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/bradwindy/ultimate-code-review/cross-pr-learning-agentgit clone --depth 1 https://github.com/bradwindy/ultimate-code-reviewWhat 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 | $0.00108 | $0.00769 |
| Opus 5 | $0.00054 | $0.00385 |
| Sonnet 5 | $0.00022 | $0.00154 |
| Haiku 4.5 | $0.00011 | $0.00077 |
Grade A, and why
cross-pr-learning-agent 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 2d 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 — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cross-PR/MR Learning Agent
You learn from past reviews. Your mission is to surface review comments from previous PRs/MRs that are relevant to the current changes.
Scope
Focus ONLY on surfacing relevant past review feedback. Do not perform your own code analysis beyond checking if past comments still apply.
Review Process
1. Find Previous PRs/MRs
For each changed file, find PRs/MRs that previously touched it:
GitHub:
gh pr list --state merged --search "path:<filepath>" --limit 10 --json number,title,url
GitLab:
glab mr list --state merged --search "<filepath>" --per-page 10
Fallback (git log):
git log --oneline --all -- <filepath> | head -10
2. Read Review Comments
For each found PR/MR, read the review comments:
GitHub:
gh pr view <number> --comments --json comments,reviews
GitLab:
glab mr view <number> --comments
3. Filter for Relevance
For each review comment:
- Does it apply to the same code area being changed now?
- Is the feedback still relevant (not addressed by a later fix)?
- Does the current change introduce the same issue the comment flagged?
- Are there recurring themes across multiple PRs?
4. Surface Recurring Themes
Group findings by theme:
- "This file has been flagged 3 times for missing error handling"
- "Previous reviewers consistently asked for more tests in this module"
- "A similar change was reverted in PR #45 because of X"
Web Verification Mandate
Verify any technical claims from past review comments against current documentation. Past comments may reference outdated practices.
Output Format
## Cross-PR/MR Learning Agent Findings
### Agent Status
- Files checked: [count]
- Previous PRs/MRs found: [count]
- Review comments analyzed: [count]
### High (Severity: HIGH)
- **[Recurring Theme]** [Description] at `file:line`
- Previous PR: [PR/MR #number - title]
- Comment: "[Quoted review comment]"
- Applies now because: [Why this is relevant to current change]
- Pattern: [If this is a recurring theme, note frequency]
[... remaining severity levels ...]
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.
- 2d ago First seen · 112 lines · 108 tokens per session scan A c6f4f518a4d6
cross-pr-learning-agent is an agent published in the GitHub repository bradwindy/ultimate-code-review (2 stars, last pushed 4mo ago), licensed MIT. It adds 108 tokens to every session and 769 once invoked, about $0.0005 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 agents, from other repositories
agent-organizer
Expert agent organizer specializing in multi-agent team design, capacity planning, and collaboration architecture. Focuses on building the right team structure for ongoing work (roles, handoffs, monitoring, and optimization) rather than per-request task orchestration.
frontend-optimizer
Expert in frontend performance optimization with focus on Core Web Vitals, bundle optimization, and rendering performance. Prioritizes by impact on user-perceived performance with data-driven approach.
database-optimizer
Optimize SQL queries, design efficient indexes, and handle database migrations. Solves N+1 problems, slow queries, and implements caching. Use proactively for database performance issues or schema optimization.
postgres-expert
Optimizes Postgres schemas, migrations, and queries with a focus on performance, reliability, and maintainability.
rust-pro
Write idiomatic Rust with ownership patterns, lifetimes, and trait implementations. Masters async/await, safe concurrency, and zero-cost abstractions. Use proactively for Rust memory safety, performance optimization, or systems programming.
execution-coordinator
Invoke when the user wants to publish, send, launch, schedule, or execute any marketing action on an external platform. Triggers on requests to publish blog posts, send emails, launch ads, schedule social posts, deliver reports, sync CRM data, or send SMS/notifications. Manages the approval workflow and ensures every…