time-estimate

<!-- tested with: claude code v2.1.122 --> # /time-estimate produce a realistic CC active-time estimate for the given task, grounded in `plugins/cc/rules/time.md`. never guesses effort level. ## what to do when invoked 1. **resolve active effort**, highest-precedence first. stop at the first rung that returns a value: - `echo "${CLAUDE_CODE_EFFORT_LEVEL:-}"` (env var) - `--effort` flag, if visible via process context - `/effort` session state, if already set this session - `effortLevel` key in `~/.claude/settings.json`, then any project-scoped `.claude/settings.json` under the repo root - otherwise, model default cite the rung that supplied the answer. if ambiguous, ask the user rather than guess. 2. **resolve active model** from the session's model id (e.g. `claude-opus-4-7`). 3. **classify session mode** using the table in the rule: - quick fix: single-file edit, typo, config tweak - standard: feature work, refactor, small module - marathon: migration, infra build, multi-file refactor 4. **apply the multiplier** from the model × effort matrix in the rule. quote the baseline at `low`, then scale. 5. **return the estimate** in the rule's format: - `CC: <range>` (always a range, never a single number) - effort rung cited (e.g. "rung 4, settings.json") - session mode named - your-time estimate for review - confidence (high / medium / low) - risks that could 2× the number 6. if the task breaks into sub-phases, estimate each ...