Ext Outline Plugin

Plugin Outline Extension

Domain-specific outline workflow for marketplace plugin development. Handles discovery, analysis, uncertainty resolution, and deliverable creation for Complex Track requests.

Loaded by: pm-workflow:phase-3-outline (Complex Track)

Input

plan_id: {plan_id}

All other data read from sinks (references.toon, request.md).

Workflow Overview

Step 1: Load Context      → Read request, module_mapping, domains, compatibility
Step 2: Discovery         → Spawn ext-outline-inventory-agent
Step 3: Determine Type    → Extract change_type from request
Step 4: Execute Workflow  → Route based on change_type (Create/Modify Flow)
Step 5: Write Solution    → Persist solution_outline.md

Detailed workflow: Load standards/workflow.md for Create Flow and Modify Flow logic.

Step 1: Load Context

Read request (uses clarified_request if available):

python3 .plan/execute-script.py pm-workflow:manage-plan-documents:manage-plan-documents request read \
  --plan-id {plan_id} \
  --section clarified_request

Read module_mapping from work directory (for bundle filtering hints):

python3 .plan/execute-script.py pm-workflow:manage-files:manage-files read \
  --plan-id {plan_id} \
  --file work/module_mapping.toon

Read domains:

python3 .plan/execute-script.py pm-workflow:manage-references:manage-references get \
  --plan-id {plan_id} --field domains

Read compatibility from marshal.json project configuration:

python3 .plan/execute-script.py plan-marshall:manage-plan-marshall-config:plan-marshall-config \
  plan phase-2-refine get --field compatibility --trace-plan-id {plan_id}

Derive the long description from the compatibility value:

• breaking → "Clean-slate approach, no deprecation nor transitionary comments"
• deprecation → "Add deprecation markers to old code, provide migration path"
• smart_and_ask → "Assess impact and ask user when backward compatibility is uncertain"

Store as compatibility and compatibility_description for inclusion in the solution outline header metadata.

Log context:

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Context loaded: domains={domains}, compatibility={compatibility}"

Step 1.5: Determine Component Scope

Analyze the request to determine which component types are affected.

Component Types (all must be considered)

Decision Logic

FOR EACH component_type IN [skills, agents, commands, scripts, plugin.json, tests, project-skills]: ANALYZE request for explicit OR implicit mentions LOG decision with evidence

Log Decisions

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Component scope: [{component_types_list}]"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Skills: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Agents: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Commands: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Scripts: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) plugin.json: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Tests: {AFFECTED|NOT_AFFECTED} - {evidence}"

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Project-skills: {AFFECTED|NOT_AFFECTED} - {evidence}"

Skill Sub-Documents

IF skills are in scope: ALSO include skill sub-documents:

• standards/*.md
• templates/*
• references/*
• knowledge/*

This uses --full mode in inventory scan to enumerate sub-documents.

Step 1.6: Derive Content Filter Criteria (if applicable)

For certain change_types, derive a content filter pattern to reduce LLM analysis load.

When to Apply

Pattern Derivation Examples

Log Decision

IF content filter is derived:

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO '(pm-plugin-development:ext-outline-plugin) Content filter: pattern={pattern}, derived from {request_keywords}'

IF no content filter applicable:

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Content filter: NONE (change_type={change_type} does not benefit from content filtering)"

Step 2: Discovery

Spawn ext-outline-inventory-agent with component scope and content filter from Steps 1.5/1.6:

Task: pm-plugin-development:ext-outline-inventory-agent
  Input:
    plan_id: {plan_id}
    component_types: [{component_types from Step 1.5}]
    content_pattern: "{pattern from Step 1.6 or empty}"
    bundle_scope: {from module_mapping or "all"}
  Output:
    inventory_file: work/inventory_filtered.toon
    scope: affected_artifacts, bundle_scope
    counts: skills, commands, agents, total

The agent:

• Runs scan-marketplace-inventory with --resource-types from component_types
• Uses --content-pattern if provided (enables pre-filtering)
• Uses --bundles filter if module_mapping specifies specific bundles
• Persists inventory to work/inventory_filtered.toon
• Stores reference as inventory_filtered in references.toon

Contract: After agent returns, work/inventory_filtered.toon exists.

Filter Result Logging

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  work {plan_id} INFO '[PROGRESS] (pm-plugin-development:ext-outline-plugin) Inventory: {total} files, content filter: {input} → {matched} ({excluded} excluded)'

Error Handling

CRITICAL: If agent fails due to API errors, HALT immediately.

IF agent returns API error (529, timeout, connection error):
  HALT with error:
    status: error
    error_type: api_unavailable
    message: "Discovery agent failed. Retry later."

  DO NOT:
    - Fall back to grep/search
    - Continue with partial data

Step 3: Determine Change Type

Extract change_type from request:

Log decision:

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Change type: {change_type}"

Validation

IF affected_artifacts is empty:
  ERROR: "No artifacts affected - clarify request"

Step 4: Execute Workflow

Load detailed workflow:

Read standards/workflow.md

The workflow routes based on change_type:

Create Flow Summary

• No analysis needed (files don't exist yet)
• Build deliverables directly from request
• One deliverable per component to create

Modify Flow Summary

• Load persisted inventory
• Spawn analysis agents per bundle/type
• Persist assessments to artifacts/assessments.jsonl
• Resolve uncertainties via AskUserQuestion
• Group into deliverables
• Write solution_outline.md

Test Deliverables (CRITICAL)

MANDATORY: Test deliverables MUST list explicit file paths, not descriptive text.

When tests are in component scope:

1. Discover test files from inventory tests section
2. For each affected script, find corresponding test files:

   # Pattern: test/{bundle}/{skill}/test_*.py
   # Example: test/pm-workflow/manage-tasks/test_manage_tasks.py
   

3. List explicit paths in the deliverable's Affected files: section

INVALID (descriptive text):

**Affected files:**
- Tests that assert on JSON output format
- Tests that use json.loads() to parse output

VALID (explicit paths):

**Affected files:**
- `test/pm-workflow/manage-tasks/test_manage_tasks.py`
- `test/pm-plugin-development/plugin-doctor/test_doctor_marketplace.py`

Note: Task contract validation rejects non-file-path steps. Descriptive text will cause task creation to fail.

Step 5: Write Solution Outline

After deliverables are built, write solution_outline.md. The header MUST include compatibility: {compatibility} -- {compatibility_description} (read from references.toon in Step 1).

CRITICAL - Plugin-Doctor Verification: Every deliverable's Verification section MUST include /plugin-doctor for each affected component path:

• For skills: /pm-plugin-development:plugin-doctor --component {skill_path}
• For agents: /pm-plugin-development:plugin-doctor --component {agent_path}
• For commands: /pm-plugin-development:plugin-doctor --component {command_path}

Additional domain-specific checks (like grep for format validation) may be included but do NOT replace plugin-doctor.

python3 .plan/execute-script.py pm-workflow:manage-solution-outline:manage-solution-outline write \
  --plan-id {plan_id} <<'EOF'
# Solution: {title}

plan_id: {plan_id}
compatibility: {compatibility} — {compatibility_description}

## Summary

{2-3 sentence summary}

## Deliverables

{deliverables content}
EOF

Log completion:

python3 .plan/execute-script.py plan-marshall:manage-logging:manage-log \
  decision {plan_id} INFO "(pm-plugin-development:ext-outline-plugin) Complete: {N} deliverables"

Output

Return minimal status - all data in sinks:

status: success
plan_id: {plan_id}
deliverable_count: {N}

Sinks Written

Impact Analysis (Optional)

Condition: Run if inventory has < 20 files AND change_type is "modify", "migrate", or "refactor".

Purpose: Discover components that depend on affected components.

python3 .plan/execute-script.py pm-plugin-development:ext-outline-plugin:filter-inventory \
  impact-analysis --plan-id {plan_id}

For detailed rules, see standards/impact-analysis.md.

Uncertainty Resolution

Trigger: Run if uncertain > 0 after analysis.

Purpose: Convert UNCERTAIN findings to CERTAIN through user clarification.

Grouping Patterns

Resolution Application

1. Query UNCERTAIN assessments:

python3 .plan/execute-script.py pm-workflow:manage-plan-artifacts:manage-artifacts \
  assessment query {plan_id} --certainty UNCERTAIN

2. Use AskUserQuestion with specific file examples
3. Log resolutions as new assessments:

python3 .plan/execute-script.py pm-workflow:manage-plan-artifacts:manage-artifacts \
  assessment add {plan_id} {file_path} {new_certainty} 85 \
  --agent pm-plugin-development:ext-outline-plugin \
  --detail "User clarified: {user_choice}" --evidence "From: {original_hash_id}"

4. Store clarifications in request.md:

python3 .plan/execute-script.py pm-workflow:manage-plan-documents:manage-plan-documents \
  request clarify \
  --plan-id {plan_id} \
  --clarifications "{formatted Q&A}" \
  --clarified-request "{synthesized request}"

Error Handling

CRITICAL: If any operation fails, HALT immediately.

DO NOT: Fall back to grep/search, skip failed bundles, continue with partial data.

Conditional Standards

Related

• workflow.md - Create Flow and Modify Flow details
• outline-extension.md - Contract this skill implements
• component-analysis-contract.md - Analysis agent contract