Skip to content
Back to skills

Verify

ASecurity

Build, launch, and drive Claudescope against sandboxed fixtures to verify a change end-to-end without touching real agent sources or ~/.claudescope.

  • 22 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 18, 2026
ai-agentsgobashnodeapi

Works with

  • claude code
  • cli
  • api

Security analysis

A92/100
  • mediumUses curl or wget to download content

Pro shows the line behind each finding and how to fix it

Scanned November 18, 2026

npx -y skills add vladar107/claudescope --skill verify --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Verify?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Verify
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vladar107-verify/badge)](https://www.skillsdirectory.com/skills/vladar107-verify)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: verify
description: Build, launch, and drive Claudescope against sandboxed fixtures to verify a change end-to-end without touching real agent sources or ~/.claudescope.
---

# Verifying Claudescope changes

## Sandboxed launch (never touch real agent homes or ~/.claudescope)

Every data source and state dir is env-overridable (`packages/server/src/config.ts`).
Build, then run the server against a scratch dir:

```bash
npm run build    # shared → web → server; server serves the built SPA

SB=$(mktemp -d)  # sandbox: fixtures + app state
mkdir -p $SB/home $SB/projects/-tmp-demo $SB/empty
# Claude Code fixture format: see writeFixtures() in packages/server/test/api.integration.test.ts

cd packages/server
CLAUDESCOPE_HOME=$SB/home CLAUDE_PROJECTS_DIR=$SB/projects \
CODEX_SESSIONS_DIR=$SB/empty JUNIE_SESSIONS_DIR=$SB/empty PI_SESSIONS_DIR=$SB/empty \
OPENCODE_DATA_DIR=$SB/empty COPILOT_SESSIONS_DIR=$SB/empty \
ANTIGRAVITY_CLI_DIR=$SB/empty ANTIGRAVITY_DIR=$SB/empty \
PORT=4390 REINDEX_INTERVAL_MS=3000 node dist/index.js &
```

- Use a non-default port (**not** 4317 — the user's real daemon may hold it).
- `REINDEX_INTERVAL_MS=3000` makes auto-reindex fast enough to observe.
- API surface: `curl http://127.0.0.1:4390/api/...` (health, sessions, search).
- **Kill the server when done** (kill the PID you started).

## Driving the UI

Playwright is already a dev dependency (perf suite). From a script anywhere:

```js
import { chromium } from '<repo>/node_modules/playwright/index.mjs';
```

Gotchas learned the hard way:

- Playwright fires `framenavigated` for SPA pushState too — to prove "no page
  reload", set `window.__marker = 1` after load and check it survived.
- Steady-state list updates arrive within reindex interval + 10s idle health
  poll; wait up to ~30s for new rows.
- Dropping a new `<cwd-slug>/sessN.jsonl` fixture while the app runs is the
  way to simulate a live agent session landing.

## Flows worth driving

- Browse → project card → session list → session page (the core read path).
- Drop a new session fixture mid-run: list + project header stats should
  update without reload (dataVersion refetch).
- Kill + restart the server with the page open: it must recover via the
  unreachable poll and keep updating (dataVersion resets on restart).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.