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 commands/betmoar/cc-proxy-plugin/modelsgit clone --depth 1 https://github.com/betmoar/cc-proxy-pluginWhat 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 | $0.00017 | $0.00528 |
| Opus 5 | $0.00009 | $0.00264 |
| Sonnet 5 | $0.00003 | $0.00106 |
| Haiku 4.5 | $0.00002 | $0.00053 |
Grade A, and why
models 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.
What it actually says
Run the cc-proxy model list and show the user its output.
Execute:
# Resolve the plugin root. $CLAUDE_PLUGIN_ROOT is only injected in hook context,
# not into a slash-command's Bash, so fall back through the legacy PROXY_PATH
# pin, then the marketplace cache (newest version wins — the tree is always
# current), then the dev repo. Never rely on env alone.
#
# Two portability traps this form avoids, both of which bit earlier versions:
# 1. A slash-command body is a TEMPLATE: Claude Code substitutes positional
# args (dollar-one through dollar-nine) into it BEFORE the shell runs — so a
# helper referencing them is blanked to a no-op. This block uses NONE.
# 2. The command runs under the user's login shell, which on macOS is ZSH,
# NOT bash. zsh does not word-split an unquoted parameter, so `for c in
# $list` iterates once over the whole blob. A newline-fed `while IFS= read
# -r` splits identically in bash and zsh.
root=""
while IFS= read -r c; do
[ -n "$c" ] || continue
if [ -f "$c/scripts/list-models.js" ]; then root="$c"; break; fi
done <<EOF
${CLAUDE_PLUGIN_ROOT:-}
$([ -n "${PROXY_PATH:-}" ] && dirname "$(dirname "$PROXY_PATH")")
$(ls -d "$HOME"/.claude/plugins/cache/*/cc-proxy/*/ 2>/dev/null | sort -V -r | sed 's:/*$::')
$PWD
EOF
[ -n "$root" ] || { echo 'cc-proxy: cannot locate plugin root; run /cc-proxy:setup or /resume'; exit 1; }
node "$root/scripts/list-models.js"
Present the script's stdout verbatim — it is already formatted. Do not summarize or reword it. If it reports the proxy down, point the user at ~/.claude/cc-proxy/cc-proxy.log and a fresh session (/exit + /resume) to re-trigger the SessionStart hook.
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 First seen · 42 lines · 17 tokens per session scan A a8f8d8ebcff8
models is a command published in the GitHub repository betmoar/cc-proxy-plugin (2 stars, last pushed yesterday), licensed MIT. It adds 17 tokens to every session and 528 once invoked, about $0.0001 per session on Opus 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
agent-status
Show a status table of all active agent worktrees (created via /worktree), their branch, age, current activity, commit count ahead of master, and rebase-lock status.
design-council
Convene a design council — role-specialized subagents debate a technical decision.
review-code
Deep audit of code for correctness, security, testing, and resilience.
review-spec
Critical review of a specification or design document.
update-architecture-docs
Review and update architecture documentation against the codebase.
build-monitor
Continuously monitor Buildkite builds, detect failures, investigate root causes, and push fixes.