mythify-worklisted

> Invocation: type `/mythify-work` in Claude Code or `$mythify-work` in Codex to run this skill. Treat any text after it as the work request. # /mythify-work Run the requested work in this chat with Mythify as the durable evidence ledger. The user should see the work unfold here: what is happening, why it is happening, what passed, what failed, and what comes next. ## Contract 1. Keep execution in the initiating chat unless the user explicitly asks for a handoff, background run, or external agent. 2. Prefer MCP tools when available. Otherwise use the installed `mythify` launcher, falling back to `python3 scripts/mythify.py` from a Mythify checkout. 3. Start with a one-sentence visible outcome. 4. Run `mythify route "TASK"` to choose the workflow unless the user already named a specific Mythify primitive. 5. For multi-step work, create or resume a plan. New plans should use a 20-step lookahead by default: follow the `mythify route "TASK"` next command, or call `mythify plan create "TASK" --horizon 20` when creating the plan directly. Then mark the chat cursor: mythify report --cursor chat --mark 6. Before each step, say the step name and success criterion in chat. 7. Mark the step in progress, do the work, then run a chat report: mythify step ID in_progress mythify report --since last --cursor chat --format chat 8. Run an executable verifier before claiming completion: mythify verify run "COMMAND" --claim "CLAIM"