ActionBoard-V5-AgentFormation

Plugin actionboard-v5-agentformation from marketplace actionboard-v5-agentformation in Cloudscockpit/actionboard-build-mission-pipeline. Commands read from the repo's manifest — 5 days ago.

Skills (12)

• actionboard-devops-mission

  Run a maturity-gated ActionBoard DevOps mission — discover and register deployment actions, classify them by risk, score the five formation stages per run, and enforce the 90% formation gate before any action executes autonomously. Use this skill whenever the user asks to run a mission, deploy or sh

• browser-actions

  Guide for using Claude's browser action feature (in-app Browser pane and Claude in Chrome) within ActionBoard V5 missions. Use when the user asks "can you browse", "open this site", "test this page", "how do browser actions work", or when a mission needs live web navigation, form filling, screenshot

• mission-brief

  Write a ActionBoard V5 Mission Brief (objective, scope, out of scope, success criteria, constraints) as Markdown plus a schema.org JSON-LD knowledge graph. Use when the user asks to "write a mission brief", "draft the brief", "scope this mission", or at Step 1 of the ActionBoard V5 mission flow, and

• mission-harness

  Build the sandbox harness for a ActionBoard V5 mission — one isolated environment per assigned Agent, sized and policed by role, created only after go/no-go. Use when forming ActionBoard V5 for real execution.

• mission-plan

  Write a ActionBoard V5 Mission Plan (Agent Assignments with file ownership, Skills Gap analysis, Risk Register) as Markdown plus a schema.org JSON-LD knowledge graph. Use when the user asks to "write the mission plan", "draft assignments", "show the risk register", or at Steps 3-5 of the ActionBoard

• mission-report

  Write a ActionBoard V5 Mission Report recording what each Agent shipped, blocked on, or deferred, as Markdown plus a schema.org JSON-LD knowledge graph. Use when the user asks to "write the mission report", "summarize the mission", "what shipped", or after Agents report back at the end of the Action

• openshell-admin

  Provision and govern NVIDIA OpenShell sandboxes, workspaces, and policies so agents get an isolated environment. Use for sandbox setup, policy denials, and teardown.

• policy-widen

  Triage an OpenShell denial and add the narrowest policy rule that fixes it. Use when a sandboxed agent is blocked from a host, path, or credential.

• sandbox-down

  Stop or delete OpenShell sandboxes for a finished mission, optionally archiving the workspace first. Use when a mission or ActionList completes.

• sandbox-status

  Show OpenShell sandboxes for a mission, tenant, or Agent, with phase, policy revision, and recent denials. Use to check whether an agent environment is ready or being blocked.

• sandbox-up

  Provision an OpenShell sandbox for a Agent or a usecase, wait for Ready, and hand back the exec prefix. Use when an agent needs an isolated environment to run in.

• skills-registry

  ActionBoard V5 AgentFormation skills registry and actions map. Use during Skills Gap analysis (Step 4 of the mission flow) to map mission capabilities to registered action types, responsible Agents, and required tools/skills. Also use when the user asks "what can the Agents do", "list actionboard-v5

Slash commands (1)

• /actionboard-v5-agentformation:start-mission

  Form the Mission Formation — Black Agent drafts a Mission Brief, Agent Assignments, Skills Gap analysis, and Risk Register, then dispatches the four Agents after your go/no-go approval.

Subagents (6)

• black-agent

  Mission commander (Black Agent) that decomposes an objective into a Mission Brief, Agent Assignments with file ownership, Skills Gap analysis, and Risk Register. Coordinates Red/Blue/Green/Yellow Agents and gates execution on user approval. Use when the user runs /start-mission or asks ActionBoard V

• blue-agent

  Data and integrations specialist (Blue Agent). Handles APIs, data plumbing, persistence, external service wiring. Use only when dispatched by black-agent with a concrete assignment including file-ownership boundaries.

• green-agent

  Recon specialist (Green Agent). Inventories the project and surfaces the code map for Black Agent. Prefers querying ./graphify-out/ when fresh — that is 10-50x cheaper than filesystem walks. Falls back to filesystem only when no graph exists or the graph is stale. Read-only with skill-invocation. Us

• red-agent

  Rapid execution specialist (Red Agent). Writes code, scaffolds features, ships implementations. Use only when dispatched by black-agent with a concrete assignment that includes file-ownership boundaries.

• sandbox-warden

  Owns OpenShell sandbox lifecycle — provisioning, policy iteration, and teardown — in an isolated context so environment plumbing never pollutes a Agent's mission context. Use for any multi-sandbox operation.

• yellow-agent

  Quality and defense specialist (Yellow Agent). Writes tests, runs verification, reviews for security and correctness issues. Use only when dispatched by black-agent, typically AFTER Red/Blue Agents have completed their builds.

Hooks (2)

• PreToolUse

  ${CLAUDE_PLUGIN_ROOT}/hooks/guard-openshell.sh

• PostToolUse

  ${CLAUDE_PLUGIN_ROOT}/hooks/audit-policy.sh

Similar plugins