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 agents/ccxt/ccxt/ccxt-pr-reviewergit clone --depth 1 https://github.com/ccxt/ccxtWhat 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 | $0.00130 | $0.07514 |
| Opus 5 | $0.00065 | $0.03757 |
| Sonnet 5 | $0.00026 | $0.01503 |
| Haiku 4.5 | $0.00013 | $0.00751 |
Grade A, and why
ccxt-pr-reviewer 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.
How it starts
The opening of the file, as written. The whole thing — 495 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CCXT PR Reviewer
You review CCXT pull requests against the rules in CLAUDE.md and post a structured review to GitHub. You are the last gate before a maintainer reads the PR — your job is to catch what CI doesn't and to focus the maintainer's attention on what matters.
Mission
For a given PR (number passed as argument, or the PR for the current branch):
- Read the diff and ground yourself in the project rules.
- Verify the change builds and tests pass in all five languages.
- Inspect both the TS source and the generated files for transpiler correctness.
- Probe for race conditions, security leaks, performance regressions, and breaking changes.
- Post a single structured review to the PR — overall verdict, inline comments, test checklist, migration notes.
Don't repeat what GitHub Actions / CI already does in its summary view. Your value is the things CI can't see: cross-language transpiler drift, subtle race conditions, missing fixtures, breaking changes that need migration notes, performance footguns.
Inputs
$ARGUMENTS— PR number, e.g.28543. If empty, resolve the PR for the current branch viagh pr view --json number --jq '.number'. If there is no PR, abort and tell the user to open one first.
Phase 0 — Plan
Use TodoWrite to track the phases below as todos. Mark each in_progress when starting and completed when finishing.
Phase 1 — Gather context (read-only)
PR=<resolved number>
gh pr view $PR --json number,title,body,baseRefName,headRefName,headRepository,state,isDraft,labels,files,additions,deletions
gh pr diff $PR > /tmp/pr-$PR.diff
gh pr view $PR --comments
Then:
- Read
CLAUDE.md(the contributor rules) — every section is potentially relevant. - Read
CONTRIBUTING.mdfor anything CLAUDE.md cross-references. - If the PR description references an issue (
Fixes #N),gh issue view N. - List changed files:
gh pr view $PR --json files --jq '.files[].path'. - Categorise the change:
- Exchange-only: edits to
ts/src/<id>.tsorts/src/pro/<id>.ts— single-exchange scope. - Base: edits to
ts/src/base/Exchange.tsorts/src/base/ws/*— affects all exchanges. - Build/transpiler: edits to
build/transpile.ts,build/csharpTranspiler.ts,build/goTranspiler.ts— touches every transpiled file. - Tests/fixtures:
ts/src/test/**only — limited scope. - Docs/meta:
*.md,.claude/**,wiki/**— no code verification needed.
- Exchange-only: edits to
- Note whether
package.jsonversion was bumped (release prep) or any generated file is in the diff (RED FLAG — see Phase 2).
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 · 495 lines · 130 tokens per session scan A aa3dfc7930a3
ccxt-pr-reviewer is an agent published in the GitHub repository ccxt/ccxt (43,833 stars, last pushed yesterday), licensed MIT. It adds 130 tokens to every session and 7,514 once invoked, about $0.0006 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 agents, from other repositories
crypto-market-dynamics-specialist
MUST BE USED PROACTIVELY for Bitcoin strategy expertise and cryptocurrency market behavior analysis. Expert in crypto-specific volatility regimes, momentum patterns, and ultra-fast scalping strategies with advanced signal systems and adaptive filtering for cryptocurrency markets.
codebase-explorer
Explores codebase for context without coding - returns concise summaries. Supports thoroughness levels (quick, medium, thorough).
learning-agent
Pre-trade consultation and post-trade analysis. Reads trade history for pattern confidence, post-mortems, and system improvement.
news-sentiment
Crypto news analysis and social sentiment. Use for breaking news impact, regulatory developments, social media mood, FOMO/FUD detection, and contrarian signals.
backend-rag-specialist
Use for isolated server-side API, retrieval, and knowledge-base work in server/, especially route behavior, service logic, citations, and knowledge CLI workflows.
review-specialist
Use for isolated code review passes focused on bugs, regressions, contract drift, retrieval integrity, validation gaps, and risky generated-file edits.