flow-deliver

Host: Codex CLI — This skill was designed for Claude Code and adapted for Codex. Cross-reference commands use installed skill names in Codex rather than /octo:* slash commands. Use the active Codex shell and subagent tools. Do not claim a provider, model, or host subagent is available until the current session exposes it. For host tool equivalents, see skills/blocks/codex-host-adapter.md.

{{PREAMBLE}}

Pre-Delivery: State Check

Before starting delivery:

1. Read .octo/STATE.md to verify Develop phase complete
2. Update STATE.md:
   • current_phase: 4
   • phase_position: "Delivery"
   • status: "in_progress"

# Verify Develop phase is complete
if [[ -f ".octo/STATE.md" ]]; then
  develop_status=$("${HOME}/.claude-octopus/plugin/scripts/octo-state.sh" get_phase_status 3)
  if [[ "$develop_status" != "complete" ]]; then
    echo "⚠️ Warning: Develop phase not marked complete. Consider completing development first."
  fi
fi

# Update state for Delivery phase
"${HOME}/.claude-octopus/plugin/scripts/octo-state.sh" update_state \
  --phase 4 \
  --position "Delivery" \
  --status "in_progress"

⚠️ EXECUTION CONTRACT (MANDATORY - CANNOT SKIP)

This skill uses ENFORCED execution mode. You MUST follow this exact sequence.

STEP 1: Detect Work Context (MANDATORY)

Analyze the user's prompt and project to determine context:

Knowledge Context Indicators:

• Document terms: "report", "presentation", "PRD", "proposal", "document", "brief"
• Quality terms: "argument", "evidence", "clarity", "completeness", "narrative"

Dev Context Indicators:

• Code terms: "code", "implementation", "API", "endpoint", "function", "module"
• Quality terms: "security", "performance", "tests", "coverage", "bugs"

Also check: What is being reviewed? Code files -> Dev, Documents -> Knowledge

Capture context_type = "Dev" or "Knowledge"

Step 1b: Detect Dev Subtype (if Dev context)

When context_type is Dev, determine the subtype to inject domain-appropriate validation criteria into the review prompt. Append the matching validation supplement after the user's prompt when calling orchestrate.sh in Step 4.

Subtype	Trigger keywords	Validation supplement
frontend-ui	"page", "widget", "component", "UI", "HTML", "CSS", "form", "dashboard", "layout"	Verify: all referenced files exist (scripts, stylesheets, images). Check ARIA labels and roles, keyboard navigability, touch target sizes (44px min). Flag innerHTML usage. Confirm progressive enhancement (fallbacks for navigator.share, localStorage, etc). Test self-containment: does this work if opened/run with zero setup?
cli-tool	"CLI", "command-line", "terminal", "script", "flag", "argument"	Verify: --help flag works, exit codes are meaningful (0/1/2), stderr vs stdout used correctly, argument edge cases handled (missing args, invalid input, --unknown-flag).
api-service	"API", "endpoint", "REST", "GraphQL", "gRPC", "server", "route"	Verify: input validation at every endpoint, consistent error response format, auth on protected routes, rate limiting considered, schema/contract documented.
infra	"deploy", "terraform", "docker", "CI", "pipeline", "Kubernetes", "helm"	Verify: operations are idempotent, no hardcoded secrets, rollback path exists, health checks included, destroy operations require confirmation.
data	"ETL", "pipeline", "migration", "schema", "database", "SQL"	Verify: migrations are reversible, data validation at ingestion, backup strategy documented, no data loss on failure.
general	Default if no subtype matches	No supplement — use standard review criteria.

How to apply: When calling orchestrate.sh in Step 4, append the validation supplement:

orchestrate.sh deliver "<user prompt>\n\nDomain-specific validation criteria:\n<supplement text>"

DO NOT PROCEED TO STEP 2 until context determined. Context type (Dev vs Knowledge) and dev subtype determine which validation supplements to inject — wrong context produces a review that checks irrelevant criteria.

