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 Alexander-Tyagunov/magician --skill divinegit clone --depth 1 https://github.com/Alexander-Tyagunov/magicianWrote 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/alexander-tyagunov/magician/divine)<a href="https://agentmods.dev/skills/alexander-tyagunov/magician/divine"><img src="https://agentmods.dev/badge/skills/alexander-tyagunov/magician/divine/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/alexander-tyagunov/magician/divine"><img src="https://agentmods.dev/badge/skills/alexander-tyagunov/magician/divine.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 | $0.00085 | $0.00480 |
| Opus 5 | $0.00043 | $0.00240 |
| Sonnet 5 | $0.00017 | $0.00096 |
| Haiku 4.5 | $0.00009 | $0.00048 |
Grade A, and why
divine 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 12d 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.
What it actually says
$divine — Codex Adapter
Read ../../references/codex-adapter.md, then read ../../../skills/divine/SKILL.md and follow the source skill through that Codex adapter. Keep the source skill's gates (the depth-selection gate and the PR-posting confirmation gate), the context contract for dispatched reviewers, the adversarial-verification pass, and the completion criteria.
Codex equivalents:
- Change context — use Codex's shell for
gh/glabandgit diff; read the PR/MR description and linked tickets for intent. - Depth gate — ask the user how deep via Codex's question/approval UI instead of AskUserQuestion.
- Lenses — when agent tools are available, run correctness/security/simplification/tests as bounded agents with generic self-contained prompts (Goal, Scope, diff/intent inputs, constraints, evidence and Return format). Do not request Claude profile names. Otherwise perform the lenses locally.
- Grounding — invoke
$magic(or available official docs/web tooling) when the change needs external/domain evidence. - Posting — writing a review to a PR/MR is a side effect: confirm explicitly and use the correct account before any
gh/glabwrite. - Implement fixes (optional) — only on explicit user confirmation, spin a Codex task to fix Critical/High findings; committing/pushing is a side effect requiring approval, the correct account, and a feature branch.
- Monitor mode — only create a recurring automation when the user explicitly asks for scheduling. Otherwise poll a running forge CLI via an exec session. Keep it idempotent by head SHA and review-only; never implement or push.
- Review artifacts — keep intermediate lens reports in a temporary directory unless the user explicitly requests durable repository artifacts. A review must not dirty the worktree.
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.
- 12d ago First seen · 19 lines · 85 tokens per session scan A eb5bde19c620
divine is a skill published in the GitHub repository Alexander-Tyagunov/magician (14 stars, last pushed 11d ago), licensed MIT. It adds 85 tokens to every session and 480 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
reviewing
Claude-on-Claude code review protocol — reviews implementation against spec requirements and code quality standards.
loop-verifier
The checker half of a maker/checker loop. Independently reviews a proposed fix before it becomes a PR. Has authority to REJECT. Never makes changes itself.
g-review
Run the review gate on the current branch diff. Runs the test suite, captures the diff, and dispatches code-lead, which verifies done conditions and reviews the diff itself. Issues MERGE READY or HOLD.
g-refactor
Guided refactor workflow — identify target, pre-analyse, spec, approve, execute, review. Accepts a scope path, an audit milestone file, or runs interactively. Safe-by-default: checks test coverage before execution and runs the full review gate after.
g-audit
Full-codebase or targeted code quality audit. Detects SOLID violations, code smells, architectural drift, dead code, and test coverage gaps. Targeted scope produces an inline report. Whole-codebase scope produces a prioritised roadmap milestone.
g-blast-radius
Analyse the blast radius of a planned change. Inputs a file path, feature name, or list of paths from a plan. Outputs the set of dependent files (forward and reverse references), a per-file volatility score (commit frequency proxy), and a total blast-radius rating (low / moderate / wide). Read-only.