43 lines
3.2 KiB
Plaintext
43 lines
3.2 KiB
Plaintext
GOAL - Restore odoo-expertos.com to a re-audited live SEO health >=75/100. ONE session supervises the launcher; the launcher runs the heavy parallel work (1 SETUP + 10 parallel BATCH worktrees + 1 FINAL). Repo root = current folder. Work ONLY here.
|
|
|
|
KEY FILES (read for context):
|
|
- seo-audit-2026-05-27\run-parallel-100.ps1 (orchestrator)
|
|
- seo-audit-2026-05-27\goals\SETUP.txt, B01..B10.txt, FINAL.txt
|
|
- seo-audit-2026-05-27\BLOG-PIPELINE.md (12-step per-article pipeline; kie.ai z-image policy)
|
|
- seo-audit-2026-05-27\OPPORTUNITIES-2026-05-28.md
|
|
- seo-audit-2026-05-27\batches\URL-BATCHES.json (100 URLs in B01..B10)
|
|
- seo-audit-2026-05-27\REVIEW.ps1 (deterministic local gate)
|
|
|
|
TURN 1 - FIRE LAUNCHER (background): use Bash run_in_background=true:
|
|
powershell -File seo-audit-2026-05-27\run-parallel-100.ps1 -Phase all -MaxConcurrent 10 -Model sonnet > seo-audit-2026-05-27\launcher.log 2>&1
|
|
Note the bg task id. Do NOT block. Runtime: 8-16 hours.
|
|
|
|
TURN 2..N - MONITOR each turn until launcher exits:
|
|
- Tail seo-audit-2026-05-27\launcher.log (last 30 lines).
|
|
- git log fix/seo-all --oneline -n 8.
|
|
- For each batch B01..B10 still active: tail ..\odoo-wt-bNN\BNN.log.
|
|
- Report phase (SETUP / BATCHES x/10 / FINAL).
|
|
- If launcher.log contains "z-image unavailable" -> ABORT and paste the kie.ai error verbatim. NO model fallback.
|
|
- A single failing batch is OK; FINAL handles partial if >=7 batches succeed.
|
|
|
|
TURN VERIFY (when launcher exits):
|
|
- powershell -File seo-audit-2026-05-27\REVIEW.ps1 -Only all - paste output; W1..W11, W5b, W12, W13 all PASS.
|
|
- git log main --oneline -n 5 - shows FINAL merged to main.
|
|
- curl -sI https://www.odoo-expertos.com/ - X-Content-Type-Options + X-Frame-Options present (W11 live).
|
|
- curl -s https://www.odoo-expertos.com/robots.txt | findstr "/\*#" - EMPTY (W2 live).
|
|
|
|
TURN AUDIT:
|
|
Invoke Skill tool: skill="seo-audit", args="https://www.odoo-expertos.com/". Paste the full report. Extract the "Overall SEO Health Score: X/100" line.
|
|
|
|
DONE WHEN: extracted Overall SEO Health Score >=75/100 AND REVIEW.ps1 -Only all all PASS AND main is deployed.
|
|
|
|
IF SCORE <75: identify lowest-scoring category (Technical/Content/On-Page/Schema/Performance/AI Search/Images), then EITHER (a) re-run powershell -File seo-audit-2026-05-27\run-parallel-100.ps1 -Phase batches -MaxConcurrent 5 (resumes failed batches in existing worktrees), OR (b) fix that category inline with the matching /blog skill. Then merge to main, wait for Vercel (poll robots.txt as in VERIFY), re-audit. Hard cap 200 total turns. If still <75 report 3 lowest categories with their recommended fixes.
|
|
|
|
CONSTRAINTS (already enforced by per-batch goals; restated for supervisor):
|
|
- blog-factcheck mandatory; NEVER fabricate stats (drop unsourced numbers).
|
|
- Preserve design + ALL D3.js refs; preserve <head> except the content-version marker + og:image + preload.
|
|
- "Equipo Odoo Expertos" replaces every "Maria Elena Rodriguez" in W12/W13 marked pages.
|
|
- >=3 z-image images per article (hero+fig1+fig2); every image prompt ends with ", NO text, NO letters, NO numbers, NO logos, NO watermarks.".
|
|
- All image gen via seo-audit-2026-05-27\scripts\kie-image.ps1 -Model z-image only (no fallback). NEVER /blog image generate.
|
|
- No edits outside this repo.
|