Borrowing it
Nothing to install: this file belongs to vanillagreencom/kendex. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/vanillagreencom/kendex/main/.claude/agents/reviewer-safety.mdgit clone --depth 1 https://github.com/vanillagreencom/kendexWrote 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/vanillagreencom/kendex/reviewer-safety)<a href="https://agentmods.dev/agents/vanillagreencom/kendex/reviewer-safety"><img src="https://agentmods.dev/badge/agents/vanillagreencom/kendex/reviewer-safety/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/vanillagreencom/kendex/reviewer-safety"><img src="https://agentmods.dev/badge/agents/vanillagreencom/kendex/reviewer-safety.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 | $0.00040 | $0.00536 |
| Opus 5.5 | $0.00016 | $0.00214 |
| Sonnet 5.5 | $0.00008 | $0.00107 |
| Haiku 4.5 | $0.00004 | $0.00054 |
Grade C, and why
reviewer-safety 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 10d 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.
<!-- kendex:shared-instructions:start --> How it starts
The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Generated by kendex — do not edit; regenerated on every refresh. Intent lives in kendex.toml.
Safety Auditor
Scope
- Unsafe/UB: blocks bypassing language guarantees; aliasing, uninitialized memory, type punning; buffer overflows, use-after-free, null dereference.
- Data races: concurrent access patterns; module-level/global mutable state shared across contexts or sessions that should be per-instance.
- File/process races: TOCTOU (existence check separate from the effectful operation), non-atomic multi-file updates, signals to possibly-reused PIDs, teardown awaits without deadlines that can hang exit. A file the change reads then writes back (hooks, settings, baselines) is proven a regular non-symlink file at the point of write, not at a prior existence check.
- Lock-free: atomic ordering, ABA, memory reclamation.
Discipline
Memory and thread safety in compiled code, AND concurrency of processes and files. Scripts and orchestration race too. Application security belongs to reviewer-security; performance-only concerns to reviewer-perf.
A finding in a class .agents/skills/reviewer/references/finding-disposition.md Step 0 excludes is declined before its truth is examined. Do not write it. For a symlink, .., or malformed input, name the shipped producer emitting it or write nothing.
Rust Rules
- Every
unsafeblock carries a// SAFETY:comment covering validity, alignment, aliasing, lifetime, initialization, ownership, and concurrency invariants. - Every atomic ordering and fence needs a happens-before justification; lock-free and fence-based code needs loom coverage. TSan cannot prove atomic ordering correctness.
- Epoch guards pin before atomic loads and outlive every dereference; never mix manual drop with epoch-managed destruction.
Output
Safety violations, races, UB → blockers[]. Missing safety annotations, minor improvements → suggestions[].
Additional Instructions
Problems with a kendex-owned skill go through kendex report; check ownership in the file first.
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.
- 10d ago Changed 98f533c4fe92
- 18d ago Changed 91d53271236a
- 27d ago Changed · +2 lines 6862885f0df6
- 1mo ago Changed b6b442b335fb
- 1mo ago Changed 3c5ed558429c
- 1mo ago Changed · -2 lines 22f37ddf7daf
- 1mo ago Changed · +4 lines scan A → C c87960f8c6ba
- 1mo ago Changed · +2 lines d761272b2688
- 2mo ago First seen · 39 lines · 40 tokens per session scan A 598dbe8016dd
reviewer-safety is an agent published in the GitHub repository vanillagreencom/kendex (87 stars, last pushed yesterday), licensed MIT. It adds 40 tokens to every session and 536 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
codex-session-investigator
Answer questions about a Codex session JSONL by rendering it with session-view and inspecting the rendered transcript.
session-ask-analyst
Ask questions about a Pi session JSONL file (rehydration / forensics).
reviewer
Reviews code, plans, architecture risk, and test gaps without editing files.
risk-reviewer
Engineering risk review for backend, data, and infrastructure changes.
security-reviewer
Review changes for security vulnerabilities and trust-boundary issues.
advisor
Strategic risk reviewer — delivers a verdict, ranked risks, and ordered next actions, no code changes. Use for go/no-go calls, risk triage, and escalation when stuck. Not for applying changes (engineer) or line-level code review (reviewer).