harnessloop-continuelisted

# Harnessloop Continue Run the Harnessloop continuation gate before any execution. This skill may continue only through an allowed next action; it must not bypass evidence, intake, control, self-audit, or human-confirmation requirements. ## Input Contract Accept an explicit skill invocation such as `$harnessloop-continue`, or natural language asking to resume/advance a Harnessloop task. Treat `harnessloop:continue` and `harnessloop continue` as natural-language aliases only; `$harnessloop:continue` is not a valid skill invocation. Useful input includes: - `target-project`: defaults to the current working directory. - Optional requested next action. - Optional human decision or unblock record. - Optional evidence, tool, or external-system confirmation needed by the current control gate. - Optional blocker classification, recovery eligibility, and safe read-only investigation target. If `.harnessloop/` is missing, stop and suggest `$harnessloop-init`. If imported intake work is pending, route to `$harnessloop-intake`. If `.harnessloop/` exists and `check_setup.py` reports `gate_blocking: true` (a core policy file — environment/control-contract/cost-context-policy — is still `template` or `missing`), stop and return `needs-setup` before evaluating any other gate (see Processing Contract step 1); suggest `$harnessloop-setup`. If `gate_blocking` is `false` but `complete` is `false` (non-blocking gaps remain, e.g. in `data-sources.md` or acknowledged via TODO), do not stop;