Menu
Show or set the autonomy dial — how much the AI may do without asking (consent grade 1–4, orthogonal to the capability level).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Domain Engineering diagnostic — shows what the classifier decides for an objective (CMIS/DAS/profile/skills/mode). Observation-only; never mutates or blocks.
Business-driven methodology entry point — classify (intake), create/advance an Operation or Business work context, and drive the intake → operation → nested-workflow flow. Host-neutral; dry-run by default.
Workflow Navigator — shows the current phase, deliverables, and next commands for an ADR-0057 workflow. Read-only; never mutates state.
Start a focused session on one objective — locks scope, blocks opportunistic refactors.
Deterministic QA gate (ADR-0055) — run the project suite; green + complete acceptance criteria ⇒ qa-approve testing cards into conclusion; red ⇒ report (and qa-reject only attributable failures).
L6 — autonomous feature pipeline. Drives the full squad: design → implement → review → test → log. Checkpoints can be manual or automatic.
| name | source-command-setup-autonomy |
| description | Show or set the autonomy dial — how much the AI may do without asking (consent grade 1–4, orthogonal to the capability level). |
Use this skill when the user asks to run the migrated source command autonomy.
Show or change how much the AI may do without asking in this project. This is
NOT the capability level (/context-level, L1–L7 — what the kit can do); the
dial is consent — what it may do on its own.
Run the script and relay its output verbatim (it echoes the first-person consequence text the user must see):
node contextkit/tools/scripts/autonomy.mjs # show current grade + what each grade means
node contextkit/tools/scripts/autonomy.mjs 3 # persist grade 3
node contextkit/tools/scripts/autonomy.mjs 3 --session # this session only (auto-expires in 8h)
node contextkit/tools/scripts/autonomy.mjs --clear # drop the session override
# grade 4 (EXPERIMENTAL, ADR-0045) — gated by the eligibility bar, SESSION-scoped by default:
node contextkit/tools/scripts/autonomy.mjs 4 # session-only grade 4 (refuses if the bar fails)
node contextkit/tools/scripts/autonomy.mjs 4 --persist --confirm # persist grade 4 (only after seeing the consequence)
node contextkit/tools/scripts/autonomy-readiness.mjs # measure self-coverage + attribution for the bar (EXPENSIVE)
Rules for you (the agent):
grade-change → always human, ADR-0042). You run this command only when
the user asks; you may suggest a grade change, never apply one.resolveAutonomy, the single read
path): 1 manual · 2 suggest+supervise · 3 auto-except-decisions
(default, ADR-0058 — ADRs, pushes, high-risk and secret paths still come to
the user) · 4 full-auto, experimental, telemetry/budget-gated (ADR-0045).--confirm is the [y] signature; omitting it is [n]/cancel).
/autonomy 4 with no flags is session-scoped (auto-reverts); persisting
needs the --persist --confirm signature after the disclaimer is shown. The
self-coverage + attribution criteria come from autonomy-readiness.mjs — tell
the user to run it if those are the blockers. Even at grade 4 the floor holds
and /ship runs the hardened quorum (see /ship): a Critical from the
security voice is a veto, and the human can yield/kill mid-run at any step.