claudecode-omc 5.5.2 → 5.6.0

package/bundled/upstream/ecc/commands/santa-loop.md

@@ -0,0 +1,175 @@
+---
+description: Adversarial dual-review convergence loop — two independent model reviewers must both approve before code ships.
+---
+
+# Santa Loop
+
+Adversarial dual-review convergence loop using the santa-method skill. Two independent reviewers — different models, no shared context — must both return NICE before code ships.
+
+## Purpose
+
+Run two independent reviewers (Claude Opus + an external model) against the current task output. Both must return NICE before the code is pushed. If either returns NAUGHTY, fix all flagged issues, commit, and re-run fresh reviewers — up to 3 rounds.
+
+## Usage
+
+```
+/santa-loop [file-or-glob | description]
+```
+
+## Workflow
+
+### Step 1: Identify What to Review
+
+Determine the scope from `$ARGUMENTS` or fall back to uncommitted changes:
+
+```bash
+git diff --name-only HEAD
+```
+
+Read all changed files to build the full review context. If `$ARGUMENTS` specifies a path, file, or description, use that as the scope instead.
+
+### Step 2: Build the Rubric
+
+Construct a rubric appropriate to the file types under review. Every criterion must have an objective PASS/FAIL condition. Include at minimum:
+
+| Criterion | Pass Condition |
+|-----------|---------------|
+| Correctness | Logic is sound, no bugs, handles edge cases |
+| Security | No secrets, injection, XSS, or OWASP Top 10 issues |
+| Error handling | Errors handled explicitly, no silent swallowing |
+| Completeness | All requirements addressed, no missing cases |
+| Internal consistency | No contradictions between files or sections |
+| No regressions | Changes don't break existing behavior |
+
+Add domain-specific criteria based on file types (e.g., type safety for TS, memory safety for Rust, migration safety for SQL).
+
+### Step 3: Dual Independent Review
+
+Launch two reviewers **in parallel** using the Agent tool (both in a single message for concurrent execution). Both must complete before proceeding to the verdict gate.
+
+Each reviewer evaluates every rubric criterion as PASS or FAIL, then returns structured JSON:
+
+```json
+{
+  "verdict": "PASS" | "FAIL",
+  "checks": [
+    {"criterion": "...", "result": "PASS|FAIL", "detail": "..."}
+  ],
+  "critical_issues": ["..."],
+  "suggestions": ["..."]
+}
+```
+
+The verdict gate (Step 4) maps these to NICE/NAUGHTY: both PASS → NICE, either FAIL → NAUGHTY.
+
+#### Reviewer A: Claude Agent (always runs)
+
+Launch an Agent (subagent_type: `code-reviewer`, model: `opus`) with the full rubric + all files under review. The prompt must include:
+- The complete rubric
+- All file contents under review
+- "You are an independent quality reviewer. You have NOT seen any other review. Your job is to find problems, not to approve."
+- Return the structured JSON verdict above
+
+#### Reviewer B: External Model (Claude fallback only if no external CLI installed)
+
+First, detect which CLIs are available:
+```bash
+command -v codex >/dev/null 2>&1 && echo "codex" || true
+command -v gemini >/dev/null 2>&1 && echo "gemini" || true
+```
+
+Build the reviewer prompt (identical rubric + instructions as Reviewer A) and write it to a unique temp file:
+```bash
+PROMPT_FILE=$(mktemp /tmp/santa-reviewer-b-XXXXXX.txt)
+cat > "$PROMPT_FILE" << 'EOF'
+... full rubric + file contents + reviewer instructions ...
+EOF
+```
+
+Use the first available CLI:
+
+**Codex CLI** (if installed)
+```bash
+codex exec --sandbox read-only -m gpt-5.4 -C "$(pwd)" - < "$PROMPT_FILE"
+rm -f "$PROMPT_FILE"
+```
+
+**Gemini CLI** (if installed and codex is not)
+```bash
+gemini -p "$(cat "$PROMPT_FILE")" -m gemini-2.5-pro
+rm -f "$PROMPT_FILE"
+```
+
+**Claude Agent fallback** (only if neither `codex` nor `gemini` is installed)
+Launch a second Claude Agent (subagent_type: `code-reviewer`, model: `opus`). Log a warning that both reviewers share the same model family — true model diversity was not achieved but context isolation is still enforced.
