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 nolte/claude-home-assistant --skill ha-discovery-augmentgit clone --depth 1 https://github.com/nolte/claude-home-assistantWrote 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/nolte/claude-home-assistant/ha-discovery-augment)<a href="https://agentmods.dev/skills/nolte/claude-home-assistant/ha-discovery-augment"><img src="https://agentmods.dev/badge/skills/nolte/claude-home-assistant/ha-discovery-augment/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/nolte/claude-home-assistant/ha-discovery-augment"><img src="https://agentmods.dev/badge/skills/nolte/claude-home-assistant/ha-discovery-augment.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 | $0.00224 | $0.02094 |
| Opus 5 | $0.00112 | $0.01047 |
| Sonnet 5 | $0.00045 | $0.00419 |
| Haiku 4.5 | $0.00022 | $0.00209 |
Grade A, and why
ha-discovery-augment 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.
How it starts
The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
HA Discovery Augment
Spec: spec/claude/ha-discovery-augment/en.md (EN canonical) / spec/claude/ha-discovery-augment/de.md (DE translation).
Why this is a skill, not an agent
- Human-visible augmentation surface — the user describes how the device announces itself and reads back the manifest matcher, the config-flow step, and the conformance report; a skill keeps this on the visible command surface, like the sibling augment skills (
ha-config-flow-augment,ha-coordinator-add,ha-repairs-add). - Mid-flow interactivity — the mechanism choice, the matcher narrowing, and the Zeroconf/Bluetooth redirect are per-run dialogues the user approves before generation.
- Bounded, inline generation — one mechanism's manifest matcher plus its discovery step fit inline; no isolated agent context is needed.
- Counter-dimension considered: the draft→validate loop could be an agent, but the mechanism decision and the report belong in the user's working context; skill wins.
When this skill activates
Use this skill to add one discovery mechanism — DHCP, SSDP/uPnP, USB, HomeKit, or MQTT — to an existing integration so the user doesn't have to type a host or pick the integration manually.
When NOT to activate
- mDNS/Zeroconf discovery →
ha/zeroconf-discovery(the scaffold already produces it) - Bluetooth discovery → separate sibling spec
ha/bluetooth - greenfield integration scaffolding →
ha-integration-scaffold - generic config-flow auth/selection steps →
ha-config-flow-augment/ha/config-flow-patterns - deploying/importing into a running HA instance → out of scope
Hard rules
- One mechanism, one run. No multi-mechanism batches.
- Read
spec/ha/discovery-mechanisms/en.mdfirst. Do not generate from memory. - Right mechanism, with delimitation. mDNS/Zeroconf →
ha/zeroconf-discovery/scaffold; Bluetooth →ha/bluetooth. Redirect rather than augmenting the wrong one. - Manifest matcher is a list of matcher dicts under the correct key (
dhcp/ssdp/usb/homekit/mqtt); discovery fires when all items of any one matcher are present. Narrow generic OUI/bridge-chip matchers (e.g. add adescriptionmatch forvid: 10C4/pid: EA60). - Typed discovery step. Implement
async_step_dhcp(self, discovery_info: DhcpServiceInfo)/async_step_ssdp(... SsdpServiceInfo)/async_step_usb(... UsbServiceInfo)/async_step_homekit(... ZeroconfServiceInfo)/async_step_mqtt(... MqttServiceInfo)inconfig_flow.py. - Always confirm. Forward into a confirm step (
async_step_discovery_confirm+self._set_confirm_only()) beforeasync_create_entry— never an entry without user confirmation. - Own the update path. Set
await self.async_set_unique_id(<stable_id>)thenself._abort_if_unique_id_configured(updates={CONF_HOST: host}); never a second entry on re-discovery. Canonicalunique_idsource per mechanism: DHCP → MAC (format_mac), SSDP →udn, USB →serial_number(elsevid:pid), HomeKit → the accessory id, MQTT → a stable id from the discovery topic/payload. For DHCP IP-updates register the MAC (CONNECTION_NETWORK_MAC) and setregistered_devices: true; for MQTT addmqtttodependenciesandawait mqtt.async_wait_for_mqtt_client(hass)before subscribing. Add theconfig.abort.already_configured/already_in_progressstrings tostrings.json. - Name per
spec/ha/naming-conventions/en.mdand verify HA internals against the official docs (seespec/ha/upstream-docs-verification/en.md).
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 · 110 lines · 224 tokens per session scan A 3e91ca3851b3
ha-discovery-augment is a skill published in the GitHub repository nolte/claude-home-assistant (1 stars, last pushed 1mo ago), licensed MIT. It adds 224 tokens to every session and 2,094 once invoked, about $0.0011 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
authentication-patterns
OAuth 2.0, JWT, SSO, MFA, NextAuth/Clerk/Supabase Auth implementation patterns.
email-systems
Transactional email (Resend, SendGrid, SES), templates (React Email, MJML), deliverability (SPF/DKIM/DMARC), and inboxing best practices. Use when building email infrastructure, designing templates, or troubleshooting deliverability.
event-driven-architecture
Kafka, RabbitMQ, SQS/SNS, event sourcing, CQRS, saga patterns, dead letter queues, and idempotency. Use when designing asynchronous systems, implementing message-driven workflows, or building event streaming pipelines.
graphql-expert
GraphQL API design and implementation. Use when building GraphQL APIs, designing schemas, implementing resolvers, or optimizing GraphQL performance.
api-design
REST and GraphQL API design best practices including OpenAPI specs. Use when designing APIs, documenting endpoints, or reviewing API architecture.
generic-fullstack-feature-developer
Guide feature development for full-stack applications with architecture focus. Covers Next.js App Router patterns, NestJS backend services, database models, data workflows, and seamless integration. Use when adding new features, refactoring existing code, or planning major changes.