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 agentmods add skills/yaalalabs/agent-kernel/ak-dev-new-queue-transportnpx skills add yaalalabs/agent-kernel --skill ak-dev-new-queue-transportgit clone --depth 1 https://github.com/yaalalabs/agent-kernelWrote 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/yaalalabs/agent-kernel/ak-dev-new-queue-transport)<a href="https://agentmods.dev/skills/yaalalabs/agent-kernel/ak-dev-new-queue-transport"><img src="https://agentmods.dev/badge/skills/yaalalabs/agent-kernel/ak-dev-new-queue-transport.svg" alt="Measured on agentmods" 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 | $0.00105 | $0.02412 |
| Opus 5 | $0.00053 | $0.01206 |
| Sonnet 5 | $0.00021 | $0.00482 |
| Haiku 4.5 | $0.00011 | $0.00241 |
Grade A, and why
ak-dev-new-queue-transport scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- `TransportConsumer`: `fetch(batch_size, wait_seconds)`, `ack`, `nack`, `dead_letter`, How it starts
The opening of the file, as written. The whole thing — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adding a New Queue Transport
This guide walks through adding a new queue transport to the pipeline
(ak-py/src/agentkernel/pipeline/). Use the shipped implementations as references, in
increasing order of complexity:
transport/in_memory.py: the semantics in their purest form, no brokertransport/sqs.py: a broker with native FIFO groups, visibility timeout, and deduptransport/nats.py: a broker where per-session ordering is built client-side (partitioned subjects, one durable consumer per partition,max_ack_pending=1)transport/kafka.py+transport/bookkeeping.py: a broker with no per-message acknowledgement model, so receive counts and dedup are rebuilt on a bookkeeping store
Read .agents/skills/ak-dev-architecture (the pipeline section) first if you have not.
The Semantics Contract
Every transport must reproduce the SQS FIFO semantics the pipeline was extracted from
(docs/specs/495-onprem-kubernetes/research/current-queue-mode.md):
- Per-group FIFO with one in-flight message per group:
group_idis the session id; a session's turns never run concurrently or out of order, while distinct sessions run in parallel. - Bounded at-least-once redelivery with an exact
receive_count: theConsumerLoopcompares it tomax_receive_countto fire the permanent-failure hook, so it must be exact, not approximate. - Publish-time deduplication on
dedup_idwithin a window (SQS parity: 5 minutes). - Attribute round-tripping:
QueueMessage.attributes(request id, user id, status code) must survive the trip byte-identically. - Batch fetch with a bounded wait, returning fewer than
batch_sizerather than blocking past the wait. - Queue isolation: INPUT and OUTPUT never see each other's messages.
Where the broker genuinely cannot provide a guarantee, the contract suite has an explicit,
documented opt-out (see timeout_redelivery in pipeline/testing.py, which Kafka sets to
False because its consumer model has no visibility timeout). Never fake a guarantee; declare
its absence and justify it in the subclass.
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 e001efeb0a58
- 5d ago First seen · 170 lines · 105 tokens per session scan A 1fc978d9d2a9
ak-dev-new-queue-transport is a skill published in the GitHub repository yaalalabs/agent-kernel (166 stars, last pushed today), licensed Apache-2.0. It adds 105 tokens to every session and 2,412 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
notion
Notion workspace integration for searching pages, managing databases, creating postmortems, and exporting RCA findings.
datadog
Datadog monitoring integration for querying logs, metrics, monitors, events, traces, hosts, and incidents during RCA investigations.
scaleway
Scaleway cloud integration for managing instances, Kapsule Kubernetes clusters, object storage, and managed databases via CLI and Terraform.
isaac-automator
Deploy and operate a cloud Isaac Workstation with Isaac Automator: provision a GPU VM running Isaac Sim, Isaac Lab, and/or Isaac Lab Arena on AWS, GCP, Azure, or Alibaba Cloud, connect to it, move data in and out, control cost with stop/start, repair, import existing deployments, and destroy. Use when the user wants a…
bitbucket
Bitbucket code repository integration for managing repos, branches, PRs, issues, and CI/CD pipelines.
cloudflare
Cloudflare integration for DNS, CDN, WAF, edge diagnostics, and remediation with zone management, analytics, security events, and cache control.