+
+In all cases, the reviewer must return the same structured JSON verdict as Reviewer A.
+
+### Step 4: Verdict Gate
+
+- **Both PASS** → **NICE** — proceed to Step 6 (push)
+- **Either FAIL** → **NAUGHTY** — merge all critical issues from both reviewers, deduplicate, proceed to Step 5
+
+### Step 5: Fix Cycle (NAUGHTY path)
+
+1. Display all critical issues from both reviewers
+2. Fix every flagged issue — change only what was flagged, no drive-by refactors
+3. Commit all fixes in a single commit:
+   ```
+   fix: address santa-loop review findings (round N)
+   ```
+4. Re-run Step 3 with **fresh reviewers** (no memory of previous rounds)
+5. Repeat until both return PASS
+
+**Maximum 3 iterations.** If still NAUGHTY after 3 rounds, stop and present remaining issues:
+
+```
+SANTA LOOP ESCALATION (exceeded 3 iterations)
+
+Remaining issues after 3 rounds:
+- [list all unresolved critical issues from both reviewers]
+
+Manual review required before proceeding.
+```
+
+Do NOT push.
+
+### Step 6: Push (NICE path)
+
+When both reviewers return PASS:
+
+```bash
+git push -u origin HEAD
+```
+
+### Step 7: Final Report
+
+Print the output report (see Output section below).
+
+## Output
+
+```
+SANTA VERDICT: [NICE / NAUGHTY (escalated)]
+
+Reviewer A (Claude Opus):   [PASS/FAIL]
+Reviewer B ([model used]):  [PASS/FAIL]
+
+Agreement:
+  Both flagged:      [issues caught by both]
+  Reviewer A only:   [issues only A caught]
+  Reviewer B only:   [issues only B caught]
+
+Iterations: [N]/3
+Result:     [PUSHED / ESCALATED TO USER]
+```
+
+## Notes
+
+- Reviewer A (Claude Opus) always runs — guarantees at least one strong reviewer regardless of tooling.
+- Model diversity is the goal for Reviewer B. GPT-5.4 or Gemini 2.5 Pro gives true independence — different training data, different biases, different blind spots. The Claude-only fallback still provides value via context isolation but loses model diversity.
+- Strongest available models are used: Opus for Reviewer A, GPT-5.4 or Gemini 2.5 Pro for Reviewer B.
+- External reviewers run with `--sandbox read-only` (Codex) to prevent repo mutation during review.
+- Fresh reviewers each round prevents anchoring bias from prior findings.
+- The rubric is the most important input. Tighten it if reviewers rubber-stamp or flag subjective style issues.
+- Commits happen on NAUGHTY rounds so fixes are preserved even if the loop is interrupted.
+- Push only happens after NICE — never mid-loop.

package/bundled/upstream/ecc/commands/save-session.md

@@ -0,0 +1,275 @@
+---
+description: Save current session state to a dated file in ~/.claude/session-data/ so work can be resumed in a future session with full context.
+---
+
+# Save Session Command
+
+Capture everything that happened in this session — what was built, what worked, what failed, what's left — and write it to a dated file so the next session can pick up exactly where this one left off.
+
+## When to Use
+
+- End of a work session before closing Claude Code
+- Before hitting context limits (run this first, then start a fresh session)
+- After solving a complex problem you want to remember
+- Any time you need to hand off context to a future session
+
+## Process
+
+### Step 1: Gather context
+
+Before writing the file, collect:
+
+- Read all files modified during this session (use git diff or recall from conversation)
+- Review what was discussed, attempted, and decided
+- Note any errors encountered and how they were resolved (or not)
+- Check current test/build status if relevant
+
+### Step 2: Create the sessions folder if it doesn't exist
+
+Create the canonical sessions folder in the user's Claude home directory:
+
+```bash
+mkdir -p ~/.claude/session-data
+```
+
+### Step 3: Write the session file
+
+Create `~/.claude/session-data/YYYY-MM-DD-<short-id>-session.tmp`, using today's actual date and a short-id that satisfies the rules enforced by `SESSION_FILENAME_REGEX` in `session-manager.js`:
+
+- Compatibility characters: letters `a-z` / `A-Z`, digits `0-9`, hyphens `-`, underscores `_`
+- Compatibility minimum length: 1 character
+- Recommended style for new files: lowercase letters, digits, and hyphens with 8+ characters to avoid collisions
+
+Valid examples: `abc123de`, `a1b2c3d4`, `frontend-worktree-1`, `ChezMoi_2`
+Avoid for new files: `A`, `test_id1`, `ABC123de`
+
+Full valid filename example: `2024-01-15-abc123de-session.tmp`
+
+The legacy filename `YYYY-MM-DD-session.tmp` is still valid, but new session files should prefer the short-id form to avoid same-day collisions.
