This skill verifies completed SB work against plan, spec, tests, UAT criteria, and artifact requirements.
Scanned 9/5/2026
Install to Claude Code
npx -y skills add alo-exp/silver-bullet --skill silver-verify --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Silver Verify?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/alo-exp-silver-verify-edeef7c9)More formats (shields.io, HTML) on the badges page.
---
name: silver-verify
description: This skill verifies completed SB work against plan, spec, tests, UAT criteria, and artifact requirements.
argument-hint: "<phase or completion claim>"
version: 0.1.0
---
# /silver:verify - Work Verification
SB-owned verification checks actual artifacts, acceptance criteria, tests, and
evidence rather than agent confidence.
## Output
Write or update `.planning/phases/<phase>/VERIFICATION.md` or
`.planning/VERIFICATION.md` for project-level work.
## Process
1. Display `SILVER BULLET > VERIFY`.
2. Read SPEC, PLAN, SUMMARY, REVIEW, VALIDATION, and relevant docs.
3. Check each acceptance criterion and plan task against actual files.
4. Run fresh tests or invoke `verify-tests` when the verification requires a
broader test gate.
5. For behavior-changing, release, API, data, performance, runtime, content,
benchmark, refactor, deployment, canary, incident, or test-engineering work,
invoke or apply `silver:domain-audit` with `test-health` plus the affected
domain packs. Verification may reuse an existing fresh DOMAIN-AUDIT.md only
when it covers the current diff and command evidence.
6. When verification discovers missing tests, route to `silver:test` instead of
accepting coverage as a vague follow-up. Use the relevant mode:
`write`, `e2e`, `repair`, `audit`, `performance`, or `mutation`.
7. If coverage gaps are found, add or request missing tests before passing the
gate.
8. Record UAT evidence, commands run, results, unverified claims, and residual
risks. When UAT requires browser interaction, follow `silver-bullet.md §8.1`
fallback hierarchy:
- **Alumnium (preferred):** when configured, use `do` / `check` / `get` for
acceptance-criteria evidence; attach results to VERIFICATION.md.
- **Host browser MCP:** when Alumnium is absent, navigate to the app, walk
UAT steps with `browser_click` / `browser_type` / `browser_scroll`,
verify state via `browser_snapshot`, and capture `browser_take_screenshot`
per criterion. task host: `host browser MCP` tools. Attach evidence to
VERIFICATION.md.
- **Text-only:** document unverified browser-dependent criteria as WARN when
neither path is available.
9. File deferred non-blocking gaps through `silver:add`.
## Exit Gate
Verification passes only when:
- must-have behavior is evidenced;
- tests are fresh enough for the change;
- missing evidence is classified as BLOCK/WARN/INFO;
- BLOCK findings are resolved before ship.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!