vk-fix-source
Skill1 starsChanged 4 months agoApache-2.0View on GitHub
⚡⚡ Fix issues from a failure source (CI url | log file | test suite)
In this repository77 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- archifySkill
- banner-designSkill
- brandSkill
- chrome-devtoolsSkill
- code-reviewSkill
- computer-useSkill
- context-engine-debuggingSkill
- context-engine-exploringSkill
- context-engine-guideSkill
- context-engine-syncSkill
- debuggingSkill
- designSkill
- design-systemSkill
- docs-seekerSkill
- docxSkill
- pdfSkill
- pptxSkill
- xlsxSkill
- kubernetesSkill
- planningSkill
- playwrightSkill
- repomixSkill
- researchSkill
- slidesSkill
- ui-ux-pro-maxSkill
- vk-askSkill
- vk-bootstrap-autoSkill
- vk-brainstormSkill
- vk-cloneSkill
- vk-contentSkill
- vk-cook-routerSkill
- vk-cookSkill
- vk-debugSkill
- vk-design-from-visualSkill
- vk-designSkill
- vk-docsSkill
- vk-docs-summarizeSkill
- vk-fix-fastSkill
- vk-fix-hardSkill
- vk-fixSkill
- vk-fix-typesSkill
- vk-fix-uiSkill
- vk-git-cmSkill
- vk-git-cpSkill
- vk-git-mergeSkill
- vk-git-prSkill
- vk-git-tagSkill
- vk-goalSkill
- vk-helpSkill
- vk-hoh-devSkill
- vk-hoh-planSkill
- vk-hoh-qaSkill
- vk-hoh-runSkill
- vk-integrate-polarSkill
- vk-integrate-sepaySkill
- vk-journalSkill
- vk-learnSkill
- vk-plan-archiveSkill
- vk-plan-fastSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 31
Use tester to verify the fix works, then report back to the main agent.
- Very specific: line 32
Use code-reviewer to review the changes against the requirements, then report back.
- Very specific: line 38
url: if gh is not available, instruct the user to install and authorize the GitHub CLI first.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Fix Source
Diagnose from a concrete failure source, then fix. One workflow, three input kinds.
<input>$ARGUMENTS</input>
Source auto-detection
Inspect <input> — a flag/keyword is honored only if its literal token appears:
| Source | Detected when | Failure evidence comes from |
| url | <input> contains a GitHub Actions / CI run URL | debugger reads the CI logs via gh |
| file | <input> names a log file, or ./logs.txt exists and no URL/suite is named | debugger reads that log file — if missing, reproduce the issue and pipe output to ./logs.txt first |
| suite | <input> names tests / a test target, or asks to run the suite | tester compiles, fixes syntax errors, runs the tests — its ACTUAL output |
Ambiguous → ask one clarifying question. --plan-only → stop after step 2 and hand the plan to the user (≥2 approaches with trade-offs + a recommendation); ask for confirmation before any implementation.
Workflow
- Parallel wave (per
workflows/parallel-execution.md): the source's evidence producer (debuggerfor url/file,testerfor suite) +scout(locates the code named in<input>) run concurrently — scout needs neither diagnosis nor test output. Both report back.- suite only:
debuggercan NOT join this wave — it genuinely needstester's actual failure output (data dependency, §2.3) — so it runs serially right after, root-causing from that real output.
- suite only:
plannercreates an implementation plan from the reports (serial, needs all of them).- Implement the fix based on the reports and the plan.
- Use
testerto verify the fix works, then report back to the main agent. - Use
code-reviewerto review the changes against the requirements, then report back. - Issues or failed tests → repeat from step 1. Max 3 cycles — if still failing after 3, STOP and report (likely wrong approach or an environment issue); never loop indefinitely.
- Closing: see
workflows/fix-report-contract.md(§Final Report).
Notes
- url: if
ghis not available, instruct the user to install and authorize the GitHub CLI first. - IMPORTANT: Analyze the skills catalog and activate the skills needed for the task during the process.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.