Track Based Task Execution Guide

Track-Based Task Execution Guide

Version: 1.0.0 Created: January 1, 2026 Standard: CODITECT-STD-001 (Agentic Communication Standard)

Overview

This guide documents the Track-based Task Management System used for all CODITECT development work. The system replaces legacy "Phase" nomenclature with a clearer Track A-G structure.

Key Principle: One source of truth for active work:

PILOT-PARALLEL-EXECUTION-PLAN.md - Authoritative task list

Quick Reference

Step	Action	Command/Location
1. Find Work	Read the PILOT plan	`cat internal/project/plans/PILOT-PARALLEL-EXECUTION-PLAN.md`
2. Pick Task	Choose unchecked `[ ]` item	Format: `X.Y.Z` (e.g., E.2.1)
3. Get Agent	Find right agent	`/which "task description"`
4. Execute	Run the work	`/agent <name> "task"` or direct implementation
5. Mark Done	Check the box	Change `[ ]` to `[x]` in PILOT plan
6. Log	Record in session log	`/session-log`

Track Nomenclature

Track Definitions

Track	Domain	Primary Agent	Repository/Scope
A	Backend API	`backend-development`	`coditect-cloud-infra/backend`
B	Frontend UI	`frontend-react-typescript-expert`	`coditect-cloud-frontend`
C	DevOps	`devops-engineer`	Infrastructure, K8s, CI/CD
D	Security	`security-specialist`	Cross-cutting security
E	Testing	`testing-specialist`	Test suites, QA
F	Docs	`codi-documentation-writer`	Documentation sites
G	DMS	`prompt-analyzer-specialist`	Document Management System

Task ID Format

Track.Section.Task[.Subtask]

Examples:
- A.3.1    = Backend, Section 3, Task 1
- E.2.1    = Testing, Section 2, Task 1
- A.3.1.1  = Backend, Section 3, Task 1, Subtask 1

Quick Track Guide

Track A = Backend    Track D = Security
Track B = Frontend   Track E = Testing
Track C = DevOps     Track F = Docs
                     Track G = DMS

Task Execution Workflow

Step 1: Start of Session

Run orientation to find the active plan and understand current state:

# Run full orientation (recommended)
/orient

# OR directly read the authoritative source:
cat internal/project/plans/PILOT-PARALLEL-EXECUTION-PLAN.md | head -200

# Check recent context
/cxq --recent 200
/cxq --decisions --limit 20

Step 2: Find Your Next Task

Tasks use format: Track.Section.Task (e.g., E.2.1, A.3.1.1)

# Find unchecked tasks in a specific track
grep -A5 "E.2" internal/project/plans/PILOT-PARALLEL-EXECUTION-PLAN.md

# Example output:
# - [ ] E.2.1: Rate limiting verification (429 responses)
# - [ ] E.2.2: Failed login lockout (5 attempts)
# - [x] E.2.3: Security headers verification (COMPLETE)

Priority Selection:

Tasks marked with **BLOCKING** or P0
Tasks with dependencies on other tracks
Sequential tasks within a track (E.2.1 before E.2.2)

Step 3: Get Agent Recommendation

Use /which to find the best agent for your task:

# Get agent recommendation
/which "rate limit security testing"

# Example output:
# Recommended: testing-specialist
# Alternative: security-specialist
# Built-in: Task(subagent_type="testing-specialist", ...)

Step 4: Execute Task

Option A: CODITECT Agent (130+ custom agents)

/agent testing-specialist "Write rate limit tests for E.2.1 - verify 429 responses"
/agent codi-documentation-writer "Update API docs for new endpoints"

Option B: Claude Code Built-in Agent (~52 types)

Task(subagent_type="testing-specialist", prompt="Write rate limit tests...")
Task(subagent_type="security-specialist", prompt="Security audit for...")

Option C: Direct Implementation

# For simple tasks, implement directly without agent delegation
# Then verify with tests
pytest backend/tests/security/test_rate_limit.py -v

Step 5: Update Status

Mark the task completed in PILOT-PARALLEL-EXECUTION-PLAN.md:

# Before
- [ ] **E.2.1**: Rate limiting verification (429 responses)

# After
- [x] **E.2.1**: Rate limiting verification (429 responses) - COMPLETE

Step 6: Log Session

Record your work for continuity:

# Add entry to session log
/session-log

# Capture context for future sessions
/cx

# For significant milestones, create checkpoint
/checkpoint "Track E.2 security tests complete"

Practical Example

Full Workflow Demonstration

# 1. FIND WORK - Read PILOT plan
cat /Users/halcasteel/PROJECTS/coditect-rollout-master/submodules/core/coditect-core/internal/project/plans/PILOT-PARALLEL-EXECUTION-PLAN.md | grep -A5 "E.2"

