@krotovm/gitlab-ai-review 1.0.41 → 1.0.43
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +2 -2
- package/dist/cli/ci-review.js +43 -32
- package/dist/cli/tooling.js +6 -0
- package/dist/cli.js +48 -50
- package/dist/gitlab/services.js +12 -75
- package/dist/gitlab/types.js +0 -2
- package/dist/prompt/index.js +3 -5
- package/dist/prompt/messages.js +5 -9
- package/dist/prompt/templates/postprocess-system.js +94 -108
- package/dist/prompt/templates/review-system.js +104 -116
- package/dist/prompt/templates/triage-system.js +42 -48
- package/dist/prompt/templates/user-prompts.js +34 -42
- package/package.json +2 -3
|
@@ -1,118 +1,104 @@
|
|
|
1
1
|
/** @format */
|
|
2
2
|
import { NO_FINDINGS_SENTENCE } from "../profile.js";
|
|
3
|
-
function
|
|
4
|
-
return
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
"GitLab-flavoured markdown.",
|
|
18
|
-
];
|
|
3
|
+
function buildConsolidateDefaultPrompt(maxFindings) {
|
|
4
|
+
return `You consolidate per-file code review findings into a final ranked list.
|
|
5
|
+
Select the top ${maxFindings} most important findings. Deduplicate overlapping issues.
|
|
6
|
+
Preserve this exact per-finding markdown block:
|
|
7
|
+
\`- [high|medium] <title>\`
|
|
8
|
+
\` File: <path>\`
|
|
9
|
+
\` Line: ~<N>\`
|
|
10
|
+
\` Why: <one concise sentence with key evidence>\`
|
|
11
|
+
Rank by: (1) [high] before [medium], (2) correctness > security > perf.
|
|
12
|
+
Do not add new findings. Do not add headings, summaries, or commentary.
|
|
13
|
+
Drop any finding that: claims a syntax error without quoting the invalid token, claims a function/variable is missing without concrete proof, or flags a refactoring rename as a bug.
|
|
14
|
+
If fewer than ${maxFindings} findings exist, return all of them.
|
|
15
|
+
If all findings are low quality or dubious after dedup, return exactly: "${NO_FINDINGS_SENTENCE}"
|
|
16
|
+
GitLab-flavoured markdown.`;
|
|
19
17
|
}
|
|
20
|
-
function
|
|
21
|
-
return
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
`If after dedup nothing survives, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`,
|
|
50
|
-
];
|
|
18
|
+
function buildConsolidateWeakPrompt(maxFindings) {
|
|
19
|
+
return `You merge per-file review findings into one ranked list.
|
|
20
|
+
Output bullets only — no preamble, no headings, no code fences.
|
|
21
|
+
Keep at most ${maxFindings} findings. Do not invent new ones.
|
|
22
|
+
|
|
23
|
+
Steps:
|
|
24
|
+
1. Drop duplicates and findings that are reworded versions of each other (keep the clearest).
|
|
25
|
+
2. Drop a finding if it has no concrete evidence (no file/line, vague reasoning, or just "this might be wrong").
|
|
26
|
+
3. Sort the remainder: all [high] before any [medium]; within a tier, correctness bugs before security before perf.
|
|
27
|
+
4. Take the top items up to the limit.
|
|
28
|
+
|
|
29
|
+
Output format — copy spacing and case exactly, do not wrap bullets in backticks:
|
|
30
|
+
- [high|medium] <short title>
|
|
31
|
+
File: <path>
|
|
32
|
+
Line: ~<N>
|
|
33
|
+
Why: <one sentence pointing at the evidence>
|
|
34
|
+
|
|
35
|
+
Example output (two findings, ranked):
|
|
36
|
+
- [high] Null pointer dereference in user lookup
|
|
37
|
+
File: src/user/service.ts
|
|
38
|
+
Line: ~42
|
|
39
|
+
Why: getUser() can return null but the next line calls user.id without a guard.
|
|
40
|
+
|
|
41
|
+
- [medium] Off-by-one in pagination loop
|
|
42
|
+
File: src/api/list.ts
|
|
43
|
+
Line: ~88
|
|
44
|
+
Why: loop runs while i <= total but indices are 0-based, so the last item is read past the array end.
|
|
45
|
+
|
|
46
|
+
If after dedup nothing survives, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`;
|
|
51
47
|
}
|
|
52
|
-
function
|
|
53
|
-
return
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
"GitLab-flavoured markdown.",
|
|
70
|
-
];
|
|
48
|
+
function buildVerificationDefaultPrompt(maxFindings) {
|
|
49
|
+
return `You are a skeptical verifier of a merge request review.
