Files
msd-core/get-shit-done/workflows/discuss-phase/modes/auto.md
Tom Boucher 48824f258c fix(#444): resolver preamble checks repo-local .claude install path (#476)
The gsd_run resolver preamble now probes
<repo-root>/.claude/get-shit-done/bin/${_GSD_SHIM_NAME} as the second
check — immediately after the existing get-shit-done/bin/ check and
before command -v / $HOME/.claude fallbacks. This covers the install
layout produced by npx @opengsd/get-shit-done-redux@latest --claude --local.

A _GSD_RUNTIME_ROOT variable is introduced to bind the repo-root
expression once and reuse it for both checks without repeating the
git rev-parse subshell.

76 workflow files regenerated via node scripts/sync-runtime-launcher.cjs.
All parity, size-budget, and new regression tests pass.

Co-authored-by: CI Rebase Check <ci@gsd-redux>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-05-29 17:12:54 -04:00

3.5 KiB

--auto mode — fully autonomous discuss-phase

Lazy-loaded. Read this file from workflows/discuss-phase.md when --auto is present in $ARGUMENTS. After the discussion completes, the parent's auto_advance step also reads modes/chain.md to drive the auto-advance to plan-phase.

Effect across steps

  • check_existing: if CONTEXT.md exists, auto-select "Update it" — load existing context and continue to analyze_phase (matches the parent step's documented --auto branch). If no context exists, continue without prompting. For interrupted checkpoints, auto-select "Resume". For existing plans, auto-select "Continue and replan after". Log every decision so the user can audit.
  • cross_reference_todos: fold all todos with relevance score >= 0.4 automatically. Log the selection.
  • present_gray_areas: auto-select ALL gray areas. Log: [--auto] Selected all gray areas: [list area names].
  • discuss_areas: for each discussion question, choose the recommended option (first option, or the one explicitly marked "recommended") without using AskUserQuestion. Skip interactive prompts entirely. Log each auto-selected choice inline so the user can review decisions in the context file:
    [auto] [Area] — Q: "[question text]" → Selected: "[chosen option]" (recommended default)
    
  • After all areas are auto-resolved, skip the "Explore more gray areas" prompt and proceed directly to write_context.
  • After write_context, auto-advance to plan-phase via modes/chain.md.

CRITICAL — Auto-mode pass cap

In --auto mode, the discuss step MUST complete in a single pass. After writing CONTEXT.md once, you are DONE — proceed immediately to write_context and then auto_advance. Do NOT re-read your own CONTEXT.md to find "gaps", "undefined types", or "missing decisions" and run additional passes. This creates a self-feeding loop where each pass generates references that the next pass treats as gaps, consuming unbounded time and resources.

Check the pass cap from config:

_GSD_SHIM_NAME="gsd-tools.cjs"; _GSD_RUNTIME_ROOT="${RUNTIME_DIR:-$(git rev-parse --show-toplevel 2>/dev/null || pwd)}"; GSD_TOOLS="${_GSD_RUNTIME_ROOT}/get-shit-done/bin/${_GSD_SHIM_NAME}"; if [ -f "$GSD_TOOLS" ]; then gsd_run() { node "$GSD_TOOLS" "$@"; }; elif [ -f "${_GSD_RUNTIME_ROOT}/.claude/get-shit-done/bin/${_GSD_SHIM_NAME}" ]; then GSD_TOOLS="${_GSD_RUNTIME_ROOT}/.claude/get-shit-done/bin/${_GSD_SHIM_NAME}"; gsd_run() { node "$GSD_TOOLS" "$@"; }; elif command -v gsd-tools >/dev/null 2>&1; then GSD_TOOLS="$(command -v gsd-tools)"; gsd_run() { "$GSD_TOOLS" "$@"; }; elif [ -f "$HOME/.claude/get-shit-done/bin/${_GSD_SHIM_NAME}" ]; then GSD_TOOLS="$HOME/.claude/get-shit-done/bin/${_GSD_SHIM_NAME}"; gsd_run() { node "$GSD_TOOLS" "$@"; }; else echo "ERROR: gsd-tools.cjs not found at $GSD_TOOLS and gsd-tools is not on PATH. Run: npx -y @opengsd/get-shit-done-redux@latest --claude --local" >&2; exit 1; fi
MAX_PASSES=$(gsd_run query config-get workflow.max_discuss_passes 2>/dev/null || echo "3")

If you have already written and committed CONTEXT.md, the discuss step is complete. Move on.

Combination rules

  • --auto --text / --auto --batch: text/batch overlays are no-ops in auto mode (no user prompts to render).
  • --auto --analyze: trade-off tables can still be logged for the audit trail; selection still uses the recommended option.
  • --auto --power: --power wins (power mode generates files for offline answering — incompatible with autonomous selection).