Overview
Transform messy, unstructured input—pasted error logs, voice dictation, screenshots, support notes—into clean, actionable GitHub issue markdown files. Each issue includes a one-line summary, environment, numbered reproduction steps, expected vs. actual behavior, error details, visual evidence references, severity-rated impact, and additional context.
When to Use
Use this skill when the user provides unstructured bug input and wants a structured GitHub issue. Trigger keywords and signals:
- Pasted stack traces, error messages, or HTTP status codes
- Voice dictation describing a bug ("so I was trying to… and it just failed")
- Screenshots or GIFs of a failure
- Support tickets, Slack threads, or rough notes needing triage
- Explicit requests: "create an issue", "file a bug", "log this on GitHub", "write up this bug"
Do not use this skill for feature requests, RFCs, or design docs—only bug/defect reports.
Prerequisites
- A local repo root where the
/issues/ directory will be created or already exists.
- On Windows host (PowerShell), the repo root is typically
~\agent-skills\library\github-issue-creator or the active project folder.
- If the user references a screenshot or GIF, confirm the file path or attachment name before referencing it inline.
- If required inputs are missing (product/service, at least one repro step, or expected behavior), stop and ask for clarification.
Procedure
1. Gather and parse raw input
- Collect all raw material from the user: error text, voice dictation transcript, screenshot paths, conversation context.
- Extract facts buried in casual language. Voice notes often contain the real sequence of events behind informal phrasing.
- Infer missing context from conversation or memory only when the user explicitly references it ("same project", "the dashboard", "that error from before"). Otherwise, use placeholders.
2. Determine environment and impact
- Identify Product/Service, Region/Version, and Browser/OS from the input or conversation context.
- If unknown, insert
[UNKNOWN] or [REGION] placeholders—never fabricate.
- Assign severity using the rubric below.
3. Assign severity
| Severity |
Criteria |
| Critical |
Service down, data loss, security issue |
| High |
Major feature broken, no workaround |
| Medium |
Feature impaired, workaround exists |
| Low |
Minor inconvenience, cosmetic |
4. Write the issue file
- Create the
/issues/ directory at the repo root if it does not exist.New-Item -ItemType Directory -Path ".\issues" -Force
- Name the file using the convention
YYYY-MM-DD-short-description.md.$date = Get-Date -Format "yyyy-MM-dd"
$slug = "agent-deployment-silent-failure"
New-Item -ItemType File -Path ".\issues\$date-$slug.md" -Force
- Write the issue using the exact output template below.
5. Output Template
## Summary
[One-line description of the issue]
## Environment
- **Product/Service**:
- **Region/Version**:
- **Browser/OS**: (if relevant)
## Reproduction Steps
1. [Step]
2. [Step]
3. [Step]
## Expected Behavior
[What should happen]
## Actual Behavior
[What actually happens]
## Error Details
[Error message/code if applicable]
## Visual Evidence
[Reference to attached screenshots/GIFs]
## Impact
[Severity: Critical/High/Medium/Low + brief explanation]
## Additional Context
[Any other relevant details]
6. Placeholder sensitive data
Replace any potentially sensitive values with bracketed placeholders:
[PROJECT_NAME]
[USER_ID]
[API_KEY]
[TENANT_ID]
[EMAIL]
Never include live secrets, tokens, or credentials in the issue file.
7. Reference attachments inline
For screenshots or GIFs, reference them inline using the format:

If the file path is unknown, write [See attached: filename.png] and flag it for the user to attach manually.
Examples
Example A — Voice dictation input
Input:
so I was trying to deploy the agent and it just failed silently no error nothing the workflow ran but then poof gone from the list had to refresh and try again three times
Output file: issues/2026-02-27-agent-deployment-silent-failure.md
## Summary
Agent deployment fails silently - no error displayed, agent disappears from list
## Environment
- **Product/Service**: Azure AI Foundry
- **Region/Version**: westus2
## Reproduction Steps
1. Navigate to agent deployment
2. Configure and deploy agent
3. Observe workflow completes
4. Check agent list
## Expected Behavior
Agent appears in list with deployment status, errors shown if deployment fails
## Actual Behavior
Agent disappears from list. No error message. Requires page refresh and retry.
