[READ-ONLY] Mirror of https://github.com/danielroe/postcss-capsize. PostCSS plugin to inject Capsize font metrics
ci: add agent-scan workflow to flag bot-authored PRs
Daniel Roe
(Jun 24, 2026, 1:43 PM +0100)
14ac60ed
89784d3d
+115
+115
.github/workflows/agent-scan.yml
+115
.github/workflows/agent-scan.yml
···
1
+
name: agent-scan
2
+
3
+
on:
4
+
# zizmor: ignore[dangerous-triggers] - DO NOT add action/checkout in this workflow as it uses pull_request_target
5
+
pull_request_target:
6
+
types:
7
+
- opened
8
+
- reopened
9
+
10
+
concurrency:
11
+
group: agent-scan-${{ github.event.pull_request.number }}
12
+
cancel-in-progress: true
13
+
14
+
permissions:
15
+
issues: write
16
+
pull-requests: write
17
+
18
+
jobs:
19
+
agentscan:
20
+
runs-on: ubuntu-latest
21
+
steps:
22
+
- name: AgentScan
23
+
id: agentscan
24
+
uses: MatteoGabriele/agentscan-action@ccb50da057e2ac7fcdb123db3110f00af97f0fef # v1.14.0
25
+
with:
26
+
github-token: ${{ secrets.GITHUB_TOKEN }}
27
+
skip-members: 'dependabot[bot],renovate[bot]'
28
+
agent-scan-comment: false
29
+
- name: Handle flagged PR
30
+
if: contains(fromJSON('["automation","mixed"]'), steps.agentscan.outputs.classification) || steps.agentscan.outputs.community-flagged == 'true'
31
+
env:
32
+
CLASSIFICATION: ${{ steps.agentscan.outputs.classification }}
33
+
COMMUNITY_FLAGGED: ${{ steps.agentscan.outputs.community-flagged }}
34
+
uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
35
+
with:
36
+
script: |
37
+
const prNumber = context.payload.pull_request.number;
38
+
const classification = process.env.CLASSIFICATION;
39
+
const communityFlagged = process.env.COMMUNITY_FLAGGED === 'true';
40
+
const shouldClose = classification === 'automation' || communityFlagged;
41
+
42
+
const issue = context.payload.pull_request
43
+
const labels = issue.labels?.map(l => l.name) || []
44
+
45
+
if (!labels.includes('possible bot')) {
46
+
await github.rest.issues.addLabels({
47
+
owner: context.repo.owner,
48
+
repo: context.repo.repo,
49
+
issue_number: prNumber,
50
+
labels: ['possible bot'],
51
+
})
52
+
}
53
+
54
+
const comments = await github.paginate(github.rest.issues.listComments, {
55
+
owner: context.repo.owner,
56
+
repo: context.repo.repo,
57
+
issue_number: prNumber,
58
+
per_page: 100,
59
+
})
60
+
61
+
const alreadyCommented = comments.some(
62
+
c => c.user.type === 'Bot' && c.body.includes('AI-assisted contribution guidelines')
63
+
)
64
+
65
+
if (!alreadyCommented) {
66
+
const closingNote = shouldClose
67
+
? "We're closing this for now as the account looks automated. If we got that wrong, please just reopen the PR and we'll take another look."
68
+
: 'If this was flagged in error, we apologise! 😳 Just let us know. 🙏'
69
+
70
+
await github.rest.issues.createComment({
71
+
owner: context.repo.owner,
72
+
repo: context.repo.repo,
73
+
issue_number: prNumber,
74
+
body: [
75
+
"We've flagged this as a potential contribution without a human behind it. We welcome the thoughtful use of AI tools when contributing, but ask all contributors to follow [two core principles](https://roe.dev/blog/using-ai-in-open-source):",
76
+
'',
77
+
'1. **Never let an LLM speak for you** - all comments, issues, and PR descriptions should be written in your own words, reflecting your own understanding.',
78
+
'2. **Never let an LLM think for you** - only submit contributions you fully understand and can explain.',
79
+
'',
80
+
'Please review these AI-assisted contribution guidelines and update this contribution if needed.',
81
+
'',
82
+
closingNote,
83
+
].join('\n'),
84
+
})
85
+
} else {
86
+
core.info('Possible-bot comment already exists - skipping comment.')
87
+
}
88
+
89
+
if (shouldClose && issue.state === 'open' && !alreadyCommented) {
90
+
await github.rest.pulls.update({
91
+
owner: context.repo.owner,
92
+
repo: context.repo.repo,
93
+
pull_number: prNumber,
94
+
state: 'closed',
95
+
title: '🚨 unwelcome pr from bot 🚨',
96
+
})
97
+
}
98
+
99
+
const actionTaken = [
100
+
'Added `possible bot` label',
101
+
alreadyCommented ? null : 'posted policy comment',
102
+
shouldClose && !alreadyCommented ? 'closed PR' : null,
103
+
].filter(Boolean).join(', ')
104
+
105
+
core.summary
106
+
.addHeading('AgentScan: Possible Bot Flag', 2)
107
+
.addTable([
108
+
[{ data: 'Property', header: true }, { data: 'Value', header: true }],
109
+
['Pull Request', `#${prNumber}`],
110
+
['Classification', classification],
111
+
['Community flagged', String(communityFlagged)],
112
+
['Action', actionTaken || 'No action (already handled)'],
113
+
])
114
+
115
+
await core.summary.write()