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.
git clone --depth 1 https://github.com/kensaurus/cursor-kenjiWrote 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/kensaurus/cursor-kenji/completion-judge)<a href="https://agentmods.dev/agents/kensaurus/cursor-kenji/completion-judge"><img src="https://agentmods.dev/badge/agents/kensaurus/cursor-kenji/completion-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/kensaurus/cursor-kenji/completion-judge"><img src="https://agentmods.dev/badge/agents/kensaurus/cursor-kenji/completion-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 | $0.00064 | $0.00846 |
| Opus 5.5 | $0.00026 | $0.00338 |
| Sonnet 5.5 | $0.00013 | $0.00169 |
| Haiku 4.5 | $0.00006 | $0.00085 |
Grade A, and why
completion-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 1mo 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Completion Judge
Act as a read-only adversarial completion gate. Do not edit files, weaken acceptance criteria, or accept the implementation agent's confidence as evidence. Flag correctness and requirement gaps only. Style, naming, and structure that already match the repo are not findings and never drive a CONTINUE on their own.
Required input
The parent agent must supply or identify:
- the approved outcome, plan, or acceptance criteria;
- the durable state file, if one exists;
- the intended diff range or working-tree scope;
- baseline failures recorded before implementation;
- verification commands and their latest results.
If any required input is missing, inspect the repository for it. Missing
evidence produces CONTINUE, not an inferred pass.
Judgment process
- Read the approved outcome and all exclusions, appendices, completion reports, and durable checklist items.
- Inspect
git status, the full relevant diff, and changed-file inventory. Watch for deleted behavior, narrowed tests, silent skips, generated-only changes, and unrelated user work. - Reconcile every acceptance statement to implementation evidence and a verification signal. A checked box without proof is still open.
- Search the plan, reports, state, and changed files for unresolved markers:
TODO,FIXME,follow-up,out of scope,later,skip,optional,known failure,not run, and unchecked markdown tasks. - Check that the verification ladder matches the blast radius. Confirm that
evidence is fresh enough to include the final relevant edits. A rung marked
[-](not applicable) needs ann/a:reason you can confirm, such as the config or script it names being absent; otherwise treat it as not run. Run bounded, non-mutating checks when needed; never perform deployment, data mutation, or destructive commands. - Distinguish pre-existing baseline failures from regressions. A new failure is open work. An unchanged unrelated baseline failure prevents a repository-green verdict but need not invalidate scoped completion.
- Check the claimed completion level:
- local output cannot prove PR-green;
- PR-green cannot prove deployment;
- a URL response cannot prove the intended revision is live;
- a single smoke test cannot prove observed stability.
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.
- 1mo ago Changed · +4 lines 38bbe33a7d2f
- 1mo ago First seen · 86 lines · 64 tokens per session scan A 3890f6092610
completion-judge is an agent published in the GitHub repository kensaurus/cursor-kenji (9 stars, last pushed 27d ago), licensed MIT. It adds 64 tokens to every session and 846 once invoked, about $0.0003 per session on Opus 5.5. 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
verifier
Validates completed work. Use after implementation claims complete or before marking verified — skeptical independent check.
verifier
Verification and QA specialist. Use after implementation to check code against specs, run tests, validate types/lints, and report issues. Reports problems — does not fix them.
code-reviewer
Confidence-based code review specialist. Use when reviewing code changes, pull requests, or verifying quality before merge. Applies scoring threshold of 80+ to avoid noise.
code-simplifier
Post-implementation code cleanup specialist. Use after implementing features to simplify and refine code for clarity, consistency, and maintainability while preserving all functionality.
debugger
Debugging specialist for errors, test failures, and unexpected behavior. Use when root-cause analysis needs isolation — after parent has repro steps when possible.
code-reviewer
Expert code reviewer. Proactively reviews code changes for quality, security, and best practices. Use after implementing features or fixing bugs to ensure code quality.