From tehik-task-board
Second-pass review — moves To-Validate tasks to Done or back to To-Do with clarifying notes (suited to running with a stronger model)
How this skill is triggered — by the user, by Claude, or both
Slash command
/tehik-task-board:validate-task-boardThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Goal:** Review tasks that have **already been completed** in the **first pass** (usually with a faster/lighter model) and sit in the **`To-Validate`** folder.
Goal: Review tasks that have already been completed in the first pass (usually with a faster/lighter model) and sit in the To-Validate folder.
Running with a stronger or more capable model improves the chance of catching logic errors, gaps, or mismatches with acceptance criteria.
Varies by project: By default this assumes the structure
Agent_reports/Tasks/{To-Do,To-Validate,Done}. If the user provides different paths, use those.
/process-task-board, when finished work was placed under To-Validate (not straight into Done).Done.Model choice: In the IDE, run this workflow in a chat where a stronger / more capable model is selected (if you have a choice), to reduce superficial approval.
| Role | Default path (customizable) |
|---|---|
| Tasks to review | <project>/Agent_reports/Tasks/To-Validate/ |
| If approved | .../Done/ |
| If fixes needed | .../To-Do/ |
If the user does not provide paths, assume these three folders from the project root.
The user may give absolute paths, e.g. ~/project/Agent_reports/Tasks/To-Validate.
You may update frontmatter (if you use YAML):
validation_status: approved # approved | needs_revision
validated_at: "2026-03-23T14:00:00+02:00" # ISO 8601
reviewer: agent-second-pass # human may replace with their initials
If there is no frontmatter: add the review summary at the beginning or end of the file with a consistent heading (see below).
For each file or task folder (To-Validate/<id>-<slug>/):
.md body, adjacent artifacts, README.md in the folder.type: code tasks: Does the work show a full cycle was followed (at least references to branch, tests, commit), not only “code changed”? If the task expected /implement-feature or /fix-bug but tests/commit are missing without a user exception → prefer needs_revision and list missing steps.If the task is a partner proposal type, you may additionally reference the /validate-external-proposal workflow logic (you do not need to rerun the whole flow if the content is already complete).
Donevalidation_status: approvedstatus: doneTo-Validate → Done (same relative structure: if it was To-Validate/PROJ-1/report.md + subfiles, keep them together).To-Validate.To-DoREVIEW.md in the same folder if the file is very long — prefer one place for clarity):## Review feedback (validation pass)
**Status:** needs_revision
**Date:** YYYY-MM-DD
### Must fix
- [ ] ...
### Suggestions (optional)
- ...
### Notes
- ...
validation_status: needs_revision, status: open.To-Validate → To-Do.If validation shows that completing the current task requires a separate new task:
id, clear scope, acceptance criteria).follow_up_tasks or a clear ID list).done by mistake if the dependency is open:
needs_revision or split scope and document which part was considered done.At the end, provide a summary:
| Task | Decision | Target folder |
|---|---|---|
| … | approved / needs_revision | Done / To-Do |
/process-task-boardGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub tehik-ee/ai-generic-skills --plugin tehik-task-board