Menu

Individual skill

adversarial-beam

Use this skill to gate Elixir/OTP systems on BEAM runtime architecture with a pass/fail adversarial review — a single blind reviewer subagent judges a diff or file set against 23 decidable rules covering supervision and failure design (restart strategy vs child dependencies, blocking init, durability in terminate, restart amnesia, unsupervised fire-and-forget), backpressure (cast on externally-driven ingest, push pipelines without demand, silenced or retried call timeouts, unbounded fan-out), event delivery semantics (PubSub treated as durable, jobs enqueued outside the creating transaction, non-idempotent consumers under at-least-once delivery, cross-source ordering, telemetry handlers that block or detach), shared-state races (check-then-act on ETS/Registry, hot reads through one mailbox, persistent_term churn), distribution (global singletons in netsplits, node-local uniqueness assumed cluster-wide), and runtime mechanics (wall-clock durations, sub-binary leaks, minted atoms). Verdicts only, never fixes.

skills/.experimental/adversarial-beam/SKILL.md

Read pinned source

Source commit cf93c57cac89d6fc3e4194686000411567f5caf3

What was checked

Matches the observed repository HEAD
Yes
Scanner rules are current
Yes
Repository scan complete
No
Skill scan complete
No
Observed behavior
Network or broad shell
Last checked
Nov 5, 2026

These are observations at the recorded commit, not a guarantee of safety. Review the source before running it.

Install this skill

Choose an agent in your Stack or copy a command below. Commands fetch the pinned commit and copy only this skill directory. Existing destinations are refused. Copying a command does not install anything.

An install command is unavailable: current rules, complete repository and skill scans, an eligible behavior level and a permitted license are required.

Repository context

Repository stars (not skill ratings)
222
Repository last updated
Aug 15, 2026
License
MIT

Observed agent integration: No specific integration evidence. Standard SKILL.md installation does not imply verified compatibility.

Files and excerpt

File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.

Browse files
Read excerpt

# Adversarial BEAM Gate

A BEAM runtime-architecture review gate for Elixir systems — pass/fail: a single blind reviewer subagent judges the work against this gate's rules with an adversarial mandate, and the work passes only when every rule is PASS or N/A. This skill renders verdicts; it never fixes the work.

The rules target one failure mode: **code that assumes guarantees the runtime never made** — that restarts restore data, that mailboxes push back, that broadcasts arrive, that delivery happens exactly once and in order, that a lookup is still true at the following insert, that node-local uniqueness holds cluster-wide, that the wall clock only moves forward. These bugs compile cleanly, pass unit tests, and surface as production incidents. Each rule carries an **Evidence of violation** paragraph so a reviewer can decide PASS/FAIL/N/A from artifact evidence alone. The sibling gate `adversarial-elixir` judges paradigm fit (OO habits ported onto the BEAM); this gate judges runtime semantics — run both for full coverage.

## When to Apply

- An Elixir feature that supervises processes, consumes events/jobs/streams, broadcasts over PubSub, or shares state via ETS/Registry is about to merge and needs an objective PASS/FAIL on its runtime assumptions.
- An agent (Claude, Codex) authored OTP-shaped code and you want an independent check that it did not assume durable broadcasts, exactly-once delivery, ordered cross-process messages, or restart-survivable in-memory state.
- A system is being prepared for clustering or higher load, and the single-node/low-traffic assumptions baked into it need to be surfaced as verdicts.
- An incident postmortem fixed one instance of a delivery/overload bug and you want the same class hunted across the affected area.

Do not apply to non-Elixir codebases (the reviewer prompt's precondition aborts with "GATE NOT APPLICABLE"), or when the user wants explanations and refactors rather than a verdict. The `dist-*` category goes N/A when the application demonstrably never clusters; Oban/Broadway appear in rules as canonical remedies but any mechanism with the same property passes — the reviewer prompt carries the applicability axes.