ds-verify-loop — Claude Code skill: checks and install
The ds-verify-loop Claude Code skill runs a real check after code changes and uses its output to guide fixes. The loop ends when the check passes or reaches its attempt limit.
MIT 589 installs ★ 320 updated 2026-10-28
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a claude-codeWhat ds-verify-loop does
ds-verify-loop picks a check before editing, such as one test, a type check, a build, or a run of the app. It can run the command first to establish the baseline. After a change, it runs that same command, reads the full error output, and targets the cause rather than a symptom.
Each iteration changes one thing. Once the narrow check passes, the skill runs a wider check and asks for the final command output in the reply or in PROGRESS.md. After five failed iterations, it stops and switches to ds-unstick; the same error after two different fixes triggers an earlier stop. Tests and lint rules must not be weakened to get a pass.
What it helps with
- check a code change with a focused test or build
- trace an error to its cause using command output
- rerun the same check after a fix
- catch regressions with a wider check
- show the passing command output as evidence
Example prompts for the agent
- Run the focused test for my changes with
/ds-verify-loopand fix the cause of any failure. - Check this change with the type checker using
/ds-verify-loop; include the final output. - Run the build after my edits with
/ds-verify-loopwithout weakening tests.
How to install ds-verify-loop
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a claude-code2Manually: put the files in place
~/.claude/skills/ds-verify-loop/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/ds-verify-loop/
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a codex2Manually: put the files in place
.agents/skills/ds-verify-loop/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a cursor2Manually: put the files in place
.cursor/skills/ds-verify-loop/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install aicodeking/deepseek-stack skills/ds-verify-loop2Manually: put the files in place
.github/skills/ds-verify-loop/SKILL.md1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a gemini-cli2Platform CLI
gemini skills install https://github.com/aicodeking/deepseek-stack --path skills/ds-verify-loopManually: put the files in place
.devin/skills/ds-verify-loop/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about ds-verify-loop
How do I install ds-verify-loop in Claude Code?
Run npx skills add aicodeking/deepseek-stack --skill ds-verify-loop -a claude-code. The skill file is at ~/.claude/skills/ds-verify-loop/SKILL.md.
What does ds-verify-loop do?
It runs a real code check, uses the reported errors to guide a fix, then repeats the same check. After a focused check passes, it runs a broader one and asks for the final output.
When does ds-verify-loop stop checking?
It stops after five failed iterations and switches to ds-unstick. If the same error returns after two different fixes, it stops sooner.
Source text
SKILL.md · MIT
--- name: ds-verify-loop description: "Run edit, real check, read the actual error, fix the root cause, rerun, until the check passes or the cap is hit. Use when you change code that has tests, a type checker, a build, or a runnable app, or before you claim any fix works. Invoke as /ds-verify-loop." --- # ds-verify-loop Prove a change works by running the real check and fixing what it reports, not what you expect it to report. ## Steps 1. Pick the check before you edit: the narrowest real command that proves the change, such as one test file, a typecheck, a build, or a run of the app with one request. Write the command down. 2. Run it once before the edit if you can, so you know the baseline (already failing, already passing, or broken for another reason). 3. Make the smallest edit that should satisfy the check. 4. Run the exact command. Read the full error output, including the first error, the file and line, and the stack frame in your code. Do not skim to the last line. 5. State the root cause in one sentence, based on the output. If you cannot, add one print or log line, or read the failing code, and rerun. Do not guess. 6. Fix the cause, not the symptom. Do not add a special case for the one input that the test uses. 7. Rerun the same command. Count this as one iteration. 8. When the narrow check passes, run the wider one once (the full test suite, the build, or a lint) to catch breakage elsewhere. 9. Paste the final passing output (the command and its last lines) in your reply or in PROGRESS.md. ## Stop when - The check passes and you have pasted the output. A claim without output does not count. - You reach 5 iterations without passing. Stop editing, report the last error, and switch to ds-unstick. - The same error appears twice after two different fixes. Switch to ds-unstick now. Do not wait for iteration 5. ## Rules - Never weaken, skip, delete or loosen a test, assertion, type or lint rule to make the check pass. If the test is wrong, say why and ask first. - Never mock away the thing under test. - Run the real command every time. Do not reason about what it would print. - Change one thing per iteration so that you know what fixed it. - If no check exists, write the smallest one that fails without your change, then use the loop.
© aicodeking · MIT · aicodeking/deepseek-stack