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.
git clone --depth 1 https://github.com/mbanderas/costguardWrote 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/commands/mbanderas/costguard/costguard-live)<a href="https://agentmods.dev/commands/mbanderas/costguard/costguard-live"><img src="https://agentmods.dev/badge/commands/mbanderas/costguard/costguard-live/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/commands/mbanderas/costguard/costguard-live"><img src="https://agentmods.dev/badge/commands/mbanderas/costguard/costguard-live.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 | $0.00043 | $0.00512 |
| Opus 5.5 | $0.00017 | $0.00205 |
| Sonnet 5.5 | $0.00009 | $0.00102 |
| Haiku 4.5 | $0.00004 | $0.00051 |
Grade A, and why
costguard-live 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 1mo 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
Run a Costguard live billing check for a provider. Requested: $ARGUMENTS
Live mode is opt-in and consent-gated. costguard only PLANS the read (it emits a read-only snippet) and INGESTS the result; the browser action is performed by the playwriter MCP server, authorized by the user. costguard never drives the browser, never submits forms or credentials, and never reads cookies, storage, or screenshots — only rendered billing figures.
Steps (via the costguard MCP server's tools):
- Call
plan_live_checkswith the provider (noconfirmLiveyet). Surface the returnedconsentNoticeto the user and report theapiFirstdecision.- If
apiFirstis true, STOP and prefer/costguard-audit <ws> --providers <provider>(the API path) — no browser is needed.
- If
- Only after the user explicitly agrees (e.g. they passed
--confirm), callplan_live_checksagain withconfirmLive:trueto obtain the read-onlyreadOnlySnippet+billingUrl+parseSpec. - Ask the playwriter MCP server to
executethat snippet over the user's logged-in session (this is playwriter's own consent gate). It returns the rendered figures. - Call
ingest_live_readingwith the provider and the returnedvaluesto get a Finding. If playwriter is unavailable or the figure is unparseable, ingest returns akind:"diagnostic"Finding (excluded from cost totals) — the audit never blocks.
Notes:
- Three consent gates apply: host MCP consent, costguard's
confirmLive:true, and playwriter's own consent beforeexecute. - A provider with a resolvable API token is API-first — browser is the fallback, never the primary path.
- This command needs the costguard MCP server (declared in
.claude-plugin/.mcp.json) and the playwriter MCP server both connected.
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.
- 1mo ago First seen · 40 lines · 43 tokens per session scan A d8858e06f491
costguard-live is a command published in the GitHub repository mbanderas/costguard (2 stars, last pushed 1mo ago), licensed MIT. It adds 43 tokens to every session and 512 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 2026-08-31.
Other commands, from other repositories
guard
Show whether the nable guard is checking your agent's commands, and turn it off or on.
brainblast-fee-configs
Fee Config Validator: the generalized Bags exploit. A curated catalog of the silent zero-revenue class — revenue-bearing config fields (fees, royalties, rewards) that, if omitted or set to zero, quietly collect nothing forever. The call succeeds, nothing reverts, and a creator/treasury/holder simply earns $0.
connect
Connect a cloud account and verify nable works, with a guided first cost number.
valuation-methods
Valuation methods analysis — multiples, DCF inputs, PEG integration, valuation assumption extraction.
cost-per-task
Compute cost-per-successful-task unit economics from usage and cost data, including retries, failures, and quality thresholds.
audit-checklist
Perform an internal audit, review controls, or prepare for an external financial audit.