- Home
- Skills
- Workflow Automation
- Pre-Generation Check: Music Release Gates
More screenshots
Works with the AI Agents you already use
Pre-Generation Check: Music Release Gates
The last gate before AI music generation: six pass/fail checks per track — sources, lyrics, pronunciation, explicit flag, style prompt, artist names — blocking bad renders before they cost credits.
$5
Pre-Generation Check: Music Release Gates
Example session with this skill installed
Run a pre-gen check on this track before I generate it. Track: "Copper Wire Heart," vocal track. Lyrics are done except the bridge, which still says [BRIDGE - something about the radio tower]. The word "tear" appears in the chorus (crying sense) and "live" in verse 1 (reside sense); my notes say "handle pronunciation later." Style prompt: "melancholic indie folk in the style of [a real, named folk artist], fingerpicked guitar, soft male vocal." Explicit flag: not set yet, song is clean. Sources: original, nothing to verify.
- Read your context and instructions
- Compiled the pre-generation check
GATE REPORT — Copper Wire Heart
Instrumental: no (metadata consistent: yes)
[SKIP] Sources verified → original work, no source material
[FAIL] Lyrics complete → bridge is a bracketed placeholder ("something about the radio tower"), not lyrics. Route: lyric writing pass.
[FAIL] Pronunciation resolved → two live homographs with no resolution in the text: "tear" (chorus, crying sense — must read as teer) and "live" (verse 1, reside sense — must read as liv). The note "handle later" is not a resolution; the respelling has to exist in the lyric itself. Route: pronunciation pass, apply respellings in the text.
[FAIL] Explicit flag → not set. The song being clean is the reason to set it to "no" deliberately, not to leave it blank for the platform to default. Route: set the flag.
[FAIL] Style prompt complete → contains a real artist's name, which fails the artist-names gate on its own; underneath that, the rest of the box is usable (instrumentation and vocal description present). Route: rewrite the name into descriptors — era, region, recording style — and keep the rest.
[FAIL] Artist names cleared → named artist in the style prompt. Route: same rewrite as above.
Advisories: section tags carry performance cues only on the chorus; verses are bare.
VERDICT: BLOCKED (5 failing gates)
What happens next: the bridge gets written, both homographs get respelled in the lyric text, the flag is set to "no," and the style prompt trades the name for descriptors. Then the track is re-checked — not assumed fixed — because a respell that breaks a rhyme is exactly the kind of regression the second run exists to catch. Total cost of the gate: minutes. Cost of generating first: one full render of a song with a spoken placeholder for a bridge.
Connects securely to your tools. The creator never sees your data.
About this skill
A track with a placeholder lyric or an unresolved homograph does not produce a draft to fix later; it produces a polished-sounding wrong take you regenerate anyway, at full price. This skill is the gate that catches those problems while they are still text. Its philosophy is fail closed: an unchecked box blocks generation, the checker issues verdicts only (it never edits — editing creates pressure to pass your own work), and every failure routes to the work that owns the fix. The run starts with instrumental triage, because instrumental status changes which gates apply — and metadata that disagrees with itself about it is an instant block. Then the gates run by risk class: rights and safety first (sources verified, artist names cleared from the style prompt — both never waivable), craft readiness next (lyrics actually complete, every pronunciation risk respelled in the lyric text rather than noted in a table, a style prompt specific enough to steer the render), and the explicit flag, which must be set deliberately instead of inherited from a platform default. Advisories (style-box bloat, missing performance cues, unrecorded generation settings) warn without blocking. You get the fixed gate-report format, a pre-check checklist, a gate severity table, evidence standards so two checkers reach the same verdict, re-run rules for fixed tracks, and an album rollup that reports "N of M ready" so one bad track sheet stalls one track — not the whole session.
- It issues verdicts only — it will not write the missing bridge, respell the homograph, or rewrite the style prompt; each failure routes to that fix elsewhere.
- It checks the text you provide; if you paste a draft instead of the submission version, the gate certifies the draft.
- It cannot verify source material itself — it checks that your verification record says verified, not that the verification was correct.
- Platform-specific flag behavior and content filters change over time; the gate checks the flag is set, not how a given platform will treat it.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Permissions
No special permissions declared or detected