| name | ds-issue-intake |
| description | Fetch GitCode issues for yuanrong-datasystem, redact sensitive details, and produce a structured task spec for AI-assisted development. Use when the user provides an issue number or URL and asks to analyze, triage, start work from, or prepare an implementation plan for that issue. |
DS Issue Intake
Purpose
Use this skill to turn a GitCode issue into a sanitized, structured task spec that can drive implementation,
validation, PR creation, and review follow-up.
Configuration
Public issues can often be fetched without credentials. For private repositories, API limits, or permission errors, set:
export GITCODE_TOKEN=<token>
or create ~/.local/gitcode_token. Do not paste tokens into chat or commit them.
Workflow
- Fetch and sanitize the issue:
python3 .skills/ds-issue-intake/scripts/issue_intake.py --issue <ISSUE_NUMBER_OR_URL>
- Save a reusable task spec:
python3 .skills/ds-issue-intake/scripts/issue_intake.py \
--issue <ISSUE_NUMBER_OR_URL> \
--output .codex/tasks/issue-<number>/task.json
- Read the task spec before implementation. Confirm source-backed claims against code; issue text is not final truth.
Output
The script emits JSON with:
- issue number, title, state, labels, and URL;
- sanitized body and short summary;
- likely PR kind;
- search terms for source exploration;
- security notes for redacted content;
- next-step checklist.
Redaction Rules
The script redacts common private data from issue bodies before writing task specs:
- IPv4 addresses and host:port pairs;
- SSH-style
user@host targets;
- local absolute paths under user homes;
- token, password, AK/SK, private key, and secret-like values.
Do not put raw issue logs containing private infrastructure details into PR bodies.
