fix: COO brief accuracy -- skeleton templates, NPS source, ALICE windows, timestamps#2371
Closed
gabrielratner-debug wants to merge 153 commits into
Closed
fix: COO brief accuracy -- skeleton templates, NPS source, ALICE windows, timestamps#2371gabrielratner-debug wants to merge 153 commits into
gabrielratner-debug wants to merge 153 commits into
Conversation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…penAI Whisper Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Syncs with upstream main (on schedule, dispatch, or push), then merges main into all skill/* branches with build+test validation. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…Forecast Briefing: removed the current-year Actuals fetch (dataset -3) from the P&L instructions. Columns are now Primary Forecast | Budget | vs Bud $ | vs Bud % | LY | vs LY % for both months. Reduced calls from 48 to 36. Prefetch: removed _act.csv fetch from both main parallel pass and retry pass. Saves one API call per hotel (12 fewer calls at 3am). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Expanded the em dash prohibition to cover all dashes used as separators in prose. Rule 4 now applies in every container session across all channels. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
coo-prefetch was OOM-killed (SIGKILL) mid-Snowflake queries after completing all 36 ProfitSword files. 2GB cap insufficient for 12 parallel Python ProfitSword procs + Snowflake MCP loading large YTD result sets. 4GB gives 2x headroom over actual peak. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Open items tracking simplified to Telegram only: removed Microsoft ToDo rules from global CLAUDE.md, removed SQLite open_items section from nanoclawrules.md. Outlook email bodies now truncated to 1500 chars to prevent container OOM from large HTML emails. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
revman@properhotel.com was CC'd on the April 22 briefing. The existing "NO CC" instruction was not explicit enough about the tool call parameters. Now bans cc/bcc parameters explicitly and names the specific address as an example of what not to add. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…O briefing - P&L: add Room Nights, Occupancy%, ADR, RevPAR KPI rows above each monthly table (RF0001/RMAVL/RF0003 ItemTags, computed values) - Revinate: add per-platform ratings table (Google, Booking.com, Expedia, TripAdvisor) with MTD/YTD counts; TripAdvisor rank via WebSearch for each property - Duetto Pace: Total/Transient/Group all three segments now required - Colors: black default, red negative variance only, green positive only Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Daily mode now shows: - P&L: KPI header (Occ/ADR/RevPAR) + full table every day, current month only (next month on Mondays only) - Duetto: all three segments (Total/Transient/Group) every day - Revinate: full platform ratings table every day (Google/Booking/ Expedia/TripAdvisor with rank), plus new-since-yesterday callout Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
High glitch volume is good (staff encouraged to log everything). Now watches for TYPE trends -- a specific issue category spiking 2x vs prior month baseline (plumbing, noise, AC, etc.). Removed volume > 600 MTD threshold. Action items weight ALICE less. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Table: DUETTO_UPLOAD.RAW.LIGHTHOUSE_RATES (all 12 properties, daily refresh at 5:20am, 365 days forward, LOS 1/2/3). Prefetch: new query 3h fetches next 30 days LOS=1 own-hotel rows to lighthouse.json. Briefing: new section 4 (between Duetto Pace and ALICE) shows 14-day rate positioning table: own rate, median comp, rank, my OTB%, market OTB%, demand%, events. Flags: near-last rank on high-demand dates, OTB gap vs market, sold-out in next 7 days, event/demand mismatches. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Gmail invalid_grant was propagating out of main() and crash-looping nanoclaw. Wrap each channel.connect() in try/catch so one channel failure degrades gracefully instead of taking down the whole process. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Captures all uncommitted local work (Slack monitor, EOD Slack, token audit skills, model override, slack-monitor mount, config .env fallbacks) as a clean checkpoint before merging upstream v2. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…piled takeaways - Remove Resolved and Resolved% from ALICE section (Table 7 and per-property 3f) - Add Top 3 Issues (not just Top 1) to ALICE table and computation - Change email delivery from inline HTML to HTML file attachment - Add Section 4: All Properties Takeaways compiled at bottom of report Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Email body now contains Portfolio Takeaways + All Properties Takeaways so the key signals are readable without opening the attachment. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
For each property takeaway that signals a problem, generate a specific action naming the right person to engage (Rowan for sales pace, Tracie for RevMan/STR watchlist, Leah for ops/NPS, Luis for F&B, plus the property GM/MD). Actions appear in section 3a, section 4, and email body. Full routing table with all 12 GMs/MDs and 4 corporate owners included. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
STR CurrWk now anchors to Sunday of the week containing max_date via (max_date.weekday() + 1) % 7 offset -- was a floating 7-day window. Revinate: renamed NPS (mislabeled) to Recommend Score (0-10), added true NPS row (Promoters >= 9 minus Detractors <= 6, as signed integer), renamed Rating (0-5) to Rating (1-5). Table 6 updated to match. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Prefetch: STR Monthly query now selects "RANK" (double-quoted, SQL reserved word) alongside indices. Briefing: YTD rank cells now read from str_monthly.json instead of showing "--". CurrWk/R28 rank still taken from latest-day in str_daily.json. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Prefetch: adds survey_topics (RAW_JSON:surveyTopics) to Revinate query. Briefing: computes Survey Overall Exp (1-5) from surveyTopics.Overall Experience ratingAnswer (survey source only), displayed as a separate row from Platform Rating (1-5) which is the public star average. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
send-email MCP tool has no attachment support -- send complete HTML report as email body instead of trying to attach a file. Cache-miss guard: if manifest.json is absent, abort with send_message instead of attempting live-fetch (which times out at 30 min). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Prefetch: added execution rules to prevent agent from exiting early on stale background notifications. Must write manifest AND send Telegram before concluding. Briefing: added Graph API Python fallback (scripts/email/send_email.py) if mcp__outlook__send-email fails. Falls back automatically. container-runner: also mounts Outlook token at /root/.outlook-mcp-tokens.json so the Python fallback script finds it regardless of running user. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
MTD Forecast Accuracy now uses RF0003 Amt (Rooms Revenue) for both actuals and forecast/budget comparisons. TOTOPRV was including F&B and spa, inflating the gap vs Duetto rooms-only OTB. Added note clarifying Leah Edwards dual role: GM Hotel June Malibu (property actions) vs Corporate Director of Operations (portfolio actions). Never "Managing Director". Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ows, timestamps - Replace property detail and executive summary sections with locked fill-in skeletons to prevent model from improvising extra sections (P&L tables, full-month OTB tables, KPI header cards) - Fix NPS filter: 'Revinate Surveys' not 'Survey' (was matching nothing, producing garbage counts and wrong NPS) - Replace MTD-vs-full-month forecast comparison with Actual % of Fcst + Month Elapsed benchmark column so mid-month data is interpretable - Add 3:00am cache timestamps to P.Fcst and Duetto sections so readers know figures may lag if forecast was updated after prefetch ran - Expand ALICE from last-7d only to Last 7/30/60/90d windows; remove MTD/YTD counts which were wrong due to query scope mismatch - Add Overall bundled rating row to platform reviews table - Require exact computed values in narrative bullets (no rounding) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Author
|
Superseded by cleaner branch off main -- no merge conflicts. |
This was referenced May 10, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
'Survey', data uses'Revinate Surveys'-- was matching nothing and producing garbage NPS countsTest plan
🤖 Generated with Claude Code