codex-project-pulse

TL;DR

Project dashboard brain. Maintains persistent state in .codex/pulse/ — sprint, priorities, blockers, risks, milestones. When user asks "hôm nay thế nào?" or "$today", read all state files, run quality check, and generate a comprehensive daily brief. This is the skill that makes the agent know WHERE the project is and WHAT to do next.

Activation

Activate on $pulse, $today, $daily, $status, $brief.
Activate on "hôm nay thế nào?", "status update", "what's next?", "what should I work on?", "project status", "daily standup", "morning brief".
Activate on $sprint-init, $done, $block, $unblock, $milestone, $priority, $risk.
Activate on "bắt đầu sprint", "hoàn thành story", "bị block", "deadline", "rủi ro".
Auto-load with codex-workflow-autopilot when routing daily check-ins.
Auto-load with codex-scrum-subagents during sprint ceremonies.
Auto-suggest on session start if .codex/pulse/sprint-state.json exists.

Announce at start: "Loading project pulse — checking sprint state, priorities, and blockers."

State Files

All project state lives in .codex/pulse/ at the project root:

.codex/pulse/
├── sprint-state.json       # Current sprint: goal, stories, status, dates
├── priority-queue.json     # Ordered list of what to do next
├── blockers.json           # Active blockers with owners and age
├── risk-register.json      # Risks with severity, probability, mitigation
├── milestones.json         # Deadlines and milestone tracking
└── daily-brief.md          # Latest generated morning brief

Command Reference

Daily Operations

Command	Action	Example
`$today` / `$pulse`	Generate daily brief	"Hôm nay thế nào?"
`$priority`	Show priority queue	"Việc gì cần làm trước?"
`$priority reorder`	Reorder priorities	"Đổi thứ tự ưu tiên"

Sprint Management

Command	Action	Example
`$sprint-init`	Initialize new sprint	"$sprint-init Sprint 4 --goal 'Complete payment flow' --end 2026-05-10"
`$sprint-close`	Close current sprint, calculate velocity	"$sprint-close"
`$add-story`	Add story to sprint backlog	"$add-story PAY-001 'Stripe checkout' --points 5 --owner @dev-a"
`$done <id>`	Mark story complete	"$done PAY-001"
`$wip <id>`	Mark story in-progress	"$wip PAY-002"

Blocker Management

Command	Action	Example
`$block <id> <reason>`	Add blocker	"$block PAY-001 'Waiting for Stripe API key'"
`$unblock <id>`	Remove blocker	"$unblock PAY-001"
`$blockers`	List all active blockers	"$blockers"

Risk & Milestone Management

Command	Action	Example
`$risk <desc>`	Add risk	"$risk 'Payment integration delay' --severity high"
`$risk resolve <id>`	Close a risk	"$risk resolve RISK-003"
`$milestone <name> <date>`	Add milestone	"$milestone 'Demo day' 2026-05-15"
`$milestones`	Show all milestones with proximity	"$milestones"

Daily Brief Protocol

When $today or $pulse is triggered:

Step 1: Read State

1. Read .codex/pulse/sprint-state.json → sprint progress
2. Read .codex/pulse/priority-queue.json → what's next
3. Read .codex/pulse/blockers.json → what's stuck
4. Read .codex/pulse/risk-register.json → active risks
5. Read .codex/pulse/milestones.json → upcoming deadlines

Step 2: Gather Signals

6. Check git log --since=yesterday → recent commits
7. Run quality check if project has test/lint commands
8. Read latest .codex/sessions/ summary → yesterday's context
9. Count open TODOs/FIXMEs if available

Step 3: Generate Brief

10. Calculate sprint progress (stories done / total, time elapsed / total)
11. Identify overdue milestones
12. Identify stale blockers (> 2 days old)
13. Generate priority recommendations
14. Write .codex/pulse/daily-brief.md
15. Present to user

Step 4: Recommend Action

16. If blockers exist → recommend unblocking action first
17. If deadline < 3 days → flag urgency
18. If quality degrading → recommend quality fix before features
19. Otherwise → recommend top priority from queue

Daily Brief Output Format

# 📊 Daily Brief — [Date]

## Sprint [N]: [Goal]
**Progress:** [X/Y stories] ([Z]%) | **Time:** Day [D/T] ([P]%)
**Velocity risk:** [On track / Behind / Ahead]

