Entity Mission Control Bootstrap

Live Operations Advanced External Bundle v1.3.2

Turn a loose cluster of agents into a repeatable ops surface. Installs the Entity Mission Control helper runtime for a crew of agents — shared scripts, structured task intake, per-agent manifests, and supervised rollout.

Canonical source: github.com/henrino3/enterprise-crew-skills/entity-mc

Install

git clone https://github.com/henrino3/enterprise-crew-skills.git /tmp/enterprise-crew-skills
mkdir -p skills
cp -R /tmp/enterprise-crew-skills/entity-mc skills/entity-mc
bash skills/entity-mc/install-auto.sh

This clones the bundle, copies it into your workspace, and runs the auto installer. The installer:

Creates an auto manifest for the current workspace
Installs wrapper scripts
Installs portable MC/intake setup context
Writes auto-pull and stall-check cron entries
Runs verification

Verify

bash skills/entity-mc/verify.sh --manifest skills/entity-mc/manifests/auto.env

Per-Agent Install

After the auto install validates locally, apply per-agent manifests to each target host:

bash skills/entity-mc/install.sh --manifest skills/entity-mc/manifests/<agent>.env

Then verify on the target:

bash skills/entity-mc/verify.sh --manifest skills/entity-mc/manifests/<agent>.env

What You Get

skill

SKILL.md

Defines the bootstrap workflow, shared runtime assumptions, and rollout guardrails.

bundle

Canonical helper runtime

mc.sh, mc-auto-pull.sh, mc-assign-model.sh, mc-build-context.sh, mc-stall-check.sh, mc-intake.sh — shared across all agents.

script

mc-intake.sh

Conservative JSON/JSONL intake helper. Dedupes and creates MC tasks from explicit structured signals.

doc

Portable MC operating context

Public-safe MC operating memory: task lifecycle, review/blocker contract, evidence rules, runtime behavior. Injected by mc-build-context.sh.

doc

Onboarding flow

End-to-end explanation of install, default crons, portable context, intake policy setup, watcher JSONL flow, and verification.

manifest

Per-agent manifests

Agent-specific .env manifests that map the canonical runtime onto each operator without duplicating scripts.

script

install.sh / verify.sh / rollback.sh

Idempotent install, verification, and rollback helpers for supervised rollout.

Verification Steps

Source bundle review

Confirm the linked GitHub bundle exposes SKILL.md, source-scripts/, context/, manifests/*.env, install.sh, verify.sh, rollback.sh, and VERSION.

Bundle install present

test -f skills/entity-mc/SKILL.md && ls skills/entity-mc

Expected: The installed workspace path contains SKILL.md, source-scripts/, context/mc-intake-setup.md, manifests/auto.env, install-auto.sh, install.sh, verify.sh, rollback.sh, and VERSION. The crontab contains exactly one ENTITY_MC block for the agent.

Runtime scripts present

ls scripts/mc*.sh

Expected: All six runtime scripts are present and executable (mc.sh, mc-auto-pull.sh, mc-assign-model.sh, mc-build-context.sh, mc-stall-check.sh, mc-intake.sh).

Portable context present

ls .entity-mc/context/*.md
bash scripts/mc-build-context.sh '{"task_id":"0","task_name":"context smoke","task_description":"smoke","skill":"none","context":""}' | grep "Entity MC Operating Context"

Expected: Context files exist and the generated prompt includes "Entity MC Operating Context".

Intake dry run

bash scripts/mc-intake.sh create --title "Entity MC intake smoke" --description "Dry run" --assignee Ada --dry-run

Expected: JSON output with action=dry_run and a payload containing metadata.intake=true.

Requirements

OpenClaw runtime — host needs a functioning OpenClaw environment with skill install access
MC connectivity — target agents must reach the Mission Control API
Operator review — validate hostnames, agent ids, and manifest selection before rollout

Current Limits

OpenClaw 2026.5.4 rejects github:owner/repo/path skill source specs — use the clone/copy/install-auto fallback above
Auto-pull and stall-check crons install by default; structured intake requires explicit enablement with --enable-intake true because it needs an inbox policy
Portable context is intentionally small — host-specific secrets, private memory, and local paths belong in manifests or local memory, not the public bundle
Auto-pull executes existing tasks; automatic task creation requires explicit structured input or a source-specific watcher feeding mc-intake.sh

Operator Notes

Verify the linked source layout before install
Verify one host first before rolling the bundle out everywhere
Keep the previous helper runtime accessible until the new manifest proves healthy
Compare generated agent state against the canonical runtime to catch drift