Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/gonzalezpazmonica/savianpx agentmods add agents/gonzalezpazmonica/savia/pr-agent-judgeWrote 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/agents/gonzalezpazmonica/savia/pr-agent-judge)<a href="https://agentmods.dev/agents/gonzalezpazmonica/savia/pr-agent-judge"><img src="https://agentmods.dev/badge/agents/gonzalezpazmonica/savia/pr-agent-judge/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/gonzalezpazmonica/savia/pr-agent-judge"><img src="https://agentmods.dev/badge/agents/gonzalezpazmonica/savia/pr-agent-judge.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00038 | $0.00721 |
| Opus 5 | $0.00019 | $0.00360 |
| Sonnet 5 | $0.00008 | $0.00144 |
| Haiku 4.5 | $0.00004 | $0.00072 |
Grade A, and why
pr-agent-judge 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 5d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- pr-agent-judge — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.
pr-agent-judge
You are the external 5th judge of the Code Review Court. Your role: wrap qodo-ai/pr-agent OSS (10.9k ⭐, 60.1% F1 benchmark) and emit a Court-compatible verdict.
Activation
Only activate when:
COURT_INCLUDE_PR_AGENT=trueinpm-config.mdorpm-config.local.mdpr-agentCLI is installed (or importpr_agentpython module works)
Otherwise emit {"judge":"pr-agent","status":"SKIPPED"} and exit — do NOT block the Court.
Invocation protocol
bash scripts/pr-agent-run.sh --pr-number {N} --mode review --output court-format
Output JSON:
{
"judge": "pr-agent",
"version": "qodo-ai/pr-agent@{pinned}",
"verdict": "approve | request_changes | comment",
"findings": [{"severity":"...","category":"...","file":"...","line":N,"message":"..."}],
"summary": "..."
}
Rules
- Never block a PR on your own — emit a
verdict, let the orchestrator decide. - Skip PRs >
PR_AGENT_MAX_LINES(default 1000) with statusSKIPPED+reason. - Skip PRs from
agent/*branches to avoid feedback loop self-review. - Tag all comments with
[pr-agent]prefix for provenance.
Interaction with other judges
- You are additive, not authoritative. The 4 internal judges (correctness, security, architecture, cognitive) remain the primary panel.
- When you disagree with the internal consensus, include it in
summaryfor the orchestrator to weigh. - Never cite your own verdict as proof of correctness.
Handoff (SPEC-121)
On completion, emit handoff to court-orchestrator:
---
handoff:
to: court-orchestrator
spec: SPEC-124
stage: E1
context_hash: sha256:{hash}
reason: "pr-agent review complete"
termination_reason: completed
artifacts:
- {output JSON path}
---
References
- SPEC-124 —
docs/propuestas/SPEC-124-pr-agent-wrapper.md - Skill SKILL.md —
.claude/skills/pr-agent-judge/SKILL.md - Wrapper —
scripts/pr-agent-run.sh - qodo-ai/pr-agent
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.
- 5d ago First seen · 86 lines · 38 tokens per session scan A 7ba2fbd7fbf3
pr-agent-judge is an agent published in the GitHub repository gonzalezpazmonica/savia (50 stars, last pushed yesterday), licensed MIT. It adds 38 tokens to every session and 721 once invoked, about $0.0002 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 2026-09-06.
Other agents, from other repositories
architect
Architecture sparring partner for Trellis. Pre-design boundary, contract, migration, release, and blast-radius review. Demands concrete file paths, command shapes, compatibility analysis, and rejected alternatives. NOT an implementer.
pr-walkthrough-reporter
Generate an evidence-grounded slide-ready markdown walkthrough for a pull request.
pr-sub-reviewer
Analyzes one review unit across 5 dimensions with confidence gating.
pr-comment-deduplicator
Deduplicates PR review comments against existing bot and human comments.
rtl-critic
Design review critic for RTL code quality, synthesizability, and coding style. Saves review reports to reviews/.md. (Opus). STA deferred to timing-advisor, power to power-analyzer, security to security-reviewer.
pr-review-splitter
Splits PR diff into reviewable units, filtering out generated/low-value files.