STEP 2: Display Visual Indicators (MANDATORY - BLOCKING)

MANDATORY: You MUST use the native shell command tool to run this provider check BEFORE displaying the banner. Do NOT skip it. Do NOT assume availability.

bash "${HOME}/.claude-octopus/plugin/scripts/helpers/check-providers.sh"

Use the ACTUAL results below. PROHIBITED: Showing only "🔵 Claude: Available ✓" without listing all providers.

If OCTO_ALLOWED_PROVIDERS is set, treat it as the source of truth for which providers may participate. Providers filtered out by that allowlist are intentionally reported as unavailable; do not invoke or recommend them in the workflow.

Display this banner BEFORE orchestrate.sh execution:

For Dev Context:

🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation mode
✅ [Dev] Deliver Phase: [Brief description of code review]

Provider Availability:
🔴 Codex CLI: [Available ✓ / Not installed ✗] - Code quality analysis
🟡 Antigravity CLI: [Available ✓ / Not installed ✗] - Security and edge cases
🧭 Antigravity CLI: [Available ✓ / Not installed ✗] - Additional external-model challenge
🔵 Claude: Available ✓ - Synthesis and recommendations

💰 Estimated Cost: $0.02-0.08
⏱️  Estimated Time: 3-7 minutes

For Knowledge Context:

🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation mode
✅ [Knowledge] Deliver Phase: [Brief description of document review]

Provider Availability:
🔴 Codex CLI: [Available ✓ / Not installed ✗] - Structure and logic analysis
🟡 Antigravity CLI: [Available ✓ / Not installed ✗] - Content quality and completeness
🧭 Antigravity CLI: [Available ✓ / Not installed ✗] - Additional external-model challenge
🔵 Claude: Available ✓ - Synthesis and recommendations

💰 Estimated Cost: $0.02-0.08
⏱️  Estimated Time: 3-7 minutes

DO NOT PROCEED TO STEP 3 until banner displayed. The banner shows users which providers will run and what costs they'll incur — starting API calls without this visibility violates cost transparency.

STEP 3: Read Prior State (MANDATORY - State Management)

Before executing the workflow, read full project context:

# Initialize state if needed
"${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" init_state

# Set current workflow
"${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" set_current_workflow "flow-deliver" "deliver"

# Get all prior decisions (critical for validation)
prior_decisions=$("${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" get_decisions "all")

# Get context from all prior phases
discover_context=$("${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" get_context "discover")
define_context=$("${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" get_context "define")
develop_context=$("${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" get_context "develop")

# Display what you found (validation needs full context)
echo "📋 Validation Context Summary:"

if [[ "$discover_context" != "null" ]]; then
  echo "  Discovery: $discover_context"
fi

if [[ "$define_context" != "null" ]]; then
  echo "  Definition: $define_context"
fi

if [[ "$develop_context" != "null" ]]; then
  echo "  Development: $develop_context"
fi

if [[ "$prior_decisions" != "[]" && "$prior_decisions" != "null" ]]; then
  echo "  Decisions to validate against:"
  echo "$prior_decisions" | jq -r '.[] | "    - \(.decision) (\(.phase))"'
fi

This provides full context for validation:

• Requirements and scope (from define phase)
• Implementation decisions (from develop phase)
• Research findings (from discover phase)
• All architectural decisions to validate against
• If claude-mem is installed, its MCP tools (search, timeline, get_observations) are available — use them to check for past delivery issues or quality patterns

DO NOT PROCEED TO STEP 4 until state read.

STEP 4: Execute orchestrate.sh deliver (MANDATORY - Use Bash Tool)

You MUST execute this command via the native shell command tool:

${HOME}/.claude-octopus/plugin/scripts/orchestrate.sh deliver "<user's validation request>"

CRITICAL: You are PROHIBITED from:

• ❌ Reviewing directly without calling orchestrate.sh — adversarial multi-AI review catches blind spots that a single reviewer misses; Codex finds code quality issues while Antigravity catches security and edge cases
• ❌ Doing single-perspective analysis instead of multi-provider
• ❌ Claiming you're "simulating" the workflow
• ❌ Proceeding to Step 4 without running this command

You MUST use the native shell command tool to invoke orchestrate.sh.