|
|
50
|
+
Your job is to remove weak, speculative, or unsupported findings from the draft list.
|
|
51
|
+
Tools get_file_at_ref and grep_repository are available. Use them to check claims about current code against the repository at the MR head ref.
|
|
52
|
+
Drop a finding if file contents at refs.head contradict it, or if it cannot be verified after reasonable tool use.
|
|
53
|
+
Do not add new findings. Keep, rewrite for clarity, or remove existing findings only.
|
|
54
|
+
A finding can stay only if supported by the per-file evidence pool and not contradicted by tools when the claim is about code that exists at refs.head.
|
|
55
|
+
If confidence is not high, drop the finding.
|
|
56
|
+
Preserve this exact per-finding markdown block:
|
|
57
|
+
\`- [high|medium] <title>\`
|
|
58
|
+
\` File: <path>\`
|
|
59
|
+
\` Line: ~<N>\`
|
|
60
|
+
\` Why: <one concise sentence with key evidence>\`
|
|
61
|
+
Do not add headings, summaries, or extra commentary.
|
|
62
|
+
Return at most ${maxFindings} findings.
|
|
63
|
+
If no findings survive verification, return exactly: "${NO_FINDINGS_SENTENCE}"
|
|
64
|
+
GitLab-flavoured markdown.`;
|
|
71
65
|
}
|
|
72
|
-
function
|
|
73
|
-
return
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
`If after verification nothing survives, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`,
|
|
101
|
-
];
|
|
66
|
+
function buildVerificationWeakPrompt(maxFindings) {
|
|
67
|
+
return `You verify a draft list of merge-request review findings against the actual repository.
|
|
68
|
+
Output bullets only — no preamble, no headings, no code fences. Do not invent new findings.
|
|
69
|
+
Return at most ${maxFindings} findings.
|
|
70
|
+
|
|
71
|
+
For each draft finding, follow these steps in order:
|
|
72
|
+
1. Look it up in the per-file evidence pool. If it is not in the evidence pool, drop it.
|
|
73
|
+
2. If the finding makes a claim about code that exists at refs.head, call get_file_at_ref(path, refs.head) and confirm the offending text really sits at the cited line (±10 lines is fine).
|
|
74
|
+
3. If the file contents at refs.head do not contain the offending text, or contain a corrected version of it, drop the finding.
|
|
75
|
+
4. If you cannot confirm the claim with one or two tool calls, drop the finding.
|
|
76
|
+
|
|
77
|
+
Keep a finding only after step 3 succeeds. Otherwise drop it.
|
|
78
|
+
|
|
79
|
+
Output format — copy spacing and case exactly, do not wrap bullets in backticks:
|
|
80
|
+
- [high|medium] <short title>
|
|
81
|
+
File: <path>
|
|
82
|
+
Line: ~<N>
|
|
83
|
+
Why: <one sentence pointing at the evidence you confirmed>
|
|
84
|
+
|
|
85
|
+
Example — KEEP this finding because get_file_at_ref shows the cited code at refs.head:
|
|
86
|
+
- [high] Null pointer dereference in user lookup
|
|
87
|
+
File: src/user/service.ts
|
|
88
|
+
Line: ~42
|
|
89
|
+
Why: getUser() returns null on miss and the next line calls user.id without a guard (confirmed in src/user/service.ts at refs.head).
|
|
90
|
+
|
|
91
|
+
Example — DROP a finding when refs.head shows the bug is already gone (do not output it).
|
|
92
|
+
|
|
93
|
+
If after verification nothing survives, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`;
|
|
102
94
|
}
|
|
103
|
-
export function
|
|
104
|
-
return buildConsolidateDefaultLines(maxFindings);
|
|
105
|
-
}
|
|
106
|
-
export function buildVerificationSystemLines(maxFindings) {
|
|
107
|
-
return buildVerificationDefaultLines(maxFindings);
|
|
108
|
-
}
|
|
109
|
-
export function getConsolidateSystemLines(profile, maxFindings) {
|
|
95
|
+
export function getConsolidateSystemPrompt(profile, maxFindings) {
|
|
110
96
|
return profile === "weak"
|
|
111
|
-
?
|
|
112
|
-
:
|
|
97
|
+
? buildConsolidateWeakPrompt(maxFindings)
|
|
98
|
+
: buildConsolidateDefaultPrompt(maxFindings);
|
|
113
99
|
}
|
|
114
|
-
export function
|
|
100
|
+
export function getVerificationSystemPrompt(profile, maxFindings) {
|
|
115
101
|
return profile === "weak"
|
|
116
|
-
?
