version: "1.0.3" name: Plan Templates description: Sidecar data for /plan workflow — Detail Plan and Roadmap templates loaded just-in-time. source: catalog

Skill: Plan Templates

Sidecar Skill for the /plan workflow. Contains plan document templates

that the Agent loads only when writing the plan file (Step 9).

Pattern: Workflow = Logic → Sidecar Skill = Data Router.

The /plan workflow instructs the Agent to read this skill at template-write time.

When to Load

• /plan create → Step 9 (Write Plan File): load the appropriate template from references/
• /plan review → NOT needed (no templates)
• /plan update → NOT needed (no templates)

References

Convention: Data files live in references/ (not templates/).

This follows the Sidecar Skill convention formalized in v1.7.6.3.

Template Selection Logic

Agent MUST select the correct template based on plan scope:

IF plan type = Roadmap
  → load references/roadmap.md
ELIF plan type = Template Plan
  → load references/detail-plan-template.md
ELIF plan type = Session Plan (DSP)
  → load references/session-plan.md
ELIF plan scope is documentation-only
     (target files are in docs/, no code changes, no repo/ modifications)
  → load references/detail-plan-docs.md
ELIF user specifically requests strict TDD (Test-Driven Development)
  → load references/detail-plan-tdd.md
ELIF user requests --hardened OR plan involves mixed code+docs with audit needs
  → load references/detail-plan-hardened.md
ELIF user requests csa-audit OR plan is for CSA Spec-to-Code audit
  → load references/detail-plan-csa-audit.md
ELSE
  → load references/detail-plan.md

Template Adaptation Rules

When generating a plan, Agent MUST load context in this order:

⛔ Log-First & Pre-Execution Skill Binding Protocol (hybrid-3-file-integrity.md C1 & context-rules.md)

1. Log Before Act (Plan Task Registration): When the User reports a bug, gap, or requests a UI/code fix during plan execution or review, Agent MUST update/append the Task List in the Active Plan file (`artifacts/plans/*.md`) FIRST (log-first principle) BEFORE making code changes.

2. Pre-Execution Context & Skill Binding: BEFORE executing any registered task or ad-hoc fix, Agent MUST scan project/workspace `skills.md` & `rules.md` trigger tables and read all relevant skills (e.g. tdd, csa, page-map, harness) using view_file.

3. Post-Execution Task & Phase Append Protocol: Small bugs/gaps in scope ➔ Append Phase N+1 (e.g. Phase N+1: Post-Execution Fixes & Polish) directly under active plan file FIRST (Log-First) with complete 5-step Post-Task Cascade Protocol (Code fix ➔ tsc typecheck ➔ npm test ➔ git commit/push ➔ Full-Stack Live Deployment Gate: npx wrangler deploy & npm run deploy).

4. Anti-Overreach User Proposal Gate (+ask): When bugs or new gaps occur, Agent MUST NOT modify any code file directly. Agent MUST draft the proposed diagnosis & task list in Chat, register the task in the plan, and obtain explicit user confirmation (+ask) BEFORE making any source code modifications.

5. No Silent Code Modifications: Modifying source files without registering an explicit task item in the active plan AND reading triggered skills is a severe workflow violation.

☁️ Techstack Domain Rules & Stack Routing

When drafting or executing plans for specific technology stacks (Cloudflare, Firebase, Android, etc.), Agent MUST load the corresponding stack domain file from domains/[stack].md (e.g. domains/cloudflare.md) listed in the References table.

Pre-requisite: Project Governance, Roadmap & Spec Baseline Loading

⛔ MANDATORY — Before writing ANY plan content:

1. Roadmap Sync: Agent MUST check project.md for the roadmap field.

- IF roadmap exists → Agent MUST read the roadmap file BEFORE drafting the plan to determine the correct Phase name, target Version number, and Dependency flow. Failure to read the roadmap leads to hallucinated versions and incorrect phase tracking.

2. Governance Sync: Agent MUST check project.md for agent.rules and agent.skills fields.

- IF agent.rules: true → read project .agents/rules.md index → load ALL triggered rules

- IF agent.skills: true → read project .agents/skills.md index → load ALL triggered skills

3. Spec Baseline Evaluation: Agent MUST check artifacts/specs/ or project specs/ for baseline specification files (spec-*.md).

- IF baseline specs exist → Agent MUST read the target spec file BEFORE drafting or updating plan tasks to evaluate functional scope, acceptance criteria, and CSA double-binding anchors.

This step ensures the plan respects project-specific governance:

- Maintenance rules (git scope, template-first flow, release process)

- Spec acceptance criteria & CSA double-binding integrity

- Review checklists (plan review §7, ecosystem integration)

- Naming conventions, documentation flow, quality standards

Failure to load project rules/skills/specs → plan WILL miss critical steps.

Adaptation Table

After loading project governance, read project.md and adapt the chosen template:

Principle: Template = clean skeleton. Adaptation = Skill responsibility.

Status lifecycle: 📝 Draft → 🔨 Active → ✅ Done. Transition from Draft → Active requires explicit user approval at /plan create Step 10 or /plan update.

⛔ Post-Draft Audit Gate & Interactive Pause Protocol (CP-3)

⛔ CP-3 CHECKPOINT (Hard Barrier):

1. Immediate Execution Stop: After writing the draft plan file using write_to_file (Step 14), Agent MUST STOP tool execution immediately.

2. No Auto-Audit in Same Turn: Agent MUST NOT auto-run the post-draft audit or modify the draft plan within the same tool turn.

3. Ask Permission First: Agent MUST present the raw draft plan link to User in chat and ask permission (Y/N) before initiating the Post-Draft Quality Audit (Step 15).

Difficulty Rating & Model Switching (v0.8.2+)

Each Phase header MUST include a difficulty tag to help users choose the appropriate AI model per phase:

Format: ### Phase N. [Name] ⚙️ \Difficulty: [🟢 Low | 🟡 Medium | 🔴 High]\``

Convention rules:

1. Final phase after `git push` SHOULD be 🟢 Low — reserved for low-thinking tasks (docs, readme, changelog) so user can switch to a cheaper model.
2. Phase with Difficulty: 🟢 Low MUST include a Model Hint blockquote:

`` > 💡 **Model Hint:** This phase is primarily documentation — consider switching to a lighter model to save costs. ``

1. Agent SHOULD assess difficulty based on the most complex task in the phase (not the average).

Output Checklist

After writing the plan, verify:

• [ ] Project contract analyzed
• [ ] Backlog items mapped to phases
• [ ] Project knowledge scanned (docs index, RFCs, architecture baseline, project rules)
• [ ] Architecture designed with component diagram (extended if baseline exists)
• [ ] Data schema defined (if applicable)
• [ ] Plan type selected: Roadmap vs Detail Plan (Step 2.8)
• [ ] Strategy/roadmap context loaded if applicable (Step 2.9)
• [ ] Phases defined (4-7 phases recommended)
• [ ] Code reuse documented (if reference projects exist)
• [ ] Plan saved to artifacts/plans/
• [ ] active_plan field set in project.md (detail plans only)
• [ ] Roadmap auto-updated if exists (Step 10)
• [ ] /backlog sync suggested (or auto-triggered)
• [ ] Session log updated