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 agentmods add skills/mathews-tom/armory/usage-auditnpx skills add Mathews-Tom/armory --skill usage-auditgit clone --depth 1 https://github.com/Mathews-Tom/armoryWrote 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/mathews-tom/armory/usage-audit)<a href="https://agentmods.dev/skills/mathews-tom/armory/usage-audit"><img src="https://agentmods.dev/badge/skills/mathews-tom/armory/usage-audit.svg" alt="Measured on agentmods" 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 | $0.00063 | $0.01854 |
| Opus 5 | $0.00032 | $0.00927 |
| Sonnet 5 | $0.00013 | $0.00371 |
| Haiku 4.5 | $0.00006 | $0.00185 |
Grade A, and why
usage-audit scanned grade A with 1 finding 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 today.
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.
Reads agent configuration directorieslowAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- Count configured servers in `settings.json` and `~/.claude/settings.json`. Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
How it starts
The opening of the file, as written. The whole thing — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Usage Audit
Bloated context costs twice: you burn usage limits faster and output quality drops because models attend most to the start and end of context. This skill finds the waste and tells you what to cut.
Credit: adapted from the context-audit skill in the Claude Code Context
Cleanup Guide (2026). Armory port adds package-aware checks and treats the
scoring rubric as an overridable reference, not a verdict.
Step 1: Get /context Data
Check the conversation for recent /context output. If the user has not run
it in this session, ask:
"Run
/contextin this terminal and paste the output. I can't run slash commands myself — once I can see the breakdown I'll audit everything it flags."
STOP HERE. Do not proceed to Step 2 until real /context data is
available. The breakdown determines what to audit and in what order. Without
it, the audit is guessing. Output the message above and wait.
Step 2: Audit What's Bloated
Work the categories from largest to smallest in the /context output. Run
independent checks in parallel.
MCP Servers
Each connected server loads full tool definitions into context every turn
(~15,000-20,000 tokens each), whether you invoke a tool or not. Under Opus 4.7 this
figure can run 1.0–1.35× higher for the same schemas due to a tokenizer update —
prefer measured values from /context over these static estimates.
- Count configured servers in
settings.jsonand~/.claude/settings.json. - Report total MCP overhead from
/context. - Flag any server with a CLI equivalent (Playwright, GitHub, Google Workspace, Notion, Slack, etc.) — the CLI costs zero tokens when idle.
- Cross-reference installed armory packages: if
mcp-to-skillis installed, recommend it explicitly for the heaviest server.
CLAUDE.md Files
Read every CLAUDE.md in scope (project root, .claude/, ~/.claude/). For
each file: count lines, then test every rule against the five filters.
| Filter | Flag when... |
|---|---|
| Default | Claude already does this without being told ("write clean code", "handle errors") |
| Contradiction | Conflicts with another rule in the same or a different file |
| Redundancy | Repeats something already covered elsewhere |
| Bandaid | Added to fix one specific bad output, not improve outputs generally |
| Vague | Interpreted differently every time ("be natural", "use good tone") |
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.
- today First seen · 189 lines · 63 tokens per session scan A 93e6b7ef3235
usage-audit is a skill published in the GitHub repository Mathews-Tom/armory (316 stars, last pushed today), licensed MIT. It adds 63 tokens to every session and 1,854 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
daily-briefing
Proactive daily briefing that fires on a recurring schedule, pulls recent memory and workspace context, composes a structured summary (action items, progress, radar, next steps), and delivers it to all active channels. Enable with a time like "set up my daily briefing at 9am". Disable, reschedule, or check status at…
relevance-coarse-filter
Cheap, high-recall first-pass filter that removes obvious junk from a detector candidate pool before expensive story-origin research and PR judgment. Decides keep, monitoronly, or reject — never ranks, writes angles, verifies dates, or decides whether to pitch.
weekly-retro
Use when the user is closing an ISO week, reviewing planned outcomes against evidence, interviewing work Areas, resolving unfinished commitments, or asking for "ретро", "weekly retro", "week review", or lessons from the week.
idea
Use when capturing ONE new idea the user voices and wants recorded — "save this idea", "I have an idea", "log this idea", "/idea", "idea: ...". Creates a provenance-tracked folder (one folder per idea) in your ideas repo, dedups against an index, optionally mirrors to a GitHub Project view filtered by label:idea. NOT…
meeting-copilot
Use when preparing for, running, or closing a live meeting with an AI assistant dashboard. Triggers on "meeting copilot", "live copilot", "prepare for a call", "update copilot", "close the session", or requests to turn transcript chunks into meeting questions, topic maps, decisions, and follow-ups.
daily
Use when the user asks "что сегодня", "задачи на сегодня", "план на день", or wants the current department's living weekly plan refreshed from task facts.