autonomy-readinesslisted

# Autonomy readiness - go/no-go review Review an autonomous workflow before it is activated. This skill is a gate review, not an implementation shortcut. ## Readiness dimensions - **Authority:** who approved the workflow, what it may do, and what it must not do. - **Scope:** objective, trigger, stop condition, and maximum runtime. - **Action policy:** read/write boundary, irreversible actions, external services, secrets, network, and code execution. - **Evidence:** dry runs, tests, smoke checks, evals, observed burn-in, and reviewer verdicts. - **Controls:** rollback, pause/kill switch, rate limits, budget limits, audit logs, alerts, and failure classification. - **Context hygiene:** prompt-injection containment, progressive disclosure, source provenance, and memory/update rules. - **Ownership:** who reviews outputs, who merges or tags, and what requires operator approval. ## Workflow 1. State the proposed autonomous behavior in concrete terms. 2. Identify every action the workflow can take without a live human. 3. Classify risk by blast radius, reversibility, cost, and credential exposure. 4. Check that each readiness dimension has evidence, not just assertions. 5. Require a dry-run or bounded pilot when evidence is weak. 6. Return one verdict and list blocking conditions. ## Verdicts - **READY:** evidence and controls are sufficient for the stated scope. - **READY-WITH-CONDITIONS:** safe only if named conditions are met first. - **NOT-READY:** material gaps make acti