-platform/.gitkeep
  pricing-and-store/.gitkeep
  privacy-and-legal/.gitkeep
  qa-and-repro/.gitkeep
  marketing-and-demo/.gitkeep

Automation optional; copy-paste is fine for micro-studios.

Ninety-minute bootstrap sprint

Done is better than perfect—empty folders with README rows beat heroic last-minute ZIP files.

Weekly maintenance (15 minutes)

Add to operating review Block 5 — Cert and compliance:

• README freshness date updated? Y/N
• New tag needed this week? Y/N
• Any bullet unmapped? Y/N

If “unmapped” is yes twice in a row, stop feature work until mapped—cert debt compounds.

Tag naming discipline

Tags are immutable labels—never reuse a tag for a different binary.

Size and secrets guardrails

• Target 40–60 MB per cert packet zip—not gigabyte screen recordings.
• No API keys, .env, or partner NDAs in git—use redacted PDFs.
• Large captures live in object storage; README links with expiry notes.

Three-person RACI (lightweight)

Solo dev? You wear all hats—README Owner column still helps future you.

Integration with publisher diligence

Publishers in 2026 increasingly ask: “Show me your evidence cadence.” Point them to:

1. Root README version table
2. Latest tag folder
3. Operating review sheet mentioning Block 5

This is stronger than claiming “we are organized.”

Common mistakes

1. Storing evidence only on one laptop
2. Mixing release-evidence/ with Assets/Plugins/
3. Deleting old tags when embarrassed by bugs
4. Uploading unredacted player emails in repro zips
5. Skipping marketing-and-demo/ until trailer controversy
6. Writing novels in README instead of paths
7. Forgetting to update map after scope cut

When to fork a new repo root

Stay in game repo unless:

• Legal insists separate compliance repo (publisher contract)
• Open-source game with private console binaries—use private submodule

Default: one repo, one evidence root.

Decision tree

Cert or partner intake in next 90 days?
  no → Still bootstrap (fest and refunds happen anyway)
  yes → Do ninety-minute sprint this week
        ↓
Rejection received?
  yes → New tag + cert-and-platform map before patch
  no → Keep weekly 15-min freshness

Engine-specific notes (brief)

Unity: Export ProjectVersion.txt + scripting define list into build-and-binary/.
Godot: Pin export_presets.cfg snippet + feature tags.
Unreal: Cook manifest + staged build ID in same folder.

Engine wars do not change folder names—reviewers think in bullets, not engines.

File naming inside each tag

Consistent names beat clever ones:

Avoid spaces in filenames; use hyphens. Reviewers on Windows thank you.

Packing a cert zip without rage

When a platform portal wants one upload:

1. Create cert-and-platform/<tag>/outbound/
2. Copy only paths listed in README map rows for active bullets
3. Add README.md from tag root at zip top level
4. Target 40–60 MB—compress videos separately with links in README
5. Name zip studio-title-<tag>-cert-packet.zip

If the zip exceeds portal limits, split by bullet group, not by random folders.

Worked example (synthesized pattern)

A two-person narrative roguelite team opened tag 2026-07-quest-1 after a first Quest rejection with nine bullets—three cert-blocking. Instead of patching immediately, they spent day one only on README rows:

Day two they filled capture slots using replay hooks. Day six they updated the same README rows—no new mystery folders. Second submission referenced one README URL in the cover letter.

Forum post-mortems describe this as the fastest path—not because patches were magic, but because reviewers stopped asking “where is artifact three?”

Q3 2026 intake calendar (planning aid)

Align with festival application calendar so fest and cert tags do not collide.

Relationship to studio/weekly-reviews/

The operating review lives in studio/weekly-reviews/<date>.md. Each Friday footer should link:

## Evidence freshness
- release-evidence README last touched: 2026-08-09
- Active tag: 2026-08-rc1
- Unmapped bullets: 0

Rhythm sheet points to proof vault—neither replaces the other.

Telemetry and privacy cross-links

If you log privacy-safe telemetry, store:

• Data processing summary in privacy-and-legal/
• Event schema version in build-and-binary/<tag>/

Do not paste raw player IDs into qa-and-repro/—hash or redact.

Patch notes and human-gated AI copy

Store two-pass verification artifacts in cert-and-platform/<tag>/:

• Pass 1 draft
• Pass 2 signed-off CHANGELOG excerpt
• Per-platform verification lines

Tie to human-gated AI patch notes so cert reviewers see process, not just prose.

Godot and Phaser web exports

Web builds add marketing-and-demo/ proof that WASM URL matches store “coming soon” language. Pair engine guides:

• Godot threaded loader for floor-transition stability evidence
• Phaser tilemap streaming for browser OOM repro notes in qa-and-repro/

Challenges and sprints that feed the tree

Running a seven-day build stability challenge or devlog habit challenge? Export daily gate results into build-and-binary/<tag>/stability-log.md—free evidence for partners.

Listicle resources belong linked, not duplicated

Your privacy-and-legal/ folder should point to template sources from the 14 templates listicle, not fork legal text twice. Version the snapshot date in README.

