Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add gtapps/claude-code-hermit/plugin install claude-code-homeassistant-hermitWrote 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/gtapps/claude-code-hermit/ha-snapshot-restore)<a href="https://agentmods.dev/skills/gtapps/claude-code-hermit/ha-snapshot-restore"><img src="https://agentmods.dev/badge/skills/gtapps/claude-code-hermit/ha-snapshot-restore/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/gtapps/claude-code-hermit/ha-snapshot-restore"><img src="https://agentmods.dev/badge/skills/gtapps/claude-code-hermit/ha-snapshot-restore.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 | $0.00069 | $0.00632 |
| Opus 5 | $0.00034 | $0.00316 |
| Sonnet 5 | $0.00014 | $0.00126 |
| Haiku 4.5 | $0.00007 | $0.00063 |
Grade A, and why
ha-snapshot-restore 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 5d 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
HA Snapshot / Restore
Purpose
Make changes reversible: capture an entity set's current state before testing a new automation or applying a risky change, then restore those exact states if it misbehaves. Capture is read-only; restore actuates devices in one scene.apply call.
Capture (read-only)
${CLAUDE_PLUGIN_ROOT}/bin/ha-agent-lab ha snapshot-states --name <label> [--domains light,cover,climate,switch] [--entities <id> ...]
--namelabels the snapshot (defaultsnapshot); it becomes part of the artifact filename.--domainsselects which domains to capture (comma-separated; defaultlight,cover,climate,switch).--entitiescaptures an explicit list instead of whole domains.- Writes
.claude-code-hermit/raw/snapshot-ha-states-<label>-latest.json(plus a timestamped copy). Only restore-relevant attributes are stored.
Restore (actuation — gated)
${CLAUDE_PLUGIN_ROOT}/bin/ha-agent-lab ha restore-states <artifact> [--confirm]
Restore is the plugin's one direct device-actuation path, so it runs through the same ha_safety_mode policy as every other actuation:
- strict (default): if the snapshot contains any sensitive entity (lock, alarm, security-keyworded cover/switch), restore is blocked and exits non-zero with
blocked:trueand a suggestion to surface it as a proposal. Non-sensitive entities (lights, climate) restore normally. - ask: a snapshot touching sensitive entities requires
--confirm. Without it, the command refuses and reportsneeds_confirm:true. The main session must confirm with the operator before re-running with--confirm— the CLI never prompts. - A successful restore writes an audit report under
.claude-code-hermit/raw/audit-ha-restore-*.
Output contract
Capture prints { ok, name, captured, entities, report_path, message }.
Restore prints { ok, blocked, needs_confirm, applied, entities, sensitive, reason, ... }. Exit code is 0 only when ok is true.
Failure modes
- HA unreachable → CLI exits non-zero with an error message.
- A snapshot with no matching entities → capture reports
captured: 0; restore reports an error. - Restore replays state + captured attributes via
scene.apply; for cover position and climate setpoints the result depends on what HA's scene integration can reproduce — verify the outcome rather than assuming an exact match.
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.
- 5d ago First seen · 49 lines · 69 tokens per session scan A 9563936fee6a
ha-snapshot-restore is a skill published in the GitHub repository gtapps/claude-code-hermit (74 stars, last pushed yesterday), licensed MIT. It adds 69 tokens to every session and 632 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 2026-09-03.
Other skills, from other repositories
kicad-jlcpcb-workflow
Use this skill when the user asks to "design a PCB", "make a board", "build a PCB project", "order from JLCPCB", "create a schematic", "source LCSC parts", "wire up a board", "generate a PCB from a BOM", or any KiCad + JLCPCB workflow. Also use when working with .kicadpro or .kicadpcb files. Phase 1.6 handles project…
especialista-em-engenharia-de-hardware
Especialista em Engenharia de Hardware. Use para design eletrônico, microcontroladores, PCB, firmware embarcado, sinais, energia e prototipagem. Palavras-chave: hardware, eletrônica, microcontrolador, PCB, embarcado, firmware, Arduino.
qector-hardware-engineer
Physical-layer quantum circuits, detector-error-model preparation, qLDPC reasoning, and honest target-device hardware reporting for QECTOR. The bundled library MCP server is app-free; the companion bench server adds DEM inspection, DEM collapse, code-family introspection, and hardware probes (all Provisional).…
jensen-skills
Invoke Jensen Huang — Hardware & AI Infrastructure. Use for AI infrastructure planning, GPU and data center strategy, compute TCO analysis, and build-vs-buy decisions. Sets Claude into the Jensen Huang persona for the current conversation.
actual-setup
Set up Actual Computer (actual.inc) inference in Hermes.
cco-tools
Show what tools actually cost in tokens — learned per-tool averages from observed results, replacing the hardcoded MCP/Agent guesses.