+
+### Step 4: Populate the file with all sections below
+
+Write every section honestly. Do not skip sections — write "Nothing yet" or "N/A" if a section genuinely has no content. An incomplete file is worse than an honest empty section.
+
+### Step 5: Show the file to the user
+
+After writing, display the full contents and ask:
+
+```
+Session saved to [actual resolved path to the session file]
+
+Does this look accurate? Anything to correct or add before we close?
+```
+
+Wait for confirmation. Make edits if requested.
+
+---
+
+## Session File Format
+
+```markdown
+# Session: YYYY-MM-DD
+
+**Started:** [approximate time if known]
+**Last Updated:** [current time]
+**Project:** [project name or path]
+**Topic:** [one-line summary of what this session was about]
+
+---
+
+## What We Are Building
+
+[1-3 paragraphs describing the feature, bug fix, or task. Include enough
+context that someone with zero memory of this session can understand the goal.
+Include: what it does, why it's needed, how it fits into the larger system.]
+
+---
+
+## What WORKED (with evidence)
+
+[List only things that are confirmed working. For each item include WHY you
+know it works — test passed, ran in browser, Postman returned 200, etc.
+Without evidence, move it to "Not Tried Yet" instead.]
+
+- **[thing that works]** — confirmed by: [specific evidence]
+- **[thing that works]** — confirmed by: [specific evidence]
+
+If nothing is confirmed working yet: "Nothing confirmed working yet — all approaches still in progress or untested."
+
+---
+
+## What Did NOT Work (and why)
+
+[This is the most important section. List every approach tried that failed.
+For each failure write the EXACT reason so the next session doesn't retry it.
+Be specific: "threw X error because Y" is useful. "didn't work" is not.]
+
+- **[approach tried]** — failed because: [exact reason / error message]
+- **[approach tried]** — failed because: [exact reason / error message]
+
+If nothing failed: "No failed approaches yet."
+
+---
+
+## What Has NOT Been Tried Yet
+
+[Approaches that seem promising but haven't been attempted. Ideas from the
+conversation. Alternative solutions worth exploring. Be specific enough that
+the next session knows exactly what to try.]
+
+- [approach / idea]
+- [approach / idea]
+
+If nothing is queued: "No specific untried approaches identified."
+
+---
+
+## Current State of Files
+
+[Every file touched this session. Be precise about what state each file is in.]
+
+| File              | Status         | Notes                      |
+| ----------------- | -------------- | -------------------------- |
+| `path/to/file.ts` | PASS: Complete    | [what it does]             |
+| `path/to/file.ts` |  In Progress | [what's done, what's left] |
+| `path/to/file.ts` | FAIL: Broken      | [what's wrong]             |
+| `path/to/file.ts` |  Not Started | [planned but not touched]  |
+
+If no files were touched: "No files modified this session."
+
+---
+
+## Decisions Made
+
+[Architecture choices, tradeoffs accepted, approaches chosen and why.
+These prevent the next session from relitigating settled decisions.]
+
+- **[decision]** — reason: [why this was chosen over alternatives]
+
+If no significant decisions: "No major decisions made this session."
+
+---
+
+## Blockers & Open Questions
+
+[Anything unresolved that the next session needs to address or investigate.
+Questions that came up but weren't answered. External dependencies waiting on.]
+
+- [blocker / open question]
+
+If none: "No active blockers."
