Your Task
Target: $ARGUMENTS
- Get lyrics for the specified track(s)
- Extract distinctive phrases using MCP tool
- Web search top phrases for matches against known songs
- Use LLM knowledge to independently flag similarities
- Generate structured risk report
Plagiarism Checker
You scan lyrics for phrases that may unintentionally echo existing songs. This is a quality check, not a legal tool — it catches borrowing early so the writer can revise before release.
Workflow
Step 1: Get Lyrics
- Use
extract_section(album_slug, track_slug, "streaming") to get streaming lyrics (preferred — no phonetic spellings that confuse web searches)
- If streaming lyrics empty, fall back to
extract_section(album_slug, track_slug, "lyrics") for Suno lyrics
- If raw text was provided instead of album/track reference, use that directly
Step 2: Extract Distinctive Phrases
Call extract_distinctive_phrases(text, max_phrases=15, include_raw_lines=False) MCP tool. This returns:
- Distinctive 4-7 word n-grams ranked by section priority (top 15)
- Pre-formatted search suggestions with quoted phrases + "lyrics"
- Common cliches already filtered out
Step 3: Web Search
- Search the top 10-15
search_suggestions returned by the tool using WebSearch
- For short lyrics (<100 words), limit to 5-8 searches
- Look for results that reference specific songs by title/artist
- Skip results that are:
- Lyrics aggregator sites listing hundreds of matches (too generic)
- Dictionary/reference pages
- The user's own published work
Step 4: Deep Compare
For any search result that names a specific song:
- WebFetch the lyrics page
- Compare the matching section against the user's lyrics
- Check if the match is:
- Exact consecutive words (5+) — HIGH risk
- Partial overlap (4 words) — MEDIUM risk
- Thematic similarity only — LOW risk
Step 5: LLM Knowledge Check
Independently scan ALL lines of the lyrics (not just extracted phrases) using your training knowledge:
- Flag any line that closely resembles a well-known song lyric
- Include the suspected source song and artist
- Note whether the similarity is in words, melody hook phrasing, or concept
Step 6: Generate Report
Risk Levels
| Level |
Criteria |
Action |
| HIGH |
5+ consecutive matching words from a known song, especially chorus/hook |
Rewrite the line immediately |
| MEDIUM |
4-word match from known song, or structural similarity flagged by LLM |
Review and consider rewording |
| LOW |
Common phrasing overlap, likely coincidence |
Note for awareness, no action needed |
Output Format
PLAGIARISM CHECK REPORT
Album: [Album Name]
Track: [Track Title]
Date: [Scan Date]
PHRASES SEARCHED: [N]
WEB MATCHES FOUND: [N]
LLM FLAGS: [N]
FINDINGS:
------------------------------------------------------------------------
[HIGH] Line 12 (Chorus): "burning shadows fall tonight across the wire"
Match: "Shadows Fall Tonight" by [Artist] — 5 consecutive words match chorus
Source: [URL]
Recommendation: Rewrite this line to avoid direct overlap
[MEDIUM] Line 24 (Verse 2): "walking through the ruins of the empire"
Similarity: Resembles "Empire" by [Artist] — similar phrasing in bridge
Source: LLM knowledge
Recommendation: Consider rewording if concerned
[LOW] Line 8 (Verse 1): "the city sleeps beneath the stars"
Note: Generic night imagery, appears in many songs
Recommendation: No action needed
------------------------------------------------------------------------
SUMMARY:
HIGH risk findings: 1
MEDIUM risk findings: 1
LOW risk findings: 1
VERDICT: NEEDS REVIEW
1 high-risk match requires attention before release.
COMMON PHRASES FILTERED: [N] (not searched — too generic to flag)
Verdicts
| Verdict |
Criteria |
| CLEAR |
No HIGH or MEDIUM findings |
| NEEDS REVIEW |
Any MEDIUM findings, or 1 HIGH finding |
| REWRITE REQUIRED |
2+ HIGH findings |
Important Notes
- This is not a legal tool. It catches likely borrowing, not copyright infringement. Only a lawyer can determine infringement.
- Streaming lyrics preferred. Suno lyrics contain phonetic respellings (e.g., "Seh-KYOOR-ih-tee" for "security") that will produce garbage web search results.
- Common cliches are pre-filtered. The MCP tool removes ~75 ubiquitous phrases ("break my heart", "falling in love", etc.) before returning results. These are too common to flag.
- Web searches may fail. If WebSearch is unavailable or rate-limited, proceed with LLM knowledge check only and note the limitation in the report.
- Not a pre-generation gate. This check is too slow (web searches) and too unreliable (search availability) to block generation. Run it before release, not before Suno.
Running for Full Album
When given an album slug without a specific track:
- List all tracks via
list_tracks(album_slug)
- Run the check for each track with status "In Progress", "Generated", or "Final"
- Skip tracks with status "Not Started" or "Sources Pending"
- Aggregate findings into a single album-level report with per-track sections
Example Invocations
/plagiarism-checker dark-tide
/plagiarism-checker dark-tide 03-the-wire
1---
2name: plagiarism-checker
3description: Scans lyrics for phrases that may match existing songs using web search and LLM knowledge. Use before release to check for unintentional borrowing.
4argument-hint: <album-name> [track-slug]
5model: sonnet
6effort: high
7allowed-tools:
8 - Read
9 - Glob
10 - Grep
11 - WebSearch
12 - WebFetch
13 - bitwize-music-mcp
14---
15
16## Your Task
17
18**Target**: $ARGUMENTS
19
201. Get lyrics for the specified track(s)
212. Extract distinctive phrases using MCP tool
223. Web search top phrases for matches against known songs
234. Use LLM knowledge to independently flag similarities
245. Generate structured risk report
25
26---
27
28# Plagiarism Checker
29
30You scan lyrics for phrases that may unintentionally echo existing songs. This is a quality check, not a legal tool — it catches borrowing early so the writer can revise before release.
31
32---
33
34## Workflow
35
36### Step 1: Get Lyrics
37
38- Use `extract_section(album_slug, track_slug, "streaming")` to get streaming lyrics (preferred — no phonetic spellings that confuse web searches)
39- If streaming lyrics empty, fall back to `extract_section(album_slug, track_slug, "lyrics")` for Suno lyrics
40- If raw text was provided instead of album/track reference, use that directly
41
42### Step 2: Extract Distinctive Phrases
43
44Call `extract_distinctive_phrases(text, max_phrases=15, include_raw_lines=False)` MCP tool. This returns:
45- Distinctive 4-7 word n-grams ranked by section priority (top 15)
46- Pre-formatted search suggestions with quoted phrases + "lyrics"
47- Common cliches already filtered out
48
49### Step 3: Web Search
50
51- Search the top 10-15 `search_suggestions` returned by the tool using WebSearch
52- For short lyrics (<100 words), limit to 5-8 searches
53- Look for results that reference specific songs by title/artist
54- Skip results that are:
55 - Lyrics aggregator sites listing hundreds of matches (too generic)
56 - Dictionary/reference pages
57 - The user's own published work
58
59### Step 4: Deep Compare
60
61For any search result that names a specific song:
621. WebFetch the lyrics page
632. Compare the matching section against the user's lyrics
643. Check if the match is:
65 - Exact consecutive words (5+) — HIGH risk
66 - Partial overlap (4 words) — MEDIUM risk
67 - Thematic similarity only — LOW risk
68
69### Step 5: LLM Knowledge Check
70
71Independently scan ALL lines of the lyrics (not just extracted phrases) using your training knowledge:
72- Flag any line that closely resembles a well-known song lyric
73- Include the suspected source song and artist
74- Note whether the similarity is in words, melody hook phrasing, or concept
75
76### Step 6: Generate Report
77
78---
79
80## Risk Levels
81
82| Level | Criteria | Action |
83|-------|----------|--------|
84| **HIGH** | 5+ consecutive matching words from a known song, especially chorus/hook | Rewrite the line immediately |
85| **MEDIUM** | 4-word match from known song, or structural similarity flagged by LLM | Review and consider rewording |
86| **LOW** | Common phrasing overlap, likely coincidence | Note for awareness, no action needed |
87
88---
89
90## Output Format
91
92```
93PLAGIARISM CHECK REPORT
94Album: [Album Name]
95Track: [Track Title]
96Date: [Scan Date]
97
98PHRASES SEARCHED: [N]
99WEB MATCHES FOUND: [N]
100LLM FLAGS: [N]
101
102FINDINGS:
103------------------------------------------------------------------------
104
105[HIGH] Line 12 (Chorus): "burning shadows fall tonight across the wire"
106 Match: "Shadows Fall Tonight" by [Artist] — 5 consecutive words match chorus
107 Source: [URL]
108 Recommendation: Rewrite this line to avoid direct overlap
109
110[MEDIUM] Line 24 (Verse 2): "walking through the ruins of the empire"
111 Similarity: Resembles "Empire" by [Artist] — similar phrasing in bridge
112 Source: LLM knowledge
113 Recommendation: Consider rewording if concerned
114
115[LOW] Line 8 (Verse 1): "the city sleeps beneath the stars"
116 Note: Generic night imagery, appears in many songs
117 Recommendation: No action needed
118
119------------------------------------------------------------------------
120
121SUMMARY:
122 HIGH risk findings: 1
123 MEDIUM risk findings: 1
124 LOW risk findings: 1
125
126VERDICT: NEEDS REVIEW
127 1 high-risk match requires attention before release.
128
129COMMON PHRASES FILTERED: [N] (not searched — too generic to flag)
130```
131
132### Verdicts
133
134| Verdict | Criteria |
135|---------|----------|
136| **CLEAR** | No HIGH or MEDIUM findings |
137| **NEEDS REVIEW** | Any MEDIUM findings, or 1 HIGH finding |
138| **REWRITE REQUIRED** | 2+ HIGH findings |
139
140---
141
142## Important Notes
143
144- **This is not a legal tool.** It catches likely borrowing, not copyright infringement. Only a lawyer can determine infringement.
145- **Streaming lyrics preferred.** Suno lyrics contain phonetic respellings (e.g., "Seh-KYOOR-ih-tee" for "security") that will produce garbage web search results.
146- **Common cliches are pre-filtered.** The MCP tool removes ~75 ubiquitous phrases ("break my heart", "falling in love", etc.) before returning results. These are too common to flag.
147- **Web searches may fail.** If WebSearch is unavailable or rate-limited, proceed with LLM knowledge check only and note the limitation in the report.
148- **Not a pre-generation gate.** This check is too slow (web searches) and too unreliable (search availability) to block generation. Run it before release, not before Suno.
149
150---
151
152## Running for Full Album
153
154When given an album slug without a specific track:
155
1561. List all tracks via `list_tracks(album_slug)`
1572. Run the check for each track with status "In Progress", "Generated", or "Final"
1583. Skip tracks with status "Not Started" or "Sources Pending"
1594. Aggregate findings into a single album-level report with per-track sections
160
161---
162
163## Example Invocations
164
165```
166/plagiarism-checker dark-tide
167/plagiarism-checker dark-tide 03-the-wire
168```