hcs-operation-prooflisted

# Skill: Render an operation proof Use any time an agent produces a command suggestion or operation proposal that a human will review. This is the structural antidote to "plausible but wrong CLI advice" — if the proof can't be filled in, the proposal isn't ready. ## Inputs - The operation intent (semantic, e.g., "restart this launchd service") - Current session context (cwd, resolved toolchain if known) - Available evidence (tool-resolve output, help cache, policy classification) ## Procedure 1. Read `docs/host-capability-substrate/operation-proof.md` for the canonical template. 2. Populate each section in order. **Do not skip any section.** If a section is unavailable, write "not available" with a one-line reason; do not omit. ## Template ```markdown ### Operation {semantic operation name} ### Host context - OS: {version} - cwd: {path} - Workspace: {id} - Shell mode: {login|non_interactive|interactive} - Resolved tool: {path}@{version} ### Evidence - Source: {command or doc} - Observed at: {ISO-8601 timestamp} - Parser version: {version} - Cache status: {hit|miss|stale} - Confidence: {authoritative|high|best-effort} ### Proposed invocation ```json {"command_mode": "argv", "file": "...", "argv": [...], "env_profile_id": "...", "lane": "..."} ``` ### Risk - Mutation scope: {none|write-local|write-project|write-host|write-destructive} - Target resources: {list} - Policy tier: {tier} ### Preflight {validation command, or "not available: <reason>"} ### Preview {dry-r