+
+---
+
+## Exact Next Step
+
+[If known: The single most important thing to do when resuming. Be precise
+enough that resuming requires zero thinking about where to start.]
+
+[If not known: "Next step not determined — review 'What Has NOT Been Tried Yet'
+and 'Blockers' sections to decide on direction before starting."]
+
+---
+
+## Environment & Setup Notes
+
+[Only fill this if relevant — commands needed to run the project, env vars
+required, services that need to be running, etc. Skip if standard setup.]
+
+[If none: omit this section entirely.]
+```
+
+---
+
+## Example Output
+
+```markdown
+# Session: 2024-01-15
+
+**Started:** ~2pm
+**Last Updated:** 5:30pm
+**Project:** my-app
+**Topic:** Building JWT authentication with httpOnly cookies
+
+---
+
+## What We Are Building
+
+User authentication system for the Next.js app. Users register with email/password,
+receive a JWT stored in an httpOnly cookie (not localStorage), and protected routes
+check for a valid token via middleware. The goal is session persistence across browser
+refreshes without exposing the token to JavaScript.
+
+---
+
+## What WORKED (with evidence)
+
+- **`/api/auth/register` endpoint** — confirmed by: Postman POST returns 200 with user
+  object, row visible in Supabase dashboard, bcrypt hash stored correctly
+- **JWT generation in `lib/auth.ts`** — confirmed by: unit test passes
+  (`npm test -- auth.test.ts`), decoded token at jwt.io shows correct payload
+- **Password hashing** — confirmed by: `bcrypt.compare()` returns true in test
+
+---
+
+## What Did NOT Work (and why)
+
+- **Next-Auth library** — failed because: conflicts with our custom Prisma adapter,
+  threw "Cannot use adapter with credentials provider in this configuration" on every
+  request. Not worth debugging — too opinionated for our setup.
+- **Storing JWT in localStorage** — failed because: SSR renders happen before
+  localStorage is available, caused React hydration mismatch error on every page load.
+  This approach is fundamentally incompatible with Next.js SSR.
+
+---
+
+## What Has NOT Been Tried Yet
+
+- Store JWT as httpOnly cookie in the login route response (most likely solution)
+- Use `cookies()` from `next/headers` to read token in server components
+- Write middleware.ts to protect routes by checking cookie existence
+
+---
+
+## Current State of Files
+
+| File                             | Status         | Notes                                           |
+| -------------------------------- | -------------- | ----------------------------------------------- |
+| `app/api/auth/register/route.ts` | PASS: Complete    | Works, tested                                   |
+| `app/api/auth/login/route.ts`    |  In Progress | Token generates but not setting cookie yet      |
+| `lib/auth.ts`                    | PASS: Complete    | JWT helpers, all tested                         |
+| `middleware.ts`                  |  Not Started | Route protection, needs cookie read logic first |
+| `app/login/page.tsx`             |  Not Started | UI not started                                  |
+
+---
+
+## Decisions Made
+
+- **httpOnly cookie over localStorage** — reason: prevents XSS token theft, works with SSR
+- **Custom auth over Next-Auth** — reason: Next-Auth conflicts with our Prisma setup, not worth the fight
+
+---
+
+## Blockers & Open Questions
+
+- Does `cookies().set()` work inside a Route Handler or only in Server Actions? Need to verify.
+
+---
+
+## Exact Next Step
+
+In `app/api/auth/login/route.ts`, after generating the JWT, set it as an httpOnly
+cookie using `cookies().set('token', jwt, { httpOnly: true, secure: true, sameSite: 'strict' })`.
+Then test with Postman — the response should include a `Set-Cookie` header.
+```
+
+---
+
+## Notes
+
+- Each session gets its own file — never append to a previous session's file
+- The "What Did NOT Work" section is the most critical — future sessions will blindly retry failed approaches without it
+- If the user asks to save mid-session (not just at the end), save what's known so far and mark in-progress items clearly
+- The file is meant to be read by Claude at the start of the next session via `/resume-session`
+- Use the canonical global session store: `~/.claude/session-data/`
+- Prefer the short-id filename form (`YYYY-MM-DD-<short-id>-session.tmp`) for any new session file