agentmods.dev

verify-backup-verdict

verify-backup-verdict is a skill for Claude Code, Codex from EtanHey/brainlayer. It costs 91 tokens per session (1,484 once invoked), scanned A, original, Apache-2.0.

A formal check for the rules that decide whether BrainLayer's backup-status dot is green or red. It uses TLA+, a language for describing and checking system behaviour, together with automated tests.

In plain words
What is it for?
Run it before merging changes to backup receipts, snapshot cleanup, weekly `VACUUM` maintenance, status history, or job alerts. It checks the formal cases and the accompanying Python test.
Why use it?
Backup status can be misleading if old or incomplete evidence is treated as current. This check catches changes that could make the dot claim that memory is safe when no verified copy exists, or fail to show a real risk.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

not rated 9repo yesterday A scan 91 tokens original Apache-2.0

Good fit Run it before merging changes to backup receipts, snapshot cleanup, weekly VACUUM maintenance, status history, or job alerts. It checks the formal cases and the accompanying Python test.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/etanhey/brainlayer/backup-verdict
View source ↗ EtanHey/brainlayer brainlayer.etanheyman.com
Install

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.

Any agent
npx skills add EtanHey/brainlayer --skill backup-verdict
Clone the repo
git clone --depth 1 https://github.com/EtanHey/brainlayer

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for verify-backup-verdict

README.md
[![agentmods](https://agentmods.dev/badge/skills/etanhey/brainlayer/backup-verdict/github.svg)](https://agentmods.dev/skills/etanhey/brainlayer/backup-verdict)
Your own site
<a href="https://agentmods.dev/skills/etanhey/brainlayer/backup-verdict"><img src="https://agentmods.dev/badge/skills/etanhey/brainlayer/backup-verdict/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.

agentmods 80×15 button for verify-backup-verdict

Your own site · 80×15
<a href="https://agentmods.dev/skills/etanhey/brainlayer/backup-verdict"><img src="https://agentmods.dev/badge/skills/etanhey/brainlayer/backup-verdict.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 91 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,484 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00091 $0.01484
Opus 5.5 $0.00036 $0.00594
Sonnet 5.5 $0.00018 $0.00297
Haiku 4.5 $0.00009 $0.00148

Measured 3d ago against content hash c910b1b9f330, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

verify-backup-verdict 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 3d ago.

The scan reads SKILL.md. This mod also ships 4 executable files (check.py, check.sh, replay.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

formal/backup-verdict/SKILL.md · 88 lines

How it starts

The opening of the file, as written. The whole thing — 88 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Verify the backup verdict

This skill checks one promise: the dot (plan section 3d) is red exactly when memory is at risk or a pipe stopped. It is never green without a verified copy, never red only because of a job alert, and never clears red without the evidence section 3d names. Everything lives in formal/backup-verdict/. Read its README first.

1. Does your change touch the verified code?

Run formal/backup-verdict/check.sh --list and open coverage.json. Your change is in scope if it edits any file under sources, or any path:line under actions. Look out especially for _verified_age, evaluate()'s unknown / hold / hysteresis / wake-grace branches, backup_facts, badge_alerts, _recent_local_verified_backup, the VACUUM deferred paths in run_maintenance, maintenance.main's report(...) calls and last_weekly.

2. Run it

formal/backup-verdict/check.sh            # about 3 min; exit 0 = every case matched
python3 -m pytest -q formal/backup-verdict/test_check.py -p no:cacheprovider -o pythonpath=

This was proved once: a scratch edit that put #1583 B5's persisted max back into _verified_age made bug-1583-B5 FAIL (exit 1), and restoring the line made it pass.

Each line reads ok|FAIL <case>: expected X, got Y. Under a bug case, replay at <ref> says whether the real code at the pre-fix commit reproduces the counterexample, and at the current tree whether it still does. A SYNC STALE line (exit 1 under --require-sync, which CI uses) means someone must confirm the model describes the changed source (step 4).

3. A real bug, or a modelling error?

Open .build.noindex/backup-verdict/<run>/<case>.trace.txt (the shortest path, changed variables only), or read the result with the pstack /statechart skill, which tells each violation as its shortest path in plain words.

  • A current-* case fails: the change broke the promise. For a BackupVerdict trace, confirm it against the real code: copy the bug-case pattern (a config plus a prefix ref in CASES), or run check.sh --case <case> --src-ref <your commit>. If the real code reproduces the trace, it is a real bug. Report it in the code's words (receipts, ticks, check states), as COUNTEREXAMPLES.md does.
  • A bug-* case FAILs with "UNEXPECTED" on the working-tree replay: either your change brought a fixed bug back (a real regression; fix the code), or the bug's PR just merged and its fixed_here is still "no" (flip it to "yes").
  • A bug-* case passes, or its replay "does not reproduce" at its pre-fix ref: the model drifted from the code it claims to describe. Fix the model. Never delete the case.
  • A trace walks a path the real code cannot take (for example, an unobserved fact the collectors always observe): that is a modelling error. Fix the action's precondition, and cite the code that makes it impossible.
  • Never weaken a property until it passes. If the code breaks section 3d, add a finding-* case, record it in COUNTEREXAMPLES.md, and tell the lead. F1–F3 went this way: found as findings, fixed by S1a-5, now bug-s1a5-* cases.

Read the full file on GitHub · 88 lines

Files

What ships with it

54 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.

Changes

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.

  1. 3d ago First seen · 88 lines · 91 tokens per session scan A c910b1b9f330

Subscribe to this mod's changes

verify-backup-verdict is a skill published in the GitHub repository EtanHey/brainlayer (9 stars, last pushed yesterday), licensed Apache-2.0. It adds 91 tokens to every session and 1,484 once invoked, about $0.0004 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 2026-11-03.

Claude CodeDebuggingLLMTestingai-agentsai-memoryknowledge-graphmcp
Related

verify-backup-verdict compared with skills from other repositories

verify-backup-verdict next to 6 skills from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
verify-backup-verdictEtanHey/brainlayerSkill9yesterdayApache-2.0A91Claude Code, Codex
ixix-infrastructure/IxSkill1.1kyesterdayApache-2.0A44Claude Code, Codex
property-reproplastic-labs/honchoSkill8.0kyesterdayAGPL-3.0A74Claude Code, Codex
honcho-cliplastic-labs/honchoSkill8.0kyesterdayAGPL-3.0A61Claude Code
ori-memoryaayoawoyemi/Ori-MnemosSkill3351mo agoApache-2.0A54Claude Code, Codex
remnic-memory-workflowjoshuaswarren/remnicSkill231yesterdayMITA54Claude Code
remnic-entitiesjoshuaswarren/remnicSkill231yesterdayMITA41Claude Code