Redaction checklist before any upload

• Remove player emails and Steam IDs from logs
• Blur keys in screenshots
• Replace real reviewer names with “Reviewer note #3” in internal copies
• Confirm NDAs allow each PDF in the zip

Auditor questions you should answer from README alone

1. Which build ID is this packet about?
2. Which bullets are in scope for this submission?
3. Where is the repro for bullet #2?
4. When was evidence last verified?
5. Who owns the next tag?

If README cannot answer all five, delay submit one day—not one week of rework later.

Scaling to four people without new philosophy

Add reviewer column to README table—still seven folders. Optional release-evidence/inbox/ for screenshots from Discord that get sorted Friday, not stored forever in inbox.

Anti-patterns from 2026 post-mortems

Post-cert: what to keep alive

After pass, do not archive discipline:

• Keep replay hooks in QA builds
• Keep README version table growing
• Keep marketing-and-demo/ updated when store strings change

Cert pass is a checkpoint, not graduation from evidence.

Key takeaways

1. One release-evidence/ root beats scattered Drive links.
2. Seven subfolders cover 2026 indie cert reality.
3. README bullet → path table is the highest-signal artifact.
4. Tags are immutable submission labels.
5. Bootstrap in 90 minutes; maintain in 15 weekly.
6. Pair folders with existing checklists—do not duplicate content, link it.
7. qa-and-repro/ is where second rejections are won or lost.
8. marketing-and-demo/ prevents store-truth drift.
9. Q3 2026 intake rewards teams who started in May.
10. Operating review Block 5 keeps the system alive.
11. Separate tags for Deck vs Quest vs fest demos—never merge unrelated bullets.
12. Redact before zip; link heavy media instead of stuffing portals.

Block-by-block operating review prompts

Copy these into your Friday sheet when Block 5 runs:

Engineering (feeds build-and-binary/):
“What build ID shipped to testers this week, and is it in the manifest folder?”

Art (feeds marketing-and-demo/):
“Do screenshots in the store still match the current lighting pass?”

Production (feeds cert-and-platform/):
“Any new rejection bullets without README rows?”

Marketing (feeds marketing-and-demo/ + pricing-and-store/):
“Did we change store strings without updating the pricing tag?”

Finance (feeds pricing-and-store/):
“Is regional table version pinned for this tag?”

Five questions, five folders—no sixth mystery bucket.

Steam Deck Verified vs console tags

Use separate tags when parallel tracks run:

• 2026-08-deck-verify-1
• 2026-08-quest-resubmit-1

Shared artifacts (build manifest) may link across tags in README (“manifest duplicated from…”) but do not merge unrelated bullets into one map—Deck and Quest reviewers care about different failures.

Wishlist and discovery evidence (optional eighth note)

Discovery diagnostics from post-Next-Fest plateau playbook can live as marketing-and-demo/<tag>/discovery-notes.md when you are not in cert—still under the same root so partners see one system.

Branching and CI hooks (lightweight)

Optional release-evidence/ci/ (outside the seven) for:

• Screenshot of green CI on tag commit
• Export of test counts

Not required day one; add when build stability challenges become habit.

FAQ

We are PC-only on Steam. Do we need cert-and-platform/?
Keep the folder with a README note “N/A until console”—structure stays stable when you port.

Can we use Notion instead of git?
Export weekly PDF snapshots into the matching subfolder; git is simpler for diffs.

How is this different from the operating review sheet?
The sheet is weekly rhythm; release-evidence/ is durable proof.

What if we have no rejections yet?
Fill build-and-binary/ and marketing-and-demo/ pre-launch—cheapest time.

Partner asked for SOC2. Does this cover it?
No—this is game shipping evidence, not enterprise security certification.

Should voice actors contracts live here?
Summary only in privacy-and-legal/; full contracts offline with reference row.

We use Perforce. Same folders?
Yes—path at depot root, not inside art streams.

Maximum tags before clutter?
Archive tags older than 12 months to _archive/<year>/—never delete.

Can one bullet map to two paths?
Yes—comma-separate in README; rare and should be explained in one sentence.

Do we version README in git tags?
Tag releases with git annotated tags matching release-evidence tag names when possible.

What about Early Access?
Treat each public EA build as a new tag; players compare patch notes to folder evidence internally even if reviewers do not.

Our publisher uses SharePoint.
Mirror README table weekly; git remains source of truth for engineers.

Is Loom ever acceptable?
Supplement only—pair every Loom with repro steps and save file path in qa-and-repro/.

How does AI art fit?
ai-disclosure/ must list every generative touchpoint; link checklist article, do not restate law.

We failed internal dry run. New tag?
Yes—2026-08-rc2 with map diff highlighting only new failures.

Solo dev with no git?
Use dated zip backups off-machine; README still required inside zip root.

Does this replace GDD or Trello?
No—design docs stay elsewhere; only shipping proof lives here.

We are open-source on GitHub. Private evidence?
Use a private repo or private submodule for console binaries; public repo holds README structure with redacted examples.

Close: Cert intake in Q3 2026 is a paperwork boss fight. A boring folder tree is your best weapon—build it in May, dry-run in July, and hand August reviewers a README they can read in five minutes instead of twelve Slack threads you already lost. Start with the tree tonight; your October self will not volunteer to organize evidence during fest week.