Borrowing it
Nothing to install: this file belongs to praveenjuge/teak. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/praveenjuge/teak/main/.agents/skills/verify-teak/SKILL.mdgit clone --depth 1 https://github.com/praveenjuge/teakWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/praveenjuge/teak/verify-teak)<a href="https://agentmods.dev/skills/praveenjuge/teak/verify-teak"><img src="https://agentmods.dev/badge/skills/praveenjuge/teak/verify-teak/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/praveenjuge/teak/verify-teak"><img src="https://agentmods.dev/badge/skills/praveenjuge/teak/verify-teak.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00086 | $0.01881 |
| Opus 5.5 | $0.00034 | $0.00752 |
| Sonnet 5.5 | $0.00017 | $0.00376 |
| Haiku 4.5 | $0.00009 | $0.00188 |
Grade A, and why
verify-teak scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verify Teak
Prove a Teak change in the running app, not by a passing build. Launch the stack, check it is healthy, drive the features the change touches through their user entry points, keep the evidence, and clean up what the run made. The recipes for each feature live in features/; read its README first.
The primary surface is the web app at http://localhost:3000 (a worktree has its own port; bun run dev --status prints it). The REST API, CLI and MCP server are secondary surfaces with their own recipes, and so are the Apple and Android apps and Raycast, which run against the local E2E stack. The browser extension is not mapped yet.
Launch
bun run dev --detach
It sets up the checkout, starts the web app against the shared dev deployment (dev:reminiscent-kangaroo-59), signs in through WorkOS staging, and returns once the app is ready: about 50 seconds cold, 10 warm. Ready means the command returned and the doctor below passes. The first signed-in page load still compiles routes, so the card grid can show the "Loading cards" skeleton for up to a minute the first time.
- The account and its password are in
.agents/.state/stack.json(account.email,account.password). A cloud session gets its own account, seeded with about 30 sample cards (packages/convex/devSeed.ts). - When the change touches
packages/convex, the backend you test must be this checkout's. Start withbun run dev --detach --pushto take over the push lease, and confirm with the doctor. - Don't start plain
bun run devas a background command: the agent tool stops it after 30 minutes. - For a run that must not share a backend (destructive or schema-changing work), use the E2E stack instead:
bun run dev --stop, thenbun run --cwd packages/tests e2e:stack(a local Convex backend and the WorkOS emulator,packages/tests/README.md). Its accounts come from the emulator, notstack.json.
Doctor
Run this before the first drive, after anything surprising, and after any failed drive:
What ships with it
16 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- features/android.md 9.7 KB
- features/api-cli-mcp.md 5.4 KB
- features/apple.md 9.0 KB
- features/capture.md 4.7 KB
- features/card-actions.md 7.4 KB
- features/card-links.md 4.6 KB
- features/card-processing.md 5.8 KB
- features/find.md 5.4 KB
- features/link-transcripts.md 4.5 KB
- features/raycast.md 6.1 KB
- features/README.md 5.4 KB
- features/save-by-email.md 4.0 KB
- features/settings.md 4.0 KB
- features/sign-in.md 3.8 KB
- scripts/api-key.ts 5.3 KB runs code
- scripts/doctor.ts 8.5 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday Changed e524b896b9d4
- 3d ago First seen · 87 lines · 86 tokens per session scan A 62254bc293d2
verify-teak is a skill published in the GitHub repository praveenjuge/teak (25 stars, last pushed today), licensed MIT. It adds 86 tokens to every session and 1,881 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-19.
Other skills, from other repositories
browser-qa
Use when validating or debugging a workflow in the embedded browser and you need a reproducible, evidence-first loop.
playwright
Use when the task requires automating a real browser from the terminal (navigation, form filling, snapshots, screenshots, data extraction, UI-flow debugging) via playwright-cli or the bundled wrapper script.
hronaut
Use the Hronaut desktop Browser MCP for visible, persistent web workflows in isolated agent workspaces. Use for authenticated browser QA with human handoff, localhost or responsive testing, accessibility and performance diagnosis, or tasks that should survive one agent session while the user watches or takes over.
browse
Drive a real browser through Aside: open a page, read it, click through a flow, take screenshots, check console errors. (gstack).
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
create-verification-skill
Create a repo-local verification skill and exhaustive feature map for driving a real app through its UI, CLI, or API. Use for "create a verification skill", "make a verify skill for this repo", or "document how agents can verify this app".