|
|
117
|
-
:
|
|
102
|
+
? buildVerificationWeakPrompt(maxFindings)
|
|
103
|
+
: buildVerificationDefaultPrompt(maxFindings);
|
|
118
104
|
}
|
|
@@ -1,120 +1,108 @@
|
|
|
1
1
|
/** @format */
|
|
2
2
|
import { NO_FINDINGS_SENTENCE } from "../profile.js";
|
|
3
|
-
const
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
"
|
|
105
|
-
" File: src/api/list.ts",
|
|
106
|
-
" Line: ~88",
|
|
107
|
-
" Why: loop runs while i <= total but indices are 0-based, so the last item is read past the array end.",
|
|
108
|
-
"",
|
|
109
|
-
`If you cannot meet all three conditions for any finding, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`,
|
|
110
|
-
];
|
|
111
|
-
export function getMainSystemLines(profile) {
|
|
112
|
-
return profile === "weak" ? MAIN_WEAK_LINES : MAIN_DEFAULT_LINES;
|
|
3
|
+
const MAIN_DEFAULT = `You are an AI code reviewer for pull requests.
|
|
4
|
+
Find only real bugs introduced by the diff.
|
|
5
|
+
Return at most 3 findings. Prefer no finding over a weak one.
|
|
6
|
+
|
|
7
|
+
Rules:
|
|
8
|
+
- Focus on changed lines.
|
|
9
|
+
- Ignore style, refactoring suggestions, and general best practices.
|
|
10
|
+
- If uncertain, use tools: get_file_at_ref and grep_repository.
|
|
11
|
+
- Report only issues clearly visible in diff or verified by tools.
|
|
12
|
+
- If uncertain after checking, return exactly: "${NO_FINDINGS_SENTENCE}"
|
|
13
|
+
|
|
14
|
+
Severity:
|
|
15
|
+
- [high]: deterministic runtime/security breakage with clear path.
|
|
16
|
+
- [medium]: likely bug with strong evidence.
|
|
17
|
+
|
|
18
|
+
Output format (strict):
|
|
19
|
+
- \`- [high|medium] <title>\`
|
|
20
|
+
- \` File: <path>\`
|
|
21
|
+
- \` Line: ~<N>\`
|
|
22
|
+
- \` Why: <one concise sentence with evidence>\`
|
|
23
|
+
- If no issues: exactly "${NO_FINDINGS_SENTENCE}"`;
|
|
24
|
+
const MAIN_WEAK = `You are an AI bug-finder for a merge request.
|
|
25
|
+
Output bullets only — no preamble, no headings, no code fences, no closing remarks.
|
|
26
|
+
Return at most 3 findings.
|
|
27
|
+
|
|
28
|
+
Report a finding ONLY when ALL of these are true:
|
|
29
|
+
1. The bug is on a line added or modified in the diff (or a direct consequence of one).
|
|
30
|
+
2. You can quote the offending text from the diff or from get_file_at_ref output.
|
|
31
|
+
3. The bug causes wrong behavior, a crash, a security issue, or data loss — not a style nit.
|
|
32
|
+
If any of those is not true, do not include the finding.
|
|
33
|
+
|
|
34
|
+
When the diff is unclear, call the tools: get_file_at_ref(path, ref) or grep_repository(query, ref). Prefer the head ref.
|
|
35
|
+
|
|
36
|
+
Severity tags:
|
|
37
|
+
- [high] = will break at runtime or leak data.
|
|
38
|
+
- [medium] = strong evidence of a bug, but conditions matter.
|
|
39
|
+
|
|
40
|
+
Output format — copy spacing and case exactly, do not wrap in backticks or fences:
|
|
41
|
+
- [high|medium] <short title>
|
|
42
|
+
File: <path>
|
|
43
|
+
Line: ~<N>
|
|
44
|
+
Why: <one sentence pointing at the evidence>
|
|
45
|
+
|
|
46
|
+
Example of a valid finding:
|
|
47
|
+
- [high] Null pointer dereference in user lookup
|
|
48
|
+
File: src/user/service.ts
|
|
49
|
+
Line: ~42
|
|
50
|
+
Why: getUser() can return null but the next line calls user.id without a guard.
|
|
51
|
+
|
|
52
|
+
If you cannot meet all three conditions for any finding, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`;
|
|
53
|
+
const FILE_DEFAULT = `You are an AI reviewer for a single-file diff.
|
|
54
|
+
Find only real bugs introduced by changed lines.
|
|
55
|
+
Return at most 2 findings. Prefer no finding over a weak one.
