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/greenpolo/cc-multi-cli-plugin/adversarial-reviewgit clone --depth 1 https://github.com/greenpolo/cc-multi-cli-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 | $0.00022 | $0.00301 |
| Opus 5 | $0.00011 | $0.00151 |
| Sonnet 5 | $0.00004 | $0.00060 |
| Haiku 4.5 | $0.00002 | $0.00030 |
Grade A, and why
adversarial-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Run Codex as a skeptical reviewer with exactly one Bash call:
node "${CLAUDE_PLUGIN_ROOT}/scripts/multi-cli-companion.mjs" adversarial-review $ARGUMENTS 2>&1
Codex stress-tests the design decision, looks for the strongest counter-argument, and surfaces risks rather than rubber-stamping. Read-only — it never edits files. Use this when you want the approach challenged, not just a line-by-line diff review.
Execution:
- Pass
--wait/--background/--base/--scopethrough verbatim; remaining text is the focus and goes through unchanged. Never pass--write. - Small diffs: run in the foreground. Large or unclear scope: run the same call with
run_in_background: trueand report the review when it finishes. - Do not review the code yourself, read files, or summarize. Return the companion's stdout verbatim.
- If the Bash call fails or returns nothing, your entire response is one line:
Codex adversarial review failed: <one-line reason>.
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 · 22 lines · 22 tokens per session scan A ab24bf617076
adversarial-review is a command published in the GitHub repository greenpolo/cc-multi-cli-plugin (89 stars, last pushed 17d ago), licensed Apache-2.0. It adds 22 tokens to every session and 301 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 2026-08-30.
Other commands, from other repositories
delegate
Delegate to Cursor, Composer, or cursor-agent via cursor-delegate-mcp (see delegate skill).
release-harn
Run the tag-first Harn release workflow.
test
Run all PandaFilter verification steps in order. Stop and report on first failure. This is the required pre-commit gate.
autocommit
Toggle auto git commits on/off for this project. Run without args to see status, or with "on"/"off" to set.
iterate
Iteration delivery — delivery check + archive + commit & tag + bump version.
gaai-audit
Show workspace audit log with filters and optional JSON export.