You MUST use this skill before claiming any work is complete, any test passes, or any fix works — especially before arc close, before telling the user "done", or when the user asks "does it work?", "did the tests pass?", "is it fixed?". Requires fresh verification evidence (not cached results). Always prefer this over ad-hoc verification when the project uses arc issue tracking.
From arcnpx claudepluginhub sentiolabs/arc --plugin arcThis skill uses the workspace's default tool permissions.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Guides agentic engineering workflows: eval-first loops, 15-min task decomposition, model routing (Haiku/Sonnet/Opus), AI code reviews, and cost tracking.
Run proof commands and read their output before making any completion claim.
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE.
"It should work" is not evidence. "Tests pass" without output is not evidence. Satisfaction expressed before running the proof command is a red flag.
Create a TodoWrite checklist with these steps:
What command proves the claim? Examples:
make test or go test ./...make buildarc show <id> (check status)Execute the full command. At minimum, run all tests affected by the change; running the full suite is safer. Not a subset from memory. Not "I ran it earlier."
Fresh. Complete. Now.
Read the FULL output. Not just the last line. Check:
Does the output actually confirm the claim?
Make the claim. Reference the evidence:
go test ./... shows 47 passed, 0 failed"make build exits 0, binary at ./bin/arc"You are skipping verification if you:
# ONLY after verification passes:
arc close <id> -r "Verified: <evidence summary>"
If verification fails, do NOT close the issue. Instead:
implement to fix the failuredebug if the failure is unexpectedfinish (session end) or back to implement (next task)skills/arc/_formatting.md