What Users See During Execution (v7.16.0+)

If running in Claude Code v2.1.16+, users will see real-time progress indicators in the task spinner:

Phase 1 - External Provider Execution (Parallel):

• 🔴 Analyzing code quality and patterns (Codex)...
• 🟡 Validating security and edge cases (Antigravity)...

Phase 2 - Synthesis (Sequential):

• 🔵 Synthesizing validation results...

These spinner verb updates happen automatically - orchestrate.sh calls update_task_progress() before each agent execution. Users see exactly which provider is working and what it's doing.

If NOT running in Claude Code v2.1.16+: Progress indicators are silently skipped, no errors shown.

STEP 5: Verify Execution (MANDATORY - Validation Gate)

After orchestrate.sh completes, verify it succeeded:

# Find the latest validation file (created within last 10 minutes)
VALIDATION_FILE=$(find ~/.claude-octopus/results -name "ink-validation-*.md" -mmin -10 2>/dev/null | head -n1)

if [[ -z "$VALIDATION_FILE" ]]; then
  echo "❌ VALIDATION FAILED: No validation file found"
  echo "orchestrate.sh did not execute properly"
  exit 1
fi

echo "✅ VALIDATION PASSED: $VALIDATION_FILE"
cat "$VALIDATION_FILE"

If validation fails:

1. Report error to user
2. Show logs from ~/.claude-octopus/logs/
3. DO NOT proceed with presenting results
4. DO NOT substitute with direct review — fallback to single-model review defeats the adversarial multi-provider validation that catches blind spots

STEP 6: Update State (MANDATORY - Post-Execution)

After validation is complete, record final metrics:

# Update deliver phase context with validation summary
validation_summary=$(head -30 "$VALIDATION_FILE" | grep -A 2 "## Summary\|Pass\|Fail" | tail -2 | tr '\n' ' ')

"${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" update_context \
  "deliver" \
  "$validation_summary"

# Update final metrics (completion of full workflow)
"${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" update_metrics "phases_completed" "1"

# Display final state summary
echo ""
echo "📊 Session Complete - Final Metrics:"
"${HOME}/.claude-octopus/plugin/scripts/state-manager.sh" show_summary

DO NOT PROCEED TO STEP 7 until state updated.

STEP 7: Present Validation Report & Post to PR (Only After Steps 1-6 Complete)

Read the validation file and present:

• Overall status (✅ PASSED / ⚠️ PASSED WITH WARNINGS / ❌ FAILED)
• Quality score (XX/100)
• Summary
• Critical issues (must fix)
• Warnings (should fix)
• Recommendations (nice to have)
• Validation details from all providers
• Quality gates results
• Next steps

Include attribution:

*Multi-AI Validation powered by Claude Octopus*
*Providers: available external providers + 🔵 Claude*
*Full validation report: $VALIDATION_FILE*

Post to PR (v8.44.0)

After presenting the report, check if the current branch has an open PR and post the validation summary as a PR comment:

CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "")
PR_NUM=""

if [[ -n "$CURRENT_BRANCH" && "$CURRENT_BRANCH" != "main" && "$CURRENT_BRANCH" != "master" ]]; then
    if command -v gh &>/dev/null; then
        PR_NUM=$(gh pr list --head "$CURRENT_BRANCH" --json number --jq '.[0].number' 2>/dev/null || echo "")
    fi
fi

if [[ -n "$PR_NUM" ]]; then
    # Extract summary section from validation file for PR comment
    REVIEW_SUMMARY=$(head -60 "$VALIDATION_FILE")

    REPO_SLUG=$(gh repo view --json nameWithOwner --jq .nameWithOwner)