## Impact
**High** - Blocks agent deployment workflow, no feedback on failure cause
## Additional Context
Required 3 retry attempts before successful deployment
Example B — Error paste input
Input:
Error: PERMISSION_DENIED when publishing to Teams channel. Code: 403. Was working yesterday.
Output file: issues/2026-02-27-teams-publish-permission-denied.md
## Summary
403 PERMISSION_DENIED error when publishing to Teams channel
## Environment
- **Product/Service**: Copilot Studio → Teams integration
- **Region/Version**: [REGION]
## Reproduction Steps
1. Configure agent for Teams channel
2. Attempt to publish
## Expected Behavior
Agent publishes successfully to Teams channel
## Actual Behavior
Returns `PERMISSION_DENIED` with code 403
## Error Details
Error: PERMISSION_DENIED
Code: 403
## Impact
**High** - Blocks Teams integration, regression from previous working state
## Additional Context
Was working yesterday - possible permission/config change or service regression
Pitfalls
- Fabricating environment details: Never invent region, version, or browser/OS. Use
[UNKNOWN] placeholders and ask the user.
- Missing repro steps: If the input does not contain enough detail for at least 2 numbered repro steps, stop and ask the user for clarification.
- Including live secrets: Error logs sometimes contain API keys, tokens, or connection strings. Always scrub these to
[REDACTED] or the appropriate placeholder before writing the file.
- Vague severity: Do not default to "Medium" when unsure. Map severity strictly to the rubric. If impact is unclear, ask the user.
- Wrong file location: Issues must go in
/issues/ at the repo root, not in a subfolder or the user's home directory. On Windows, confirm the working directory with Get-Location before writing.
- Inconsistent naming: Always use
YYYY-MM-DD-short-description.md. Do not use spaces or uppercase in the slug portion.
- Over-writing existing issues: Use
New-Item without -Force on the file if you want to avoid clobbering; check first with Test-Path.
- Treating output as validated: The generated issue is a draft. It does not substitute for environment-specific validation, testing, or expert review.
Verification
After writing the issue file, verify:
File exists and is named correctly:
Get-ChildItem .\issues\*.md | Select-Object Name
Confirm the filename matches YYYY-MM-DD-short-description.md.
All required sections are present:
$content = Get-Content ".\issues\YYYY-MM-DD-short-description.md" -Raw
$sections = @("## Summary","## Environment","## Reproduction Steps","## Expected Behavior","## Actual Behavior","## Error Details","## Visual Evidence","## Impact","## Additional Context")
foreach ($s in $sections) { if ($content -notmatch [regex]::Escape($s)) { Write-Host "MISSING: $s" } }
No live secrets leaked:
$content = Get-Content ".\issues\YYYY-MM-DD-short-description.md" -Raw
if ($content -match '(?i)(sk-[a-z0-9]{20,}|Bearer\s+[A-Za-z0-9\.\-_]+|password\s*[:=]\s*\S+)') { Write-Host "WARNING: Possible secret detected" }
Repro steps are numbered: Confirm at least 2 numbered items exist under ## Reproduction Steps.
Severity is set: Confirm the Impact section contains one of Critical, High, Medium, or Low.
Related skills
- github-pr-template: For structuring pull request descriptions once a fix is ready.
- bug-triage-labeler: For assigning labels and priority after issue creation.
1---
2name: github-issue-creator
3description: "Turns pasted errors, screenshots, voice notes, and rough tickets into dated GitHub issue markdown covering environment, numbered repro, expected vs actual, evidence, and severity. Trigger when the user asks to file, create, or log a GitHub issue from unstructured bug input. Do not use for feature RFCs, design docs, or pull-request writeups."
4version: 1.0.1
5risk: unknown
6source: community
7date_added: "2026-02-27"
8metadata:
9 source: https://skillmd.com/skills/kayforkind/github-issue-creator
10---
11
12## Overview
13
14Transform messy, unstructured input—pasted error logs, voice dictation, screenshots, support notes—into clean, actionable GitHub issue markdown files. Each issue includes a one-line summary, environment, numbered reproduction steps, expected vs. actual behavior, error details, visual evidence references, severity-rated impact, and additional context.
