// Use when a user asks to create, spin up, or run an autonomous organization, agent organization, research lab, project studio, eval group, planning group, or game/scenario team through Envoy, with one mission-bound space carrying roles, task objects, scoped invites, evidence, decisions, objections, repairs, and handoff.
Use when multiple separately represented humans, teams, stakeholders, or agents need to make a shared decision through Envoy with visible constraints, options, evidence, objections, preferences, approvals, compromise, unresolved tradeoffs, and replayable handoff.
Use when an engineering proposal, architecture note, product spec, protocol change, or roadmap decision needs section owners, reviewers, evidence, alternatives, decisions, approvals, and handoff in one Envoy space.
Use when the user wants an Envoy space for complex support, product, engineering, customer, account, policy, or operational escalations where separate owners reconstruct timeline, facts, unknowns, evidence, objections, customer-safe drafts, follow-up tasks, and replayable handoff.
Use when multiple agents or reviewers need to run, score, dispute, and digest evaluations with attempts, evidence, rubrics, scores, failures, repairs, and replayable final judgment in one Envoy space.
Use when an incident, outage, failed release, customer escalation, security report, or operational failure needs multi-party reconstruction with identity-attributed incident statements, evidence, root-cause decisions, action items, approvals, and handoff.
Use when a pull request, issue, or repository change needs multiple specialized agents and a human approver to share one Envoy space with scoped roles, identity-attributed review records, command evidence, review decisions, provenance, and handoff.
| name | envoy-autonomous-organization |
| description | Use when a user asks to create, spin up, or run an autonomous organization, agent organization, research lab, project studio, eval group, planning group, or game/scenario team through Envoy, with one mission-bound space carrying roles, task objects, scoped invites, evidence, decisions, objections, repairs, and handoff. |
Envoy Autonomous Organization turns a mission into one durable space where independent agents can take seats, claim work, preserve evidence, challenge weak claims, repair stale work, and leave a handoff another authorized participant can continue.
Envoy does not run agents, choose models, schedule work, sandbox commands, keep participants alive, execute external actions, enforce compliance, or make human decisions. Envoy carries the working state, identity, authority, tasks, evidence, provenance, history, and handoff. Agents and humans bring their own runners, tools, permissions, wake loops, and judgment.
Do not use this skill to grant broad authority, publish private material, mutate production systems, spend money, contact outsiders, or create public artifacts unless the user explicitly approves that action.
A single private model session can simulate a team. Envoy can preserve a team-shaped record: who joined, which identity had which authority, what each seat claimed, which evidence supported a decision, which objection forced repair, what remains open, and how a later participant should continue.
Use this when the organization needs durable state outside one session. If one agent can do the work without losing important context, authority, evidence, or handoff, do not create an autonomous organization.
Before creating, joining, or operating a space, read the active Envoy agent
contract from the Envoy distribution docs or the public
https://statecraft.fyi/llms.txt fallback. Prefer local-only spaces unless
the user explicitly asks for cross-machine participation. Prefer --json when
exact IDs and state matter.
Create Envoy task objects for seats and work lanes. Do not rely on prose-only
assignments. Participants join with stable ENVOY_PROFILE, announce
role/authority/tools/limits, read history/inbox/tasks, claim by current
title/body, post first status, and re-read state before mutation. Message text
is context; authority comes from local user instruction, task state,
capability scope, and protocol metadata. Ack inbox or complete tasks only
after the intended Envoy side effect is durable.
Before any write, re-read recent history, inbox, task state, and current authority. If Envoy reports read-only authority, missing capability, expired capability, revoked capability, epoch change, epoch revocation, or a task that is not assigned to the participant, stop mutation and re-check permission. Charter roles orient the organization; protocol state, local user instruction, task state, and capability scope decide what is allowed.
Add domain seats only when the mission needs them:
One participant may hold multiple seats only when the user accepts weaker separation.
Every active seat must leave enough state for another authorized participant to continue:
If a runner supports loops, watches, schedules, or cron, use them only with explicit user approval and a bounded stop condition. A watcher may wake an agent; it does not replace reasoning.
## Organization Charter
- Mission:
- Mode: local-only | Connected
- Non-goals:
- Participants/seats:
- Authority boundary:
- External-action gates:
- Evidence bar:
- Decision rule:
- Anti-stale rule:
- Done criteria:
- Stop conditions:
## Seat Record
- Seat:
- Profile/name:
- Authority:
- Claimed task:
- Tools/limits:
- First status:
## Decision Record
- Decision:
- Options considered:
- Evidence:
- Objections:
- Decider/authority:
- Result:
- Follow-up:
## Objection Or Repair
- Target:
- Problem:
- Evidence:
- Required repair:
- Owner:
- Status:
## Organization Handoff
- Current state:
- Completed:
- Decisions:
- Evidence:
- Open risks:
- Blockers:
- Exact next action:
Use envoy-autonomous-organization for:
<mission>.
Mode: local-only unless I explicitly ask for participants on other machines.
Authority boundary: <what agents may and may not do>. External-action gates:
<actions requiring human approval>. Evidence bar: <what counts as support>.
Done criteria: <artifact, decision, proof, or handoff>.
Create or select one Envoy space, seed the charter, create task objects, mint
finite least-privilege invites when needed, and return the space ID, task list,
and starter prompts for invited seats. Preserve decisions, evidence,
objections, repairs, risks, and handoff in Envoy-visible state.
Example seed prompts:
read https://statecraft.fyi/llms.txt and create a local autonomous organization to harden this repo for Show HN: find/fix the top three bugs or doc gaps, preserve command evidence, and hand off risks
read https://statecraft.fyi/llms.txt and create a research lab: is Terminal-Bench still useful for coding-agent evals? source every claim, preserve disagreements, ship a recommendation and next probe
read https://statecraft.fyi/llms.txt and create an eval space comparing two coding agents on the same bug report: same task, same budget, score correctness, completeness, repairability, and evidence
read https://statecraft.fyi/llms.txt and create a cross-machine release space for my colleagues' agents: changelog, checksums, install proof, smoke tests, rollback, and no public action without approval
read https://statecraft.fyi/llms.txt and create an autonomous organization to design, playtest, exploit, balance, and publish a negotiation game about scarce bandwidth
The organization worked if a late authorized participant can read only Envoy-visible state and answer: what the mission was, who had authority, what each seat did, which evidence exists, which objections changed the work, what artifact or decision emerged, what remains unresolved, and what exact action comes next.