[Progress bar visualization]

## 🔴 Blockers ([count])
| Story | Blocker | Age | Owner | Action Needed |
|---|---|---|---|---|
| [id] | [reason] | [days] | [who] | [next step] |

## ⚠️ Risks ([count active])
| Risk | Severity | Probability | Mitigation |
|---|---|---|---|
| [desc] | 🔴/🟡/🟢 | H/M/L | [action] |

## ⏰ Upcoming Deadlines
| Milestone | Date | Days Left | Status |
|---|---|---|---|
| [name] | [date] | [N days] | 🟢/🟡/🔴 |

## 📈 Quality Pulse
| Signal | Status | Trend |
|---|---|---|
| Tests | [pass rate] | [↑/↓/→] |
| Lint | [error count] | [↑/↓/→] |
| Tech debt | [signal count] | [↑/↓/→] |
| Security | [finding count] | [↑/↓/→] |

## 📋 Today's Priorities
1. 🔴 [Urgent: unblock/fix] — [reason]
2. 🟡 [High: continue/start] — [story or task]
3. 🟢 [Normal: next in queue] — [story or task]

## 💡 Yesterday's Progress
[Summary from latest session or git log]

Sprint State Management

Sprint Lifecycle

stateDiagram-v2
    [*] --> Planning: $sprint-init
    Planning --> Active: Stories added
    Active --> Active: $done, $wip, $block
    Active --> Closing: All stories done or sprint end date
    Closing --> Review: $sprint-close
    Review --> [*]: Velocity recorded

    Active --> Active: Daily $pulse

Velocity Tracking

After each sprint close, record:

{
  "sprint": "Sprint 3",
  "planned_points": 21,
  "delivered_points": 18,
  "velocity": 18,
  "stories_planned": 7,
  "stories_completed": 6,
  "carryover": ["PAY-003"]
}

Use velocity history to forecast future sprint capacity:

recommended_commitment = average(last 3 sprints velocity) × 0.85

Priority Calculation Rules

When ordering the priority queue:

Priority Score = base_priority + urgency_bonus + blocker_penalty + deadline_bonus

base_priority:
  - critical: 100
  - high: 70
  - medium: 40
  - low: 10

urgency_bonus:
  - blocked_dependency: +30 (something else waiting on this)
  - stale_wip > 3 days: +20

blocker_penalty:
  - currently_blocked: -50 (deprioritize until unblocked)

deadline_bonus:
  - milestone < 3 days: +40
  - milestone < 7 days: +20
  - milestone < 14 days: +10

State Initialization

When no .codex/pulse/ exists, on $sprint-init:

Create .codex/pulse/ directory
Ask user for:
- Sprint name and number
- Sprint goal (one sentence)
- Start date and end date
- Stories to include (or import from backlog)
Generate initial state files
Generate first daily brief

When .codex/pulse/ already exists but sprint is closed:

Archive previous sprint to .codex/pulse/archive/sprint-N/
Initialize new sprint state
Carry over incomplete stories (if user confirms)

Integration with Other Skills

Skill	Integration
`codex-project-memory`	Read session summaries for "yesterday's progress"; read decisions for context
`codex-execution-quality-gate`	Run quality scripts for "quality pulse" section of daily brief
`codex-scrum-subagents`	Use ceremony workflows for sprint events; share sprint state
`codex-workflow-autopilot`	Route `$today` to pulse; use priority queue for "what next?" decisions
`codex-context-engine`	Load genome.md for project structure awareness alongside sprint state
`codex-verification-discipline`	Verify story completion claims with evidence before marking $done
`codex-document-writer`	Use report templates for sprint reports and progress updates

Hard Rules

Never mark a story done without verification evidence (tests, build, demo).
Never remove a blocker without confirming the blocking condition is resolved.
Never ignore a milestone < 3 days away — always flag in daily brief.
Always read state files before generating any brief — no guessing from memory.
Always persist state changes to files — memory alone decays between sessions.
Stale blockers (> 2 days) get escalation flag automatically.
Sprint velocity is calculated from actual delivered points, not planned.
When quality degrades (test failures, new critical debt), flag before feature work.

Reference Files

references/state-schemas.md: JSON schemas for all .codex/pulse/ state files.
references/daily-brief-template.md: daily brief format, quality rules, and examples.
references/priority-engine.md: priority calculation, reordering rules, and queue management.