15
16## When to Use
17
18Use this skill when the user provides unstructured bug input and wants a structured GitHub issue. Trigger keywords and signals:
19
20- Pasted stack traces, error messages, or HTTP status codes
21- Voice dictation describing a bug ("so I was trying to… and it just failed")
22- Screenshots or GIFs of a failure
23- Support tickets, Slack threads, or rough notes needing triage
24- Explicit requests: "create an issue", "file a bug", "log this on GitHub", "write up this bug"
25
26Do **not** use this skill for feature requests, RFCs, or design docs—only bug/defect reports.
27
28## Prerequisites
29
301. A local repo root where the `/issues/` directory will be created or already exists.
312. On Windows host (PowerShell), the repo root is typically `~\agent-skills\library\github-issue-creator` or the active project folder.
323. If the user references a screenshot or GIF, confirm the file path or attachment name before referencing it inline.
334. If required inputs are missing (product/service, at least one repro step, or expected behavior), stop and ask for clarification.
34
35## Procedure
36
37### 1. Gather and parse raw input
38
391. Collect all raw material from the user: error text, voice dictation transcript, screenshot paths, conversation context.
402. Extract facts buried in casual language. Voice notes often contain the real sequence of events behind informal phrasing.
413. Infer missing context from conversation or memory only when the user explicitly references it ("same project", "the dashboard", "that error from before"). Otherwise, use placeholders.
42
43### 2. Determine environment and impact
44
451. Identify **Product/Service**, **Region/Version**, and **Browser/OS** from the input or conversation context.
462. If unknown, insert `[UNKNOWN]` or `[REGION]` placeholders—never fabricate.
473. Assign severity using the rubric below.
48
49### 3. Assign severity
50
51| Severity | Criteria |
52|----------|----------|
53| **Critical** | Service down, data loss, security issue |
54| **High** | Major feature broken, no workaround |
55| **Medium** | Feature impaired, workaround exists |
56| **Low** | Minor inconvenience, cosmetic |
57
58### 4. Write the issue file
59
601. Create the `/issues/` directory at the repo root if it does not exist.
61 ```powershell
62 New-Item -ItemType Directory -Path ".\issues" -Force
63 ```
642. Name the file using the convention `YYYY-MM-DD-short-description.md`.
65 ```powershell
66 $date = Get-Date -Format "yyyy-MM-dd"
67 $slug = "agent-deployment-silent-failure"
68 New-Item -ItemType File -Path ".\issues\$date-$slug.md" -Force
69 ```
703. Write the issue using the exact output template below.
71
72### 5. Output Template
73
74```markdown
75## Summary
76[One-line description of the issue]
77
78## Environment
79- **Product/Service**:
80- **Region/Version**:
81- **Browser/OS**: (if relevant)
82
83## Reproduction Steps
841. [Step]
852. [Step]
863. [Step]
87
88## Expected Behavior
89[What should happen]
90
91## Actual Behavior
92[What actually happens]
93
94## Error Details
95```
96[Error message/code if applicable]
97```
98
99## Visual Evidence
100[Reference to attached screenshots/GIFs]
101
102## Impact
103[Severity: Critical/High/Medium/Low + brief explanation]
104
105## Additional Context
106[Any other relevant details]
107```
108
109### 6. Placeholder sensitive data
110
111Replace any potentially sensitive values with bracketed placeholders:
112
113- `[PROJECT_NAME]`
114- `[USER_ID]`
115- `[API_KEY]`
116- `[TENANT_ID]`
117- `[EMAIL]`
118
119Never include live secrets, tokens, or credentials in the issue file.
120
121### 7. Reference attachments inline
122
123For screenshots or GIFs, reference them inline using the format:
124
125```
126
127```
128
129If the file path is unknown, write `[See attached: filename.png]` and flag it for the user to attach manually.
130
131## Examples
132
133### Example A — Voice dictation input
134
135**Input**:
136> so I was trying to deploy the agent and it just failed silently no error nothing the workflow ran but then poof gone from the list had to refresh and try again three times
137
138**Output file**: `issues/2026-02-27-agent-deployment-silent-failure.md`
139
140```markdown
141## Summary
142Agent deployment fails silently - no error displayed, agent disappears from list
143
144## Environment
145- **Product/Service**: Azure AI Foundry
146- **Region/Version**: westus2
147
148## Reproduction Steps
1491. Navigate to agent deployment
1502. Configure and deploy agent
1513. Observe workflow completes
1524. Check agent list
153
154## Expected Behavior
155Agent appears in list with deployment status, errors shown if deployment fails
156
157## Actual Behavior
158Agent disappears from list. No error message. Requires page refresh and retry.
