{"skill":{"slug":"governance-policies","displayName":"AxonFlow Governance Policies","summary":"Govern OpenClaw with AxonFlow — block dangerous commands, detect PII, prevent data exfiltration, protect agent config files, explain policy decisions, grant...","tags":{"agent-security":"1.5.2","approvals":"1.5.2","audit":"1.5.2","compliance":"1.5.2","data-loss-prevention":"1.5.2","explainability":"1.5.2","governance":"1.5.2","human -in-the-loop":"1.2.0","human-in-the-loop":"1.5.2","latest":"1.5.2","llm-governance":"1.5.2","mcp":"1.5.2","openclaw":"1.5.2","overrides":"1.5.2","pii":"1.5.2","policies":"1.5.2","prompt-injection":"1.5.2","safety":"1.5.2","security":"1.5.2","sqli":"1.5.2"},"stats":{"comments":0,"downloads":225,"installsAllTime":0,"installsCurrent":0,"stars":1,"versions":9},"createdAt":1775211296455,"updatedAt":1776606108363},"latestVersion":{"version":"1.5.2","createdAt":1776605944133,"changelog":"1.5.2 (2026-04-19)\n\n  Major content expansion covering Plugin Batch 1 features shipped in @axonflow/openclaw v1.3.0 and v1.3.1.\n\n  1. New section - \"Understand a Block: Richer Context\" - documents the structured block response shape from platform v7.1.0+ (decision_id, risk_level, policy_matches[], override_available, override_existing_id)\n  so agents render specific block reasons instead of generic \"policy violation\" strings.\n  2. New section - \"Explain a Decision\" - covers client.explainDecision(id) with the frozen DecisionExplanation shape (ADR-043). Links to the Explainability docs page.\n  3. New section - \"Grant a Session Override\" - covers client.createOverride / revokeOverride / listOverrides with platform-enforced invariants: TTL clamping (60s to 24h, default 60m), critical-risk rejection,\n  allow_override=false rejection, mandatory justification, 4-event audit lifecycle. Links to the Session Overrides docs page.\n  4. New section - \"Common Workflows\" - canonical walkthroughs for: debug a block, grant a one-off allow, audit a session.\n  5. New section - \"When to use this skill\" - triggers that help the LLM decide to activate this skill (setup, debugging blocks, granting overrides, compliance audits, hardening).\n  6. New userEmail configuration guidance - required for override and explain endpoints in v1.3.1+. Added to the Configure block.\n  7. Community-mode config wording corrected - clientSecret can be left empty in community mode (the plugin's resolveConfig only defaults clientId to \"community\"; clientSecret stays the empty string unless set).\n  Prior phrasing had implied both defaulted to \"community\", which overstated the actual runtime behavior.\n  8. Updated description and tags - added explainability, overrides, sqli, prompt-injection, llm-governance, agent-security, and data-loss-prevention for better discoverability. Total tag count now 19 (up from\n  12), covering capability (explainability, overrides, sqli, prompt-injection, pii, mcp), category (governance, llm-governance, agent-security, compliance, security, safety, data-loss-prevention), process (audit,\n  approvals, human-in-the-loop, policies), and platform (openclaw, latest).\n  9. Removed the pre-2026.4.14 CLI workaround from the primary install flow - the upstream bug is fixed and the primary install command now works unconditionally. Minimum OpenClaw CLI required stated as 2026.4.14;\n   minimum plugin version stated as 1.3.1 (for X-User-Email forwarding).","license":"MIT-0"},"metadata":{"os":null,"systems":null},"owner":{"handle":"axonflow","userId":"s17d5ef7wxwj9v980ssptrc4e1845vky","displayName":"AxonFlow: Runtime control layer for production AI","image":null},"moderation":null}