Hephaestus Cloud Routing (my own cloud / 보관함)
SkillDev toolsOnce agentlas-cloud is added, your AI can draw abilities only from your private Cloud packages, tied to the account you are signed in with. It is a dev-tool skill that keeps package selection limited to your own Cloud. You start it any time by typing /agentlas-cloud, /agentlas cloud, or /hep-cloud.
Available today. Use it from your connected AI after setup.
No other account needed.
After adding the skill, type /agentlas-cloud, /agentlas cloud, or /hep-cloud in a chat. Your AI will then work only from your private Cloud packages.
Then ask your AI: use the Hephaestus Cloud Routing (my own cloud / 보관함) skill
What your AI can do with it
- Pick abilities only from the signed-in owner's private Cloud packages
- Keep selections within your own Cloud account rather than other sources
- Start working when you type /agentlas-cloud, /agentlas cloud, or /hep-cloud
What this skill tells your AI
The instructions your AI receives, as published by agentlas-ai/agentlas-os in skills/agentlas-cloud/SKILL.md and read by ahel’s review.
Route the request through the signed-in user's OWN Agentlas cloud packages only. The active host LLM remains the staffing decision-maker; Cloud supplies a content menu and exact BYOM releases.
0. Scope rule
/hep-cloud is owner-scoped: it queries ONLY the authenticated owner's
own cloud packages (보관함) through Core's typed sourceScope: "cloud". It does not
search the public marketplace and does not search local private/plugin
cards.
- The user's own Cloud packages are restorable/owned by them. Entitlement, lease, and charged credits are server-authoritative; do not hard-code a price.
- For the public marketplace only, use
/hep-hub(sourceScope: "hub"). - For the combined Local + own Cloud + public Hub menu, use
/hep-network(sourceScope: "network").
Each unpinned Cloud discovery goes to the owner-scoped source for its current menu; a successful client cache entry must not hide a newly published, replaced, or removed package. Once the host validates and prepares a choice, the exact Cloud release remains pinned for that execution.
1. Resolve the runner
Run this resolution in a shell and use the first hit:
RUNNER=""
for c in \
"$HOME/.agentlas/runtime/current/bin/hephaestus" \
./bin/hephaestus
do [ -x "$c" ] && RUNNER="$c" && break; done
if [ -z "$RUNNER" ]; then
for cache in \
"$HOME/.claude/plugins/cache/agentlas-core-engine/hephaestus" \
"$HOME/.codex/plugins/cache/agentlas-core-engine/hephaestus"; do
newest="$(ls -d "$cache"/*/bin/hephaestus 2>/dev/null | sort -V | tail -1)"
[ -n "$newest" ] && [ -x "$newest" ] && RUNNER="$newest" && break
done
fi
If no runner exists, tell the user to run the one-touch installer:
curl -fsSL https://raw.githubusercontent.com/agentlas-ai/Agentlas-OS/main/scripts/install-all-runtimes.sh | bash
If shell execution is unavailable but the local hephaestus-network MCP is
available, use the typed Workforce sequence in section 3. If that server cannot
advertise owner-Cloud search plus exact bundle fetch, report
source_not_supported; never fall back to the legacy cargo search path.
1.5 Core project first-contact contract
The cloud ... --project . call below is a trusted plugin contact. Agentlas
Core must synchronously create or repair the same private project soul memory,
code map, ontology runtime, CareerGraph, and full .agentlas/ ignore block used
by every other host. If bootstrap is blocked, stop rather than querying Cloud
without the project architecture. The adapter never owns a second seed format.
2. Agentlas sign-in (required)
The owner cloud is sign-in-gated. Before routing, ensure Agentlas is signed in:
if [ "${HEPHAESTUS_AUTH_AUTOPOPUP:-1}" != "0" ]; then
"$RUNNER" auth ensure >/dev/null 2>&1 || true
fi
This opens the user's default browser only when there is no valid local sign-in
yet, and reuses a saved sign-in silently. For CI/headless checks only, set
HEPHAESTUS_AUTH_AUTOPOPUP=0 and skip this step.
3. Staff from owner Cloud only
In an MCP-capable host, author the complete redacted WorkOrder, then call the actual local Core tools in this order:
workforce.search_candidates(sourceScope="cloud")
workforce.validate_selection(workOrder=..., selection=...)
workforce.prepare_execution(workOrder=..., selection=..., federatedSelection=..., projectDir=..., goalId=activeGoalId?)
workforce.validate_execution_receipt(receipt=..., executionPlan=..., toolInventory=...)
The host LLM authors the final Selection. If the deployed owner-Cloud Workforce
source contract is absent, report source_not_supported; do not silently query
public Hub or legacy cargo search. A shell without an active host LLM can only
report that orchestration is required.
The default search response is a projected menu. Preserve its Cloud source
receipt and selectionSessionId, but do not echo it as a complete
federationResult; Core resolves the full pinned result by session. Receipt
validation is local and read-only and cannot create execution evidence.
4. Act on the typed result (scope: "cloud")
Preserve the Cloud source receipt, source selection session, candidate-set digest, release id, package hash, and content digest. Validate and prepare in local Core; never send the federated wrapper back to remote validation.
5. Hard rules
- Never report public marketplace agents or local private/plugin cards as if they were the user's own cloud packages.
- Deterministic Core validates governance and immutable pins but never chooses the roster. The active host LLM chooses from content evidence.
- For actual tool execution, follow the host runtime's safety and permission
model. Report the exact source receipt plus
selectionReceiptId,preparationReceiptId, and validated execution receipt status when present.
Signals
- GitHub stars
- 1k
- Forks
- 103
- Last commit
- Sep 2026
ahel review
K1info
remote-installer-piped-to-shell
Automated review, not a security audit. Ruleset v1+k2.
Others that do the same job
Advanced
- Catalog kind
- skill
- Gateway key
agentlas-cloud- Source
- github.com/agentlas-ai/agentlas-os