Menü
Use when building an automated test → issue → fix loop with Claude Code and GitHub issues — overnight auto-fixing, regression loops, self-healing CI.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
Anthropic API rate limit handling - retry logic, backoff, throttling for batch workloads against Claude models
Use when creating, editing, publishing, or deleting posts on Cyril's Workshop blog or the steponnopets.net devblog.
Use when writing or contributing a boofuzz network-protocol fuzzer in this repo — layout, formatting rules, and reading results.
Use when a task needs real-time control of a connected browser via the Browser Bridge Broker — submit JS jobs over HTTP that browsers eval and return.
Use when training a character LoRA (Chroma/Flux or Pony/SDXL) on a RunPod GPU and wiring it into the ComfyUI + pony_web render stack.
Use when writing a new skill or revising an existing one — format, trigger descriptions, and what content belongs in a skill.
| name | auto-testing |
| description | Use when building an automated test → issue → fix loop with Claude Code and GitHub issues — overnight auto-fixing, regression loops, self-healing CI. |
Pattern for a Test → Fail → Issue → Fix → Repeat cycle that runs until tests pass, developed on the iSCSI project.
LLMs rationalize failures and close issues prematurely ("the timeout is infrastructure, not my code"). Never put "close the issue if tests pass" in a prompt — the shell runs the tests and enforces tests pass && close:
./run-tests.sh full
if [ $? -eq 0 ]; then
gh issue close "$ISSUE_NUM" --comment "Tests passed"
else
gh issue comment "$ISSUE_NUM" --body "Tests failed. Leaving open."
fi
Separation of concerns: Claude writes code; the shell runs tests, checks exit codes, and manages issues. Prompts must explicitly say: do NOT run tests, do NOT close the issue — and that the tests are CORRECT and must not be modified (Claude will otherwise "fix" tests to make them pass).
Check for open issues BEFORE running tests. If an issue is open, fix it; only run tests when nothing is open. This prevents duplicate issues, issue spam, and wasted test runs.
Before each retry, commit the previous failed attempt to an auto-fix-wip/issue-N branch and reference it in the next iteration's prompt, so approaches aren't repeated.
Full test-runner, issue-fixer/implementer, and auto-fix-loop scripts: see reference-templates.md.