suede-release-linter
Find missing files, malformed metadata, artwork and stem issues, split gaps, AI involvement, and delivery blockers in local music or media release folders. Use when asked to "check this release folder" or explain a delivery rejection. Writes scored Markdown and JSON reports; unknown facts remain unknown. NOT FOR: resolving rights or building a transfer package (use suede-rights-passport), sync one-sheets (use suede-sync-packaging), or rollout campaigns (use suede-campaign-in-a-box).
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1.6k
- Forks
- 491
01 前言 Front Matter
- name
- suede-release-linter
- metadata
- { "display-name": "Suede AI Release Linter Skills" }
- description
- Find missing files, malformed metadata, artwork and stem issues, split gaps, AI involvement, and delivery blockers in local music or media release folders. Use when asked to "check this release folder" or explain a delivery rejection. Writes scored Markdown and JSON reports; unknown facts remain unknown. NOT FOR: resolving rights or building a transfer package (use suede-rights-passport), sync one-sheets (use suede-sync-packaging), or rollout campaigns (use suede-campaign-in-a-box).
02 资源文件 25
- 参考文档agents/openai.yaml363 B
- 资产文件assets/metadata.example.json906 B
- 资产文件assets/release-lint-report.template.json431 B
- 资产文件assets/release-lint-report.template.md874 B
- 参考文档CARD.md4.7 KB
- 参考文档references/fix-guidance.md1.6 KB
- 参考文档references/lint-rules.md3.3 KB
- 参考文档references/metadata-fields.md2.1 KB
- 参考文档references/passport-context.md765 B
- 参考文档references/worked-handoff.md1.6 KB
- 脚本scripts/fixtures/README.md2.1 KB
- 脚本scripts/fixtures/sample-blocked-project.expected.json6.1 KB
- 脚本scripts/fixtures/sample-blocked-project.expected.md3.9 KB
- 脚本scripts/fixtures/sample-blocked-project/metadata.json839 B
- 脚本scripts/fixtures/sample-blocked-project/notes/project-notes.txt223 B
- 脚本scripts/fixtures/sample-clean-project.expected.json2.3 KB
- 脚本scripts/fixtures/sample-clean-project.expected.md1013 B
- 脚本scripts/fixtures/sample-clean-project/artwork/cover-art.jpg39.7 KB
- 脚本scripts/fixtures/sample-clean-project/audio/paper-moons-master.wav15.7 KB
- 脚本scripts/fixtures/sample-clean-project/lyrics/paper-moons-lyrics.txt186 B
- 脚本scripts/fixtures/sample-clean-project/metadata.json1.3 KB
- 脚本scripts/fixtures/sample-clean-project/stems/bass-stem.wav15.7 KB
- 脚本scripts/fixtures/sample-clean-project/stems/drums-stem.wav15.7 KB
- 脚本scripts/fixtures/sample-clean-project/stems/vocals-stem.wav15.7 KB
- 脚本scripts/lint_release.py30.8 KB
03 SKILL.md 正文
Suede AI Release Linter Skills
Compatibility
Bundled offline linter needs Python 3 and local file access; YAML metadata additionally needs PyYAML. Without Python or a local folder, use the documented text-only report fallback.
For a concrete input, expected handoff, and failure check, read the worked handoff.
Gate policy
Every check and verdict here is advice to the user, not a control on the agent. Run the checks, report results honestly, and complete what the user asked. A failed gate changes the report, never the work. The one exception: a finding that risks data loss, credential exposure, a legal or rights violation, a payment mistake, or irreversible public damage. Pause, name the risk, give the options, and let the user choose.
Overview
Audit a music or media project folder and produce a practical release-readiness report. The linter should help creators find missing files, weak metadata, rights risks, split gaps, platform-delivery blockers, and downstream handoff issues before a release or transfer package is created.
Core principle: report what is present, missing, or unknown. Never upgrade unknown to confirmed, and never treat a clean report as clearance, ownership confirmation, or approval.
Public v1 is offline-first: inspect local files and supplied metadata, do not upload files, write to a registry, call distribution APIs, request private keys, or claim legal clearance.
Workflow
- Identify the source folder or supplied files.
- Ask for the output location if it is not obvious.
- Read
references/lint-rules.mdbefore classifying any finding: it defines the categories, severities, score, and status bands. Do not assign severities from memory. - If working on a local folder, run
scripts/lint_release.pyto generaterelease-lint-report.mdandrelease-lint-report.json. Exit-code contract:0= report written with noerror-severity findings;1= report written and at least oneerrorfinding exists, which is ablockedstatus, not a script failure: do not abort or re-run on exit 1. In both cases read the generated report rather than re-deriving findings from the folder. If the source is pasted text rather than a folder, orpython3is unavailable, hand-lint againstreferences/lint-rules.md, produce the same report shape fromassets/release-lint-report.template.md, and label the report text-only. - Read
references/fix-guidance.mdwhen turning findings into specific next actions. - If the user wants downstream intake prep, use the report to decide whether to invoke or recommend the
suede-rights-passportpackage workflow. - Do not invent release metadata. Mark uncertain facts as
unknown,missing, orneeds creator confirmation. Never resolve a rights, sample, split, or ownership question yourself: a fact moves to confirmed only when the creator supplies the confirmation, and open gaps route tosuede-rights-passportaudit mode. - End with a concise summary: report path, score, status, highest-severity findings, and next fixes.