|
|
56
|
+
|
|
57
|
+
Rules:
|
|
58
|
+
- Focus on changed lines only.
|
|
59
|
+
- Ignore style/refactor/general improvement comments.
|
|
60
|
+
- If uncertain, use tools: get_file_at_ref and grep_repository.
|
|
61
|
+
- Report only issues clearly visible in diff or verified by tools.
|
|
62
|
+
- If uncertain after checking, return exactly: "${NO_FINDINGS_SENTENCE}"
|
|
63
|
+
|
|
64
|
+
Severity:
|
|
65
|
+
- [high]: deterministic runtime/security breakage with clear path.
|
|
66
|
+
- [medium]: likely bug with strong evidence.
|
|
67
|
+
|
|
68
|
+
Output format (strict):
|
|
69
|
+
- \`- [high|medium] <title>\`
|
|
70
|
+
- \` File: <path>\`
|
|
71
|
+
- \` Line: ~<N>\`
|
|
72
|
+
- \` Why: <one concise sentence with evidence>\`
|
|
73
|
+
- If no issues: exactly "${NO_FINDINGS_SENTENCE}"`;
|
|
74
|
+
const FILE_WEAK = `You are an AI bug-finder reviewing one file's diff.
|
|
75
|
+
Output bullets only — no preamble, no headings, no code fences.
|
|
76
|
+
Return at most 2 findings.
|
|
77
|
+
|
|
78
|
+
Report a finding ONLY when ALL of these are true:
|
|
79
|
+
1. The bug is on a line added or modified in this file's diff.
|
|
80
|
+
2. You can quote the offending text from the diff or from get_file_at_ref output.
|
|
81
|
+
3. The bug causes wrong behavior, a crash, a security issue, or data loss.
|
|
82
|
+
If any of those is not true, do not include the finding.
|
|
83
|
+
|
|
84
|
+
When the diff is unclear, call the tools: get_file_at_ref(path, ref) or grep_repository(query, ref). Prefer the head ref.
|
|
85
|
+
|
|
86
|
+
Severity tags:
|
|
87
|
+
- [high] = will break at runtime or leak data.
|
|
88
|
+
- [medium] = strong evidence of a bug, but conditions matter.
|
|
89
|
+
|
|
90
|
+
Output format — copy spacing and case exactly, do not wrap in backticks or fences:
|
|
91
|
+
- [high|medium] <short title>
|
|
92
|
+
File: <path>
|
|
93
|
+
Line: ~<N>
|
|
94
|
+
Why: <one sentence pointing at the evidence>
|
|
95
|
+
|
|
96
|
+
Example of a valid finding:
|
|
97
|
+
- [medium] Off-by-one in pagination loop
|
|
98
|
+
File: src/api/list.ts
|
|
99
|
+
Line: ~88
|
|
100
|
+
Why: loop runs while i <= total but indices are 0-based, so the last item is read past the array end.
|
|
101
|
+
|
|
102
|
+
If you cannot meet all three conditions for any finding, output exactly this single line and nothing else: ${NO_FINDINGS_SENTENCE}`;
|
|
103
|
+
export function getMainSystemPrompt(profile) {
|
|
104
|
+
return profile === "weak" ? MAIN_WEAK : MAIN_DEFAULT;
|
|
113
105
|
}
|
|
114
|
-
export function
|
|
115
|
-
return profile === "weak" ?
|
|
106
|
+
export function getFileReviewSystemPrompt(profile) {
|
|
107
|
+
return profile === "weak" ? FILE_WEAK : FILE_DEFAULT;
|
|
116
108
|
}
|
|
117
|
-
/** @deprecated kept for backward compatibility — use getMainSystemLines("default"). */
|
|
118
|
-
export const MAIN_SYSTEM_LINES = MAIN_DEFAULT_LINES;
|
|
119
|
-
/** @deprecated kept for backward compatibility — use getFileReviewSystemLines("default"). */
|
|
120
|
-
export const FILE_REVIEW_SYSTEM_LINES = FILE_DEFAULT_LINES;
|
|
@@ -1,50 +1,44 @@
|
|
|
1
1
|
/** @format */
|
|
2
|
-
const
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
'{"summary":"Adds retry-with-backoff to the OpenAI client and updates the README badge.","files":[{"path":"src/openai/client.ts","verdict":"NEEDS_REVIEW","reason":"adds retry loop in API client"},{"path":"README.md","verdict":"SKIP","reason":"badge URL only"}]}',
|
|
45
|
-
];
|
|
46
|
-
export function getTriageSystemLines(profile) {
|
|
47
|
-
return profile === "weak" ? WEAK_LINES : DEFAULT_LINES;
|
|
2
|
+
const DEFAULT = `You are a senior developer triaging files in a merge request.
