Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add cboone/agent-harness-plugins/plugin install review-until-cleanWrote 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/skills/cboone/agent-harness-plugins/review-until-clean)<a href="https://agentmods.dev/skills/cboone/agent-harness-plugins/review-until-clean"><img src="https://agentmods.dev/badge/skills/cboone/agent-harness-plugins/review-until-clean/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/skills/cboone/agent-harness-plugins/review-until-clean"><img src="https://agentmods.dev/badge/skills/cboone/agent-harness-plugins/review-until-clean.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 | $0.00025 | $0.06501 |
| Opus 5.5 | $0.00010 | $0.02600 |
| Sonnet 5.5 | $0.00005 | $0.01300 |
| Haiku 4.5 | $0.00003 | $0.00650 |
Grade C, and why
review-until-clean scanned grade C with 1 finding 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.
Hidden instructionshighPrompt injection
Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.
<!-- validate-plugins: ignore /simplify /pr-review-toolkit:review-pr --> How it starts
The opening of the file, as written. The whole thing — 272 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Review Until Clean
Run one reviewer over every local change, fix the findings at or above a severity threshold, and repeat until a finished review of the snapshot that is actually on disk leaves nothing to fix. The loop owns the scope, the snapshot, the ledger, the fixes and the decision to stop. It knows nothing about how any particular reviewer works.
Options
- --reviewer
<command>: the reviewer, written as a person would type it: a skill or slash command such as/pr-review-toolkit:review-pror/review-pr-with-codex --effort high, or a shell command such ascoderabbit review --agent --fresh. Quote a value with spaces. The reviewer's own options go inside the value. Exactly one reviewer per run - --base
<ref>: take the merge base with this ref instead of with the default branch - --severity
<critical|important|nit>: lowest severity the loop fixes; defaultimportant - --max-rounds
<n>: cap the rounds; default 3 - --confirm-clean: require two consecutive clean rounds over the same snapshot rather than one. After the first clean round, run a confirming round with no fix pass in between
- --report-only: run one round, write the ledger, and fix nothing; under
--confirm-clean, also run the confirming round when the first is clean. A reviewer that edits still edits, and its changes are kept and reported - --no-save: leave the ledger at its temporary path instead of saving it to
docs/reviews/
Severity order is Critical > Important > Nit. The default fixes Critical and Important; critical fixes only Critical, and nit fixes all three. Print and record lower-severity findings as open; they do not block cleanliness. Unlabelled findings are Important and do not block under critical. Reject other thresholds before running a reviewer.
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 Changed · +13 lines a9da1cd0bcce
- 4d ago Changed · +32 lines · -1 tokens per session scan A → C 606d31993684
- 5d ago Changed · -1 lines 97ecf534a600
- 15d ago Changed · +16 lines 849cbeae5288
- 18d ago First seen · 212 lines · 26 tokens per session scan A 3d06bacd079d
review-until-clean is a skill published in the GitHub repository cboone/agent-harness-plugins (2 stars, last pushed today), licensed MIT. It adds 25 tokens to every session and 6,501 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it C with 1 finding (hidden instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-28.
Other skills, from other repositories
review-pr
Review a pull request for conceptual fit, architecture impact, adversarial failure modes, security risk, docs impact, regression risk, test coverage, and merge readiness. Use when asked to review a PR, check whether a PR is safe to merge, decide if more tests are needed, perform adversarial or security review, or…
software-code-refactoring
Improve production code quality while preserving all existing test behavior. Commonly used for the Refactor phase of TDD red-green-refactor, but applicable to any codebase with tests. Use when production code works but needs cleanup — reducing duplication, improving naming, simplifying complexity, aligning with…
drift-check
Pre-PR advisory check for deviations from the project spec. Read-only analysis.
review-preflight
Review tooling for a PR: review-preflight prints REVIEW.md's criteria and the PR's live gate state before a review; ci-triage maps a PR's failing checks to already-filed issues.
remarc
Manage Remarc session comments and contextual feedback. Use when the user mentions Remarc sessions, comments, handoff, triage, review, addressing, resolving, status updates, or summaries.
fuck-over-engineering
Use when the user asks what could be deleted or reduced to simplify a codebase, says "simplify", "over-engineered", or wants a repo-, folder-, or file-wide audit for over-engineering — hunts dead code, reinvented stdlib, needless dependencies, and single-implementation abstractions, reports ranked cuts, applies only…