skills.rest
EricAnalysis
EricAnalysisCOMMUNITY
Explore all skills by this author →
GitHub

eightforge-validation-triage

Triage EightForge validation runs and findings persisted in Supabase.

Updated Mar 8, 2026
Search Tags:#debugging#validation#supabase#blockers#triage#validator#findings
One-click install
npx skills add https://github.com/EricAnalysis/eightforge-os --skill eightforge-validation-triage-ericanalysis
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: eightforge-validation-triage
Source: https://github.com/EricAnalysis/eightforge-os/tree/main/docs/skills/eightforge-validation-triage
Command: npx skills add https://github.com/EricAnalysis/eightforge-os --skill eightforge-validation-triage-ericanalysis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Debugging EightForge project validation behavior often means re-discovering the same validator run and finding paths, chasing stale blockers, duplicate findings, and findings_count mismatches across Supabase tables. This Skill anchors triage in persisted state first so investigations stay evidence-based and fixes stay minimal. ## Core Features & Use Cases - Persisted-State-First Triage: Inspect project_validation_runs, project_validation_findings, and projects.validation_summary_json before touching code. - Finding Classification: Distinguish real blockers from stale findings, semantic duplicates, field alias issues (vendor_name vs contractor_name), and rendering problems. - Minimal Fix Guardrails: Make the narrowest fix that preserves validator authority and canonical truth, with regression tests when rule changes are explicitly requested. - Use Case: A Golden Project rerun shows findings_count = 0 but attached findings still exist. Use this Skill to inspect persistence ordering around persistValidationRun, identify the post-persist side effect, and report the exact run id and root cause. ## Quick Start Use the eightforge-validation-triage skill to investigate why the latest Golden Project validation run shows stale blockers after a rerun.

Frequently Asked Questions about eightforge-validation-triage

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug stale validation blockers after a project rerun?▼

Compare the latest completed run id in project_validation_runs against open findings in project_validation_findings and the projects.validation_summary_json. Stale blockers are open findings from an older run that were never resolved by the newest completed run.

How do I triage duplicate contract-rate validation findings?▼

Group findings by invoice document, line code, description/rate context, and rule_id before changing any rule behavior. Semantic duplicates usually come from overlapping rules firing on the same line, so classify first and only then consider a minimal fix.

Why does findings_count show zero but findings still exist?▼

This mismatch usually comes from persistence ordering or post-persist side effects around persistValidationRun. Inspect the run row's counts and timestamps against the actual rows in project_validation_findings to find where the count was written before findings were stored.

Can this Skill change validator rule logic to unblock a decision?▼

No, it does not change rule logic merely to quiet noise or unblock a decision. If persisted evidence proves a rule is systematically wrong, it reports the proposed change, and only implements it with a regression test when the user explicitly requests rule editing.

Does the Decision Queue resolve validation findings?▼

No, the Decision Queue only frames and routes work. Canonical facts, validator findings, and execution state remain authoritative in their own layers, and only Execution finalizes approval-impacting outcomes.

Related Skills

Explore More Software Engineering Skills→
apply-theapply-the
1

canon-inspect-artifacts

Return Canon-backed artifact paths under .canon/artifacts/<RUN_ID>/ for a run.

Official
Intermediate
SalesTeamToolboxSalesTeamToolbox

standup-report

Generates a structured daily standup report from git history, task-queue status, and agent activity.

Community
Basic
ai-dynamoai-dynamo
526

aiperf-code-review

Compare current branch to origin/main and document code-quality gaps with runtime evidence.

Official
Advanced
dlt-hubdlt-hub
53

view-data

Query and explore dlt pipeline data using Python and ibis expressions.

Official
Intermediate