gm-oc 2.0.400 → 2.0.402

package/.github/workflows/publish-npm.yml

@@ -31,7 +31,14 @@ jobs:
           fi
 
           echo "Publishing $PACKAGE@$VERSION..."
-          npm publish --access public
+          npm publish --access public 2>&1 | tee /tmp/npm-out.log; EXIT=${PIPESTATUS[0]}
+          if [ "$EXIT" != "0" ]; then
+            if grep -q "cannot publish over\|previously published" /tmp/npm-out.log; then
+              echo "⚠️  Version already published, skipping"
+            else
+              exit "$EXIT"
+            fi
+          fi
           echo "✅ Published $PACKAGE@$VERSION"
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "gm-oc",
-  "version": "2.0.400",
+  "version": "2.0.402",
   "description": "State machine agent with hooks, skills, and automated git enforcement",
   "author": "AnEntrypoint",
   "license": "MIT",

package/skills/gm/SKILL.md

@@ -128,11 +128,13 @@ Completing a phase is NOT stopping. After every phase: read .prd, check git, inv
 
 ## CONSTRAINTS
 
-**Tier 0**: no_crash, no_exit, ground_truth_only, real_execution
+**Tier 0**: no_crash, no_exit, ground_truth_only, real_execution, fail_loud
 **Tier 1**: max_file_lines=200, hot_reloadable, checkpoint_state
 **Tier 2**: no_duplication, no_hardcoded_values, modularity
 **Tier 3**: no_comments, convention_over_code
 
-**Never**: `Bash(node/npm/npx/bun)` | skip planning | sequential independent items | screenshot before JS exhausted | narrate past unresolved mutables | stop while .prd has items | ask the user what to do next while work remains
+**FAIL LOUD — NO FALLBACKS**: Never create fallback modes, demo modes, graceful degradation, silent error swallowing, or try/catch that returns a default value instead of propagating the error. Every error must throw and propagate. If something fails, the user must see exactly what failed and why. No `|| defaultValue`, no `catch { return null }`, no "demo mode", no "offline mode", no degraded functionality. The only acceptable error handling is: catch → log the real error → re-throw or display to user.
+
+**Never**: `Bash(node/npm/npx/bun)` | skip planning | sequential independent items | screenshot before JS exhausted | narrate past unresolved mutables | stop while .prd has items | ask the user what to do next while work remains | create fallback/demo modes | silently swallow errors
 
 **Always**: invoke named skill at every transition | snake to planning on any new unknown | witnessed execution only | keep going until .prd deleted and git clean

package/skills/gm-emit/SKILL.md

@@ -74,6 +74,7 @@ Pre-emit revealing unexpected behavior → new unknown → snake to `planning`.
 - Crash-proof: catch at every boundary, recovery hierarchy
 - No mocks/fakes/stubs anywhere
 - Files ≤200 lines, no duplicate code, no comments, no hardcoded values
+- No fallbacks, demo modes, or silent error swallowing — every error must throw and propagate
 - CLAUDE.md reflects actual behavior
 
 ## CODEBASE EXPLORATION
@@ -99,7 +100,7 @@ Never respond to the user from this phase. When all gate conditions pass, immedi
 
 ## CONSTRAINTS
 
-**Never**: write before pre-emit passes | advance with post-emit variance | absorb surprises silently | comments | hardcoded values | defer spotted issues | respond to user or pause for input
+**Never**: write before pre-emit passes | advance with post-emit variance | absorb surprises silently | comments | hardcoded values | fallback/demo modes | silent error swallowing | defer spotted issues | respond to user or pause for input
 
 **Always**: pre-emit debug before writing | post-emit verify from disk | snake to planning on any new unknown | fix immediately | invoke next skill immediately when gates pass