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 Rohit-ATS/Airlock --skill infra-safetygit clone --depth 1 https://github.com/Rohit-ATS/AirlockWrote 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/rohit-ats/airlock/infra-safety)<a href="https://agentmods.dev/skills/rohit-ats/airlock/infra-safety"><img src="https://agentmods.dev/badge/skills/rohit-ats/airlock/infra-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/skills/rohit-ats/airlock/infra-safety"><img src="https://agentmods.dev/badge/skills/rohit-ats/airlock/infra-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 | $0.00050 | $0.01996 |
| Opus 5 | $0.00025 | $0.00998 |
| Sonnet 5 | $0.00010 | $0.00399 |
| Haiku 4.5 | $0.00005 | $0.00200 |
Grade A, and why
infra-safety 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 11d 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.
How it starts
The opening of the file, as written. The whole thing — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Infrastructure safety
Most infrastructure incidents are caused by cleanup, not by building. The change that takes production down is rarely the deploy; it is the tidy-up three weeks later, performed confidently by someone who checked that the thing was unused.
The rule that governs this whole skill: "nothing reads from it" and "nothing should read from it" are different findings. A resource can be unread because the feature that reads it is broken. Establish which one you are looking at before proposing a deletion.
1. Reversible or not — be honest about which
| Operation | Reversible? | The catch |
|---|---|---|
| Scale a stateless deployment | Yes | None, if capacity allows |
| Scale down a StatefulSet | Usually | PVCs are retained by default, so data survives — unless the reclaim policy is Delete |
| Delete a PVC | No | The volume goes with it under Delete reclaim |
| Delete an object storage bucket | No | And the name may be unclaimable afterwards |
| Rotate a secret | Yes, with overlap | Not reversible if the old value was not retained |
| Change a DNS record | Yes | But not quickly — see §4 |
| Delete a security group / firewall rule | Yes | Existing connections survive; new ones do not, so the failure is delayed and confusing |
| Terminate an instance in an ASG | Yes | It comes back different, which is usually the point and occasionally the problem |
A scale-down that can be scaled back up is a legitimate UNDO certificate: apply against a shadow topology, restore it, compare. A deletion is a SCOPE certificate, always.
2. Scaling down: prove the headroom first
The question is never "can the remaining nodes serve the current load". It is "can they serve peak load, with one of them gone, during the window before autoscaling reacts".
Compute and report:
- peak utilisation over the last 30 days, not mean — the mean of a spiky workload is a number about nothing
- headroom after the change, at peak, with N-1 of the remaining nodes
- the autoscaler's reaction time, and what happens during it
- connection capacity, not just CPU: a Postgres replica pool sized for
max_connectionsfails on connection exhaustion long before CPU
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.
- 11d ago First seen · 189 lines · 50 tokens per session scan A 8e4104318ebb
infra-safety is a skill published in the GitHub repository Rohit-ATS/Airlock (1 stars, last pushed 3d ago), licensed MIT. It adds 50 tokens to every session and 1,996 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
reweave-operator
Operate a Reweave self-healing web-data fleet — run pipelines, triage structural-drift incidents, review repair proposals, and manage the human approval workflow. Use whenever a monitored source is broken, a heal is pending, or the operator asks about pipeline health or impact.
resonate-server-postgres
Run the Resonate protocol inside Postgres itself using resonate-pg — one SQL file of stored procedures, with no server process at all. Covers installing the schema, the pgcron timer dependency and its silent-failure mode, reaching the protocol through resonaterpc, starting workflows with resonate.invoke, the…
prisma-schema-conventions
Data modeling conventions for Prisma + PostgreSQL in multi-tenant systems — workspace isolation, numeric criticality, idempotent recurrence, document versioning by lineage and decoupled audit via a queue in Postgres itself. Use whenever creating or changing schema.prisma, designing new entities, planning migrations or…
postgres-patterns
Provides PostgreSQL patterns for query optimization, schema design, indexing strategies, RLS, and security. Use when working with PostgreSQL SQL files or when the user mentions PostgreSQL, Postgres, pgvector, Supabase, or database optimization.
postgres-patterns
PostgreSQL database patterns for query optimization, schema design, indexing, and security. Quick reference for common patterns, index types, data types, and anti-pattern detection. Based on Supabase best practices.
agent-harness-fault-injection
Use when an agent workflow needs deterministic recovery evidence for sandbox, MCP/tool, worker, checkpoint, memory, or orchestration failures.