Quick Start
python3 /path/to/suede-release-linter/scripts/lint_release.py \
/path/to/music-project \
--output /path/to/release-lint-output
If the source folder contains a metadata file, pass it explicitly:
python3 /path/to/suede-release-linter/scripts/lint_release.py \
/path/to/music-project \
--metadata /path/to/music-project/metadata.json \
--output /path/to/release-lint-output
Accepted metadata formats are JSON, YAML/YML when PyYAML is installed, and
public-safe key=value text files. Do not point metadata at real .env,
credential, wallet, or deployment config files.
Safety defaults:
- Hidden files, dependency folders, build outputs, caches, and secret-like files are skipped by default.
- Unrecognized file types are skipped unless
--include-otheris passed. - Absolute local paths are redacted to share-safer names unless
--include-absolute-pathsis passed. - Existing generated report files are not overwritten unless
--forceis passed. - The output folder cannot be the same folder as the source or live inside it.
- YAML metadata requires PyYAML:
python3 -m pip install PyYAML.
What To Check
Read each bundled reference at the moment it is needed, not up front:
references/lint-rules.md: before classifying findings, or when hand-linting without the script: categories, severity levels, score, and status bands.references/metadata-fields.md: when metadata is missing, malformed, or being authored: recommended fields, accepted aliases, and confirmation values.references/fix-guidance.md: when turning findings into next actions or a fix plan.references/passport-context.md: when the user asks how the lint report relates to Suede AI review or the Suede Creator Passport.
The script writes:
release-lint-report.md: human-readable report.release-lint-report.json: machine-readable findings.
Use the bundled assets when repairing or hand-writing reports:
assets/release-lint-report.template.mdassets/release-lint-report.template.jsonassets/metadata.example.json
Fixtures
Two synthetic release folders under scripts/fixtures/ (all names and metadata
fake: no real personal data) exist only to regression-check the script. Read
scripts/fixtures/README.md when changing scripts/lint_release.py; a normal
lint of a user's folder never touches them.
Public Safety Rules
- Do not say a project is legally cleared unless the user provides explicit proof.
- Do not treat a clean lint report as a legal opinion, distributor approval, registry write, or guaranteed release.
- Do not ask for private keys, seed phrases, unreleased account secrets, or full payment credentials.
- Do not include private implementation details, private endpoints, internal provider names, or non-public pricing.
- Treat generated reports as private drafts until a creator or operator reviews and redacts them for the intended audience.
- Keep public positioning focused on broadly reusable creator workflows: metadata quality, provenance, release readiness, rights, royalty routing, licensing, and agent commerce.
Completion Checklist
Before reporting a lint result:
- Confirm the source folder was inspected or state that the report is based only on supplied text.
- Confirm whether metadata was discovered, supplied, or missing.
- Report the score and severity counts.
- List all
errorfindings and the most importantwarningfindings. - State the mechanical status the findings produce:
blocked(anyerrorfinding, or score below 50),needs-work(50-74),usable-with-cleanup(75-89), orstrong(90+). Never soften ablockedstatus in prose. - Recommend a next action: fix metadata, collect rights confirmations, prepare a rights package, or package for release.
Red flags: stop
If any of these appear in your reasoning, stop and re-read the core principle:
- "The folder looks complete, skip the script." Run it. Eyeballing is not linting.
- "The artist obviously owns it." Ownership status comes from the creator, not from the folder.
- "One unconfirmed split won't block anything." Split errors block royalty routing and licensing by rule.
- "Round the score up; it's close." The score is arithmetic, not judgment.
- "A clean report means it's cleared." A clean report means fewer prep blockers. Nothing more.
Downstream Review Context
A clean release-lint report is a portable review artifact. It can support a release, registry, licensing conversation, collaborator handoff, marketplace review, label review, advisor review, or Suede AI review without claiming that any downstream system has accepted, cleared, registered, paid, or approved the work.
Routing
- Rights, sample, split, ownership, or AI-authorship gaps in the findings → suede-rights-passport audit mode to organize the evidence.
- No
errorfindings and the user wants handoff prep → suede-rights-passport package mode. - The artist site or release page needs to rank in search or get cited in AI answers → suede-seo-audit.
- Track headed for film/TV/ads → suede-sync-packaging.
- The release needs a rollout → suede-campaign-in-a-box.
Family order: suede-release-linter (files) → suede-rights-passport (audit, then package) → suede-sync-packaging (pitch); this skill is step 1.
04 同仓库更多 Skills
worktable-review-with-annotations
Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.
worktable-find-and-synthesize-context
Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.
worktable-create-or-update-docs
Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.
worktable-create-or-manage-records
Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.
worktable-create-or-manage-interactive-html
Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.
worktable-collaborate-threads
Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.