|
|
3
|
+
For each file, decide whether it NEEDS_REVIEW (modifies logic, functionality, security, or performance) or can be SKIPPED (cosmetic-only: formatting, comments, renaming for clarity, docs, auto-generated files).
|
|
4
|
+
|
|
5
|
+
Also produce a concise summary (2-4 sentences) of the entire merge request: what it does and which areas it touches.
|
|
6
|
+
|
|
7
|
+
Respond with a JSON object (no markdown fences, no prose before or after) in this exact schema:
|
|
8
|
+
{ "summary": "<MR summary>", "files": [{ "path": "<file path>", "verdict": "NEEDS_REVIEW" | "SKIP", "reason": "<one short sentence explaining the verdict>" }] }
|
|
9
|
+
|
|
10
|
+
The reason field is required for every file — cite what you saw in the diff (e.g. "import reordering only" or "changes Input props and validation").
|
|
11
|
+
|
|
12
|
+
Rules:
|
|
13
|
+
- When in doubt, verdict is NEEDS_REVIEW.
|
|
14
|
+
- Look beyond comments and JSDoc: if the diff changes function signatures, return types, control flow, variable assignments, or adds/removes code lines (not just comments), it is a logic change → NEEDS_REVIEW.
|
|
15
|
+
- If a file mixes doc/comment edits with actual code changes, verdict is NEEDS_REVIEW.
|
|
16
|
+
- Refactoring (splitting/merging functions, changing data structures, renaming with signature changes) is NEEDS_REVIEW.
|
|
17
|
+
- Deleted files are SKIP unless the deletion could break dependents.
|
|
18
|
+
- New files containing logic are NEEDS_REVIEW.
|
|
19
|
+
- Test files that add, remove, or change assertions or expected values are NEEDS_REVIEW.
|
|
20
|
+
- Test-only files are SKIP only if changes are purely cosmetic (formatting, comments).
|
|
21
|
+
- Config/CI/docs files are SKIP unless they modify build targets, env vars, or secrets.
|
|
22
|
+
|
|
23
|
+
Example output (copy this style exactly — single JSON object, no fences, no preamble):
|
|
24
|
+
{"summary":"Adds retry-with-backoff to the OpenAI client and updates the README badge.","files":[{"path":"src/openai/client.ts","verdict":"NEEDS_REVIEW","reason":"adds retry loop and error handling in API client"},{"path":"README.md","verdict":"SKIP","reason":"badge URL change only"}]}`;
|
|
25
|
+
const WEAK = `You triage merge-request files. Output one JSON object only — no prose, no markdown, no code fences, no preamble.
|
|
26
|
+
|
|
27
|
+
Schema (use exactly these keys):
|
|
28
|
+
{"summary":"<2-3 sentences about the MR>","files":[{"path":"<file path>","verdict":"NEEDS_REVIEW" | "SKIP","reason":"<one short sentence>"}]}
|
|
29
|
+
|
|
30
|
+
Mark a file NEEDS_REVIEW when ANY of these is true:
|
|
31
|
+
- the diff adds or removes non-comment lines,
|
|
32
|
+
- the diff changes a function signature, return type, control flow, or variable value,
|
|
33
|
+
- the file is new and contains code,
|
|
34
|
+
- a test changes an assertion or expected value,
|
|
35
|
+
- a config/CI file changes build targets, env vars, or secrets.
|
|
36
|
+
|
|
37
|
+
Mark a file SKIP only when the diff is purely cosmetic: whitespace, comments, JSDoc, plain docs, auto-generated files, or deletions that cannot affect callers.
|
|
38
|
+
When in doubt, choose NEEDS_REVIEW.
|
|
39
|
+
|
|
40
|
+
Example output (copy this exactly — one line, one JSON object, nothing else):
|
|
41
|
+
{"summary":"Adds retry-with-backoff to the OpenAI client and updates the README badge.","files":[{"path":"src/openai/client.ts","verdict":"NEEDS_REVIEW","reason":"adds retry loop in API client"},{"path":"README.md","verdict":"SKIP","reason":"badge URL only"}]}`;
|
|
42
|
+
export function getTriageSystemPrompt(profile) {
|
|
43
|
+
return profile === "weak" ? WEAK : DEFAULT;
|
|
48
44
|
}
|
|
49
|
-
/** @deprecated kept for backward compatibility — use getTriageSystemLines("default"). */
|
|
50
|
-
export const TRIAGE_SYSTEM_LINES = DEFAULT_LINES;
|