Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add paulinevos/claude-stuff --skill amp-usage-watchgit clone --depth 1 https://github.com/paulinevos/claude-stuffWrote 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/paulinevos/claude-stuff/amp-usage-watch)<a href="https://agentmods.dev/skills/paulinevos/claude-stuff/amp-usage-watch"><img src="https://agentmods.dev/badge/skills/paulinevos/claude-stuff/amp-usage-watch/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/paulinevos/claude-stuff/amp-usage-watch"><img src="https://agentmods.dev/badge/skills/paulinevos/claude-stuff/amp-usage-watch.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 | $0.00055 | $0.00870 |
| Opus 5.5 | $0.00022 | $0.00348 |
| Sonnet 5.5 | $0.00011 | $0.00174 |
| Haiku 4.5 | $0.00006 | $0.00087 |
Grade A, and why
amp-usage-watch 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 3d ago.
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.
What it actually says
Watch Amp usage
Use the bundled scripts/monitor.py with Python 3.9+ on macOS/Linux. It polls amp usage and, for a thread cap, amp threads usage <id>. A cap is an alert on total lifetime thread cost, including spending before monitoring started. It does not stop Amp or enforce a spending limit. State and deduplication persist under ~/.local/state/amp-monitor/<name>/.
Take the thread ID/URL and cap amount from the request. Interpret an unqualified cap as USD, and say so. Ask only if the intended thread or amount is missing or ambiguous. Do not use another currency as USD without resolving that choice. Accept --notify terminal|macos|both, defaulting to terminal. Default polling is every 60 seconds, low credits is $5, and low allowance is 10% remaining. Preserve any user-specified thresholds or interval.
Resolve scripts/monitor.py relative to this skill, not the user's checkout. Run help when needed. Check authentication and parsing with a single sample before starting a watcher:
python3 <skill-dir>/scripts/monitor.py check --thread T-actual-id --cap 10
This prints measured values and coverage warnings without sending notifications. If authentication fails, help the user sign in via amp login. If the CLI output is unfamiliar, inspect amp usage and amp threads usage <id> directly. Do not invent a balance or cost. A verified --credit-pattern, --allowance-pattern, or --thread-pattern can adapt the parser: each must match exactly once with one numeric capture. Allowance patterns capture percent remaining; money patterns capture USD. Pass patterns as quoted arguments, not shell-generated code.
For live terminal notifications, run in the foreground:
python3 <skill-dir>/scripts/monitor.py run --thread T-actual-id --cap 10 --notify terminal
When the agent must hand control back, start a background watcher:
python3 <skill-dir>/scripts/monitor.py start --thread T-actual-id --cap 10 --notify both
Use the user's chosen notification mode; do not silently switch terminal to macOS. Background terminal output/bells go to monitor.log, not a detached user's terminal; report the log path and use foreground execution or tail -f for live terminal delivery. macOS notifications use osascript and may require execution permission outside a sandbox. Delivery also depends on system notification settings. Background watchers survive this chat ending, but do not automatically restart after logout/reboot.
Thread watchers also check account balances by default. Use --thread-only when requested, or when the user wants the thread monitored despite unavailable account data. Account-only monitoring omits both --thread and --cap. Use --name to distinguish configurations; by default a watcher is named after its thread, or account. One running watcher is allowed per name. To change a running cap, stop the watcher and wait until status says it has exited before starting a new configuration.
python3 <skill-dir>/scripts/monitor.py status --thread T-actual-id
python3 <skill-dir>/scripts/monitor.py stop --thread T-actual-id
Stop requests end only the watcher, after any in-flight CLI request. Never imply that the Amp agent was stopped. Alerts fire once per threshold crossing and rearm after a balance/allowance recovers. Notification failures retry on later polls. Poll errors are visible and trigger a monitor-health notification; five consecutive failures stop the watcher by default. Check status and logs rather than claiming it is still protecting the budget.
Return the thread, cap in USD, monitored balances/allowance coverage, notification mode, interval, watcher name and status/log paths. Do not launch a watcher merely because someone asks to create this skill or script.
What ships with it
1 file 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.
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.
- 3d ago First seen · 44 lines · 55 tokens per session scan A c99ca21dccb9
amp-usage-watch is a skill published in the GitHub repository paulinevos/claude-stuff (64 stars, last pushed 4d ago), licensed MIT. It adds 55 tokens to every session and 870 once invoked, about $0.0002 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 2026-10-14.
Other skills, from other repositories
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…