# Output shows:
# - [ ] E.2.1: Rate limiting verification
# - [ ] E.2.2: Failed login lockout tests
# ...

# 2. PICK TASK
# Selected: E.2.1 (Rate limiting verification)

# 3. GET AGENT
/which "rate limit security testing"
# → Recommended: testing-specialist

# 4. EXECUTE
/agent testing-specialist "Write pytest tests for E.2.1 rate limiting:
- Test /auth/login returns 429 after 5 requests/minute
- Test /auth/register returns 429 after 3 requests/minute
- Test /licenses/validate returns 429 after 60 requests/minute
- Verify rate limit headers in response"

# OR implement directly:
# Write tests in backend/tests/security/test_rate_limit.py
# Run: pytest backend/tests/security/test_rate_limit.py -v

# 5. MARK DONE - Edit PILOT plan
# Change: - [ ] E.2.1  →  - [x] E.2.1

# 6. LOG - Update session log
/session-log
# Add entry: "E.2.1 Rate limiting tests - 5 tests passing"

# Capture context
/cx

Key Files Reference

Purpose	File Location
Active Tasks (Source of Truth)	`internal/project/plans/PILOT-PARALLEL-EXECUTION-PLAN.md`
Legacy Reference (with mapping)	`docs/project-management/project-plan.md`
Legacy Tasks (with mapping)	`docs/project-management/tasklist.md`
Session History	`docs/session-logs/SESSION-LOG-YYYY-MM-DD.md`
Checkpoints	`docs/CHECKPOINT-YYYY-MM-DD.md`
This Guide	`internal/project/TASK-EXECUTION-GUIDE.md`

Nomenclature Mapping (Legacy → Current)

For historical documents using Phase nomenclature:

Legacy Term	Current Track	Description
Phase 0, 0.5	(Historical)	Infrastructure setup - COMPLETE
Phase 0.8	Track C (DevOps)	Cloud infrastructure
Phase 0.9	Track A (Backend)	Backend API development
Phase 1.0	Track B (Frontend)	Frontend UI development
P0 Critical Path	Tracks A, B, C, D	Cross-cutting priority work
Phase 1	Tracks A-G	Active development
Phase 2+	(Future)	Post-pilot phases

Standard Reference: CODITECT-STD-001

Common Commands

Command	Purpose
`/orient`	Session orientation (5-step MoE workflow)
`/which <task>`	Agent discovery and recommendations
`/agent <name> "task"`	Invoke CODITECT agent
`/cx`	Capture session context
`/cxq "search"`	Query context database
`/session-log`	Update session log
`/checkpoint "desc"`	Create milestone checkpoint
`/classify path/`	MoE document classification

Best Practices

Do's

Always start with /orient or reading the PILOT plan
Use /which to find the right agent before executing
Mark tasks complete immediately after finishing
Log significant work in the session log
Run /cx at session end to preserve context

Don'ts

Don't work from legacy project-plan.md for active tasks
Don't skip logging - it breaks continuity
Don't leave tasks in "in progress" state across sessions
Don't create new tasks without adding to PILOT plan first

Troubleshooting

"Can't find my task"

Check PILOT-PARALLEL-EXECUTION-PLAN.md first
Use grep: grep -i "keyword" internal/project/plans/PILOT*.md
Ask: /cxq "task keyword"

"Wrong agent selected"

Re-run /which with more specific description
Check agent descriptions in agents/ directory
Use built-in Claude Code agent as fallback

"Task dependencies unclear"

Read the full Track section in PILOT plan
Check for BLOCKED BY or DEPENDS ON annotations
Ask about dependencies before starting

Status: Production Ready Last Updated: January 1, 2026 Review Frequency: Update when workflow changes

Overview​

Quick Reference​

Track Nomenclature​

Track Definitions​

Task ID Format​

Quick Track Guide​

Task Execution Workflow​

Step 1: Start of Session​

Step 2: Find Your Next Task​

Step 3: Get Agent Recommendation​

Step 4: Execute Task​

Step 5: Update Status​

Step 6: Log Session​

Practical Example​

Full Workflow Demonstration​

Key Files Reference​

Nomenclature Mapping (Legacy → Current)​

Common Commands​

Best Practices​

Do's​

Don'ts​

Troubleshooting​

"Can't find my task"​

"Wrong agent selected"​

"Task dependencies unclear"​