159
160## Impact
161**High** - Blocks agent deployment workflow, no feedback on failure cause
162
163## Additional Context
164Required 3 retry attempts before successful deployment
165```
166
167### Example B — Error paste input
168
169**Input**:
170> Error: PERMISSION_DENIED when publishing to Teams channel. Code: 403. Was working yesterday.
171
172**Output file**: `issues/2026-02-27-teams-publish-permission-denied.md`
173
174```markdown
175## Summary
176403 PERMISSION_DENIED error when publishing to Teams channel
177
178## Environment
179- **Product/Service**: Copilot Studio → Teams integration
180- **Region/Version**: [REGION]
181
182## Reproduction Steps
1831. Configure agent for Teams channel
1842. Attempt to publish
185
186## Expected Behavior
187Agent publishes successfully to Teams channel
188
189## Actual Behavior
190Returns `PERMISSION_DENIED` with code 403
191
192## Error Details
193```
194Error: PERMISSION_DENIED
195Code: 403
196```
197
198## Impact
199**High** - Blocks Teams integration, regression from previous working state
200
201## Additional Context
202Was working yesterday - possible permission/config change or service regression
203```
204
205## Pitfalls
206
2071. **Fabricating environment details**: Never invent region, version, or browser/OS. Use `[UNKNOWN]` placeholders and ask the user.
2082. **Missing repro steps**: If the input does not contain enough detail for at least 2 numbered repro steps, stop and ask the user for clarification.
2093. **Including live secrets**: Error logs sometimes contain API keys, tokens, or connection strings. Always scrub these to `[REDACTED]` or the appropriate placeholder before writing the file.
2104. **Vague severity**: Do not default to "Medium" when unsure. Map severity strictly to the rubric. If impact is unclear, ask the user.
2115. **Wrong file location**: Issues must go in `/issues/` at the repo root, not in a subfolder or the user's home directory. On Windows, confirm the working directory with `Get-Location` before writing.
2126. **Inconsistent naming**: Always use `YYYY-MM-DD-short-description.md`. Do not use spaces or uppercase in the slug portion.
2137. **Over-writing existing issues**: Use `New-Item` without `-Force` on the file if you want to avoid clobbering; check first with `Test-Path`.
2148. **Treating output as validated**: The generated issue is a draft. It does not substitute for environment-specific validation, testing, or expert review.
215
216## Verification
217
218After writing the issue file, verify:
219
2201. **File exists and is named correctly**:
221 ```powershell
222 Get-ChildItem .\issues\*.md | Select-Object Name
223 ```
224 Confirm the filename matches `YYYY-MM-DD-short-description.md`.
225
2262. **All required sections are present**:
227 ```powershell
228 $content = Get-Content ".\issues\YYYY-MM-DD-short-description.md" -Raw
229 $sections = @("## Summary","## Environment","## Reproduction Steps","## Expected Behavior","## Actual Behavior","## Error Details","## Visual Evidence","## Impact","## Additional Context")
230 foreach ($s in $sections) { if ($content -notmatch [regex]::Escape($s)) { Write-Host "MISSING: $s" } }
231 ```
232
2333. **No live secrets leaked**:
234 ```powershell
235 $content = Get-Content ".\issues\YYYY-MM-DD-short-description.md" -Raw
236 if ($content -match '(?i)(sk-[a-z0-9]{20,}|Bearer\s+[A-Za-z0-9\.\-_]+|password\s*[:=]\s*\S+)') { Write-Host "WARNING: Possible secret detected" }
237 ```
238
2394. **Repro steps are numbered**: Confirm at least 2 numbered items exist under `## Reproduction Steps`.
240
2415. **Severity is set**: Confirm the Impact section contains one of `Critical`, `High`, `Medium`, or `Low`.
242
243## Related skills
244
245- **github-pr-template**: For structuring pull request descriptions once a fix is ready.
246- **bug-triage-labeler**: For assigning labels and priority after issue creation.