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 skills add Ashkaan/contextium --skill reviewgit clone --depth 1 https://github.com/Ashkaan/contextiumWrote 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/ashkaan/contextium/review)<a href="https://agentmods.dev/skills/ashkaan/contextium/review"><img src="https://agentmods.dev/badge/skills/ashkaan/contextium/review/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/ashkaan/contextium/review"><img src="https://agentmods.dev/badge/skills/ashkaan/contextium/review.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 | $0.00179 | $0.01203 |
| Opus 5.5 | $0.00072 | $0.00481 |
| Sonnet 5.5 | $0.00036 | $0.00241 |
| Haiku 4.5 | $0.00018 | $0.00120 |
Grade A, and why
review 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 yesterday.
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 — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
.agents/skills/review — the vendor review chain
A folder of scripts, not a workflow: the loop skills call these by path, and
this file exists so the folder is a skill in the shape .agents/checks/check-skills.ts
holds every folder here to.
| Script | Reviews | Called by |
|---|---|---|
code-review.ts <base> <head> / --since <tree> / --snapshot |
a git diff, with the blast-radius pack spliced in | /implement Phase 4.7, /implement-audit, validate.ts |
spec-audit.ts <spec-or-folder> <brief> [pushback-file] |
a SPEC or spec folder | /spec-audit |
policy-review.ts <task-kind> <artifact> <brief> |
one file, under any policy row | /spec-audit, /author |
policy-chain.ts |
imported by the three above; walks the row | — |
blast-radius.ts + blast-radius-symbols.ts |
who imports, what imports, who calls | code-review.ts |
find-peers.ts <files> / --verify-sweep <regex> -- <pathspec> |
the siblings a fix left behind | run-automated-checks.ts, /implement § 3.1.5 |
Where things are found
- The repo under review is the caller's:
code-review.tsreadsCODEX_REVIEW_REPO, elsegit rev-parse --show-toplevelfrom the cwd, and exits 2 with one line when neither names a repo. It never falls back to the checkout this script sits in — that reviews the wrong tree. - The other half of a cross-repo change is named by
CODEX_REVIEW_COMPANIONS(worktree paths, comma-separated): the prompt lists each with its branch and tells the reviewer to read that repo there, not from the landed checkout, which does not carry the change yet. A path that is not a git checkout exits 2. Without it, a reviewer spends rounds on findings about code the companion worktree has already changed. - The policy is
policy.jsonbeside these scripts: rowsadversarial-review,judgment,panelandrepo-investigation, each an ordered list of vendor CLIs (codex,grok; aclaudeslot means "dispatch your own agent").tracksnames an exact model for the CLI's-m; empty runs the CLI's own default.POLICY_JSONoverrides the path;POLICY_CHAIN_RESOLVERoverrides the model resolver. - The author is left out. A review row skips the vendor recorded as
agent=in.agents/harness(the model family that writes the code); thepanelkeeps every seat. - The parsers
blast-radius-symbols.tsuses —web-tree-sitter,tree-sitter-bash,typescript— are declared inpackage.jsonhere and found in the workbench's ownnode_modules, then npm's global root (npm i -g typescript@5 web-tree-sitter tree-sitter-bash), then Node's own globalnode_modules. Without them the packer warns and falls back to a regex walk.
What ships with it
29 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.
- agents/openai.yaml 43 B
- blast-radius-symbols.test.ts 13 KB runs code
- blast-radius-symbols.ts 25 KB runs code
- blast-radius.test.ts 26 KB runs code
- blast-radius.ts 27 KB runs code
- code-review.test.ts 38 KB runs code
- code-review.ts 56 KB runs code
- evals/caller-contract.eval.test.ts 7.6 KB runs code
- evals/caller-contract.eval.ts 8.3 KB runs code
- evals/README.md 1.5 KB
- find-peers.test.ts 20 KB runs code
- find-peers.ts 22 KB runs code
- package.json 666 B
- policy-chain.test.ts 49 KB runs code
- policy-chain.ts 40 KB runs code
- policy-review.test.ts 5.3 KB runs code
- policy-review.ts 3.1 KB runs code
- policy.json 2.0 KB
- run-tests.test.ts 3.2 KB runs code
- run-tests.ts 3.1 KB runs code
- spec-audit.test.ts 8.3 KB runs code
- spec-audit.ts 22 KB runs code
- tests/fixtures/planted-bugs/metrics.ts 2.0 KB runs code
- tests/stubs/stub-clean.ts 495 B runs code
- tests/stubs/stub-empty.ts 497 B runs code
- tests/stubs/stub-fail.ts 579 B runs code
- tests/stubs/stub-hang.ts 378 B runs code
- tests/stubs/stub-narrate-then-answer.ts 1.5 KB runs code
- tests/stubs/stub-prose.ts 526 B runs code
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.
- yesterday Changed · +7 lines 8f2954d096a3
- 15d ago Changed · +1 lines 601433549601
- 16d ago First seen · 57 lines · 179 tokens per session scan A 24d42d7cc608
review is a skill published in the GitHub repository Ashkaan/contextium (36 stars, last pushed yesterday), licensed MIT. It adds 179 tokens to every session and 1,203 once invoked, about $0.0007 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 2026-09-30.
Other skills, from other repositories
code-review-loop
Use when opening a PR for review or when receiving review feedback. Activate for keywords like "code review", "PR review", "request review", "review feedback", "address comments", "reviewer said". Covers both ends of the loop: preparing a reviewable PR and acting on feedback rigorously. Always engage with every…
Review Changes
Review or explain the code changes in a working tree or branch, grouped by severity, with file and line references and an architecture diagram of what the change does.
red-team-review
Unified adversarial review, two modes. Strategic Matrix v4.3 (MTA-004) for arguments/plans/decisions: Priors → Rubric → Adversarial Lenses → SWOT/TOWS → MCDA → Blind Spot/Kill Switch → Summary. Deliverable Audit for built artefacts (assignments, reports, code outputs): isolated fresh-context hunt against the brief…
decision-journal
Unified decision lifecycle: Pre-decision logging, post-decision review, failure classification, and calibration tracking. Absorbs: post-mortem-engine.
release-pr-review
Review pass on an open release PR (release/ → main) — the step between git-wrapup and release-and-publish when a project releases in gated release PR mode. Reads the PR's commit range through the code-simplifier lens plus a correctness review, verifies whatever an automated reviewer left on the PR, lands fixes as…
git-worktree-swarm
Automates the complex setup of parallel git worktrees for agentic swarms with dependency-aware wave execution.