Skip to content
Back to skills

Correct Course

ASecurity

Manage mid-sprint changes and pull deferred items forward into the active workstream. Use when the user says "correct course", "change of plans", "we need to adjust", or "something came up".

  • 2 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 19, 2026
ai-agentsgogit

Security analysis

A100/100

Scanned September 19, 2026

npx -y skills add rterakedis/leanwheel-skills --skill correct-course --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Correct Course?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Correct Course
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/rterakedis-correct-course/badge)](https://www.skillsdirectory.com/skills/rterakedis-correct-course)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: correct-course
description: Manage mid-sprint changes and pull deferred items forward into the active workstream. Use when the user says "correct course", "change of plans", "we need to adjust", or "something came up".
---

# Correct Course Skill

**Goal:** When discovery/requirements/bugs break plan, update `docs/`, `docs/epics/`, and backlog coherently. Every impact: fix now, schedule as story, or defer with date/condition.

**Deferred problem:** `[Defer]` items in stories go nowhere. This skill pulls them forward and forces decisions.

## Activation

Ask: "What triggered this?"
1. Requirement changed
2. Bug in done work
3. Architecture wrong mid-implementation
4. External dependency changed
5. Scope creep
6. Other

Store trigger. Read `docs/epics.md` for status.

---

## Phase 1 — Triage the Trigger

If `docs/project/decisions.md` exists, record the trigger there first (**PARK** from
`skills/decision-log/SKILL.md`), then run the `/ideate` loop scoped to it — resolve only the
questions this trigger raises, recording each impact decision with `source: /correct-course`.
The recorded decisions are the triage output. (No decision log yet → assess inline as below;
consider creating the log if the change is direction-setting.)

Assess impact across: PRD (what/who), architecture (tech/patterns), epics (ACs/tasks/ordering), active story, done stories.

Note what changes and why for each affected artifact.

---

## Phase 2 — Classify Each Impact

`fix-in-place`: not implemented yet (ready-for-dev). Update docs/story directly.

`new-story`: done/review work OR large fix. Don't reopen done stories; schedule new story instead (keeps history clean).

`defer-explicitly`: real but out of scope. Must include: issue, why deferred, specific condition/date for revisit. No vague deferrals.

---

## Phase 3 — Pull Forward Deferred Items

Execute **SCAN** then **RESOLVE** from `skills/deferred/SKILL.md`. All open deferrals must be resolved (no orphaned items).

---

## Phase 4 — Execute Changes

**Fix-in-place:** Edit doc/story; add inline `<!-- corrected {date}: {reason} -->` if non-obvious.

**New stories:** Execute **SCHEDULE** from `skills/deferred/SKILL.md` (writes to epics.md + GitHub).

**Explicit deferrals:** Add to `## Deferred Items` in `docs/epics.md`: `- {date} — {issue} — Deferred until: {condition}. Trigger: {event}`.

---

## Phase 5 — Update Planning Docs

PRD changed: bump last-updated, one-line comment at section top.
Architecture changed: bump last-updated. If pattern changed, `touch docs/architecture.md` to invalidate caches.

---

## Phase 6 — Summary

Report: artifacts updated, new stories, deferred items, resolved deferrals, next action.

---

## When to Run This Skill vs. Others

Bug in **done** story: `/correct-course` (add remediation story forward).
Bug in **active** story: inline auto-patch.
Bug needs investigation: `/investigate` → `/correct-course`.
Small post-MVP: `/quick-dev`.
Requirements changed: `/correct-course` (may trigger `/spec` prd update).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.