music-composition— agent skill

Modular music-composition skill for AI assistants — harmony, melody, form, 24+ genres.

by SJY051·Agent Skill·github.com/SJY051/music-composition

Is music-composition safe to install?

SaferSkills independently audited music-composition (Agent Skill) and scored it 66/100 (yellow). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 8 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

Score
66/100
●●●●●●●○○○
↑ +0 since first scan (66 → 66)Re-scan~30s
Latest scan
ScannedJun 23, 2026 · 33d ago
Scans run1 over 90 days
Detectors55 checks · 5 categories
Findings8 warnings · 0 high
EngineSaferSkills 2b638c6
View methodology →
SaferSkills installs
This week0
This month0
All time0
CategoryWeightCategory scoreContribution
Securityprompt, exec, net, exfil, eval
35%
4
1.4 pts
Supply chainhash, typosquat, maintainer, lockfile
20%
100
20.0 pts
Maintenancestaleness, pinning, CI
15%
100
15.0 pts
TransparencySKILL.md, perms, README
15%
100
15.0 pts
Communityinstalls, verify, response
15%
100
15.0 pts

Findings & checks · 8 flagged

Securityscore 4 · 8 findings
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · assets/web-search-cheatsheet.md
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptassets/web-search-cheatsheet.md· markdown
213- Do not treat global charts as local taste.
214- Do not use SEO listicles as primary evidence when official charts, local press, or practit
… (23 chars elided on L214)
215- Do not ask for the user's streaming login, password, tokens, or private account access.
216- Do not copy melodies, lyrics, riffs, samples, or signature artist identifiers from referen
… (4 chars elided on L216)
217 
Occurrences
1 occurrence · at L215
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha25668b030f55bcd484arubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · references/00-navigation.md×2
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptreferences/00-navigation.md· markdown
249- The user's level is critical and unclear (a beginner asking about "modes" needs different
… (55 chars elided on L249)
250- The request is ambitious and could go in opposing directions ("write a song" — you need to
… (79 chars elided on L250)
251- Personal listening context would materially change the answer; ask whether the user can sh
… (101 chars elided on L251)
252 
253Don't ask for the sake of asking. If context is clear or the answer doesn't depend on the mi
… (24 chars elided on L253)
Occurrences
2 occurrences · first at L251, also L253
Show all 2 locations
Line
File
L251
references/00-navigation.md
L253
references/00-navigation.md
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha25614b7852251bf4666rubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · references/creative-workflows/musical-brainstorming.md
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptreferences/creative-workflows/musical-brainstorming.md· markdown
66## First response pattern
67 
68When the user asks for brainstorming and gives little context, do not ask five questions. As
… (86 chars elided on L68)
69 
70Good:
Occurrences
1 occurrence · at L68
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha25612211d7a427c799arubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · references/creative-workflows/user-agent-collaboration.md
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptreferences/creative-workflows/user-agent-collaboration.md· markdown
253## When not to ask a question
254 
255Do not ask a clarifying question if the user clearly wants momentum and the missing detail c
… (21 chars elided on L255)
256 
257Instead:
Occurrences
1 occurrence · at L255
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha25668b030f55bcd484arubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · references/research/user-listening-context-and-streaming-services.md×2
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptreferences/research/user-listening-context-and-streaming-services.md· markdown
28## Safety and privacy baseline
29 
30Never ask the user for passwords, account tokens, private login links, or anything that requ
… (22 chars elided on L30)
31 
32Safe inputs include:
Occurrences
2 occurrences · first at L30, also L188
Show all 2 locations
Line
File
L30
references/research/user-listening-context-and-streaming-services.md
L188
references/research/user-listening-context-and-streaming-services.md
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha2565084aa178cfd9d97rubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · references/source-bibliography.md
MEDIUMit fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptreferences/source-bibliography.md· markdown
239Use caution:
240 
241- Never ask users for passwords, account tokens, or private credentials.
242- A user's personal listening history is taste evidence, not a universal market signal.
243- Platform charts have different weighting rules and demographics.
Occurrences
1 occurrence · at L241
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha2565084aa178cfd9d97rubric 365aacaView on GitHub
Supply chainscore 100 · 0 findings
All supply chain checks passedNo findings in this category for the latest scan.pass
Maintenancescore 100 · 0 findings
All maintenance checks passedNo findings in this category for the latest scan.pass
Transparencyscore 100 · 0 findings
All transparency checks passedNo findings in this category for the latest scan.pass
Communityscore 100 · 0 findings
All community checks passedNo findings in this category for the latest scan.pass
Vendor response · right of reply
Are you the maintainer? Submit a response →

Audit the pieces. Scan the whole. Decide.

~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.