Add keboola-cli plugin with 10-agent project review#47
Open
ottomansky wants to merge 19 commits intokeboola:mainfrom
Open
Add keboola-cli plugin with 10-agent project review#47ottomansky wants to merge 19 commits intokeboola:mainfrom
ottomansky wants to merge 19 commits intokeboola:mainfrom
Conversation
…ency Delete general knowledge (ERP structures, SaaS metrics, SQL examples), compress prose checklists to rule tables, apply compact temp-dir output format. financial-analyst 294->120, template-readiness 331->142, performance-optimizer 301->119, security-auditor 282->107. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…mmand, bump to v1.1.0 Compress checklists to rule tables, delete SQL/code examples, apply compact temp-dir output format to all agents and kbc-review command. data-quality 210->120, semantic-layer 206->117, consolidator 200->153, dwh-architect 165->114, sql-reviewer 159->99, config-reviewer 159->107. Total: 2377->1268 lines (47% reduction). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…diagrams, action dependencies - Pre-flight: 5 checks before spawning agents (stale cleanup, .keboola exists, manifest readable, MCP reachable, configs non-empty) - Dedup: formal definitions for same/related issues, merge procedure, cross-reference enrichment, 30% sanity check - Mermaid: graph LR data flow diagram with color-coded issue nodes replaces text-only overview - Action items: dependency-ordered tables with Requires column and common patterns Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…date-only) and orchestration resilience Scope selection: --scope=sql,security runs only specified agents, --quick skips consolidator for inline summary, --consolidate-only retries failed consolidation from preserved temp reports. Auto-quick when 1-2 agents selected. Orchestration: 5-min per-agent timeout with monitoring loop, 8-min consolidator timeout, automatic report backup before overwrite, graceful degradation with partial results, --consolidate-only recovery path for failed runs. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
… config-analyzer pre-scan Shared context: orchestrator creates SHARED_CONTEXT.md in temp dir, all 9 reviewers read and append cross-domain findings (append-only to avoid contention), consolidator reads shared context to enrich merged findings. Pre-scan: config-analyzer runs on sonnet before review team (90s timeout), writes PROJECT_OVERVIEW.md with component inventory, data flow summary, bucket structure. All reviewers read it for shared project context, reducing redundant MCP calls. Updated: kbc-review.md, all 9 reviewer agents, kbc-review-consolidator.md, keboola-config-analyzer.md Agent total: 1356 lines (under 1400 target) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
v1.2.0 includes: pre-flight validation, targeted review mode (--scope, --quick, --consolidate-only), orchestration resilience (timeouts, monitoring, graceful degradation), structured deduplication rules, Mermaid data flow diagrams, action item dependency ordering, agent cross-reference via shared context file, config-analyzer pre-scan. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…to review agents (v1.3.0) - Add shared standards/review-standards.md (142 lines, 10 sections: architecture, bucket/table/column naming, PKs, technical columns, data types, transformation template, platform best practices) - Update kbc-review.md orchestrator to copy standards file to temp dir and reference in agent prompts - Update kbc-dwh-architect.md with Actum L0/L1/L2 bucket naming, table suffixes, column suffixes, technical columns checklist - Update kbc-sql-reviewer.md with 6 new rules (technical columns, UPPERCASE, domain suffixes, FK pattern, upsert template, moustache variables) - Update kbc-config-reviewer.md with 6 new rules (component/flow naming conventions, read-only mappings, flow phases, incremental PKs) - Update all 6 remaining agents with REVIEW_STANDARDS.md reference in Team Behavior - Bump plugin version 1.2.0 -> 1.3.0 Agent total: 1389 lines (budget: <1400). Standards file: 142 lines (shared, not counted in agent budget). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
….4.0) Default /kbc-review runs 7 general agents. --fi flag adds 2 FI agents (financial-analyst, template-readiness). FI domain knowledge moved to skills/keboola-fi/. Consolidator uses glob for dynamic report discovery. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…, polish - Fix data-quality agent step numbering (duplicate 3 -> 3,4,5,6) - Generalize template-readiness value inventory (domain-agnostic) - Auto-enable --fi when --scope includes financial/template - Add /kbc-fi shorthand command - Add scope presets (quick-check, architecture, compliance) - Trim consolidator to 178 lines (Mermaid + dedup compression) - Enrich keboola-fi skill (COA patterns + FI naming conventions) - Expand README (workflow, flags, presets, version history) - Bump to v1.5.0 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…t output dir Replace docs/.review_temp/ with docs/output/review/<project-name>/ so agent reports survive consolidation. Project name sanitized from MCP project info. Consolidator now deletes only working files (SHARED_CONTEXT.md, REVIEW_STANDARDS.md) and preserves agent reports + PROJECT_OVERVIEW.md. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…ic depth, agent trims New fi-template-spec agent (FI-only): template delta, Mermaid ER diagram, source knowledge base, template gaps. Semantic-reviewer gains FI metric depth section. Trims to consolidator, template-readiness, financial-analyst, data-quality-analyst for line budget. kbc-review updated for 10-agent --fi. SKILL.md adds template specification patterns. Plugin bumped to v1.7.0. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…tep numbering, fork bomb - Replace docs/.review_temp/ with <review_output_dir>/ in 10 agent files + config-analyzer - Fix FI agent count from 2 to 3 in SKILL.md, README.md, all references - Add kbc-fi-template-spec to agent lists in SKILL.md, README.md - Fix duplicate step 3 in Team Behavior (semantic, security, performance, financial) - Remove impossible "Message consolidator" step from all 11 reviewer agents - Remove fork bomb from plugin.json author name - Change config-analyzer model from sonnet to inherit - Improve description trigger phrases in SKILL.md, keboola-fi/SKILL.md, kbc-fi.md - Fix stale output path docs/PROJECT_REVIEW_REPORT.md -> docs/output/review/<project-name>/ - Add v1.6.0, v1.7.0, v1.7.1 to README version history - Bump plugin version to 1.7.1 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…E wording Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…andards ref Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
jordanrburger
approved these changes
Feb 9, 2026
…, naming conventions report Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- PK on every L1+ table (HIGH) - Composite PK uniqueness, no NULL components (CRITICAL) - FK explicit references to table.PK (MEDIUM) - FK cardinality documented (MEDIUM) - No orphan FKs in L2 mart tables (HIGH) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…K checks dwh-architect: PK/FK Integrity checklist (5 items) + PK/FK Validation output table data-quality: 3 new checks (cross-layer FK, PK uniqueness, composite PK NULLs) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- SKILL.md: Core Template Scope section (14 core tables, root variables, core vs addon metrics) - plugin.json: bump to 1.9.0 - README: add v1.9.0 to version history Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
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
keboola-cliplugin toplugins/keboola-cli/with a full project review system/kbc-review,/kbc-fi,/kbc-pull,/kbc-push,/kbc-init,/kbc-diffkeboola-config(platform knowledge) andkeboola-fi(financial intelligence domain)--fi), and consolidationdocs/output/review/<project-name>/What it does
/kbc-reviewspawns a team of reviewer agents against a Keboola project via MCP:Validated against Quantilope project (2415): 60 findings, 34 action items, 52% template readiness.
Version history
Test plan
/kbc-review --fiagainst Quantilope project 2415docs/output/review/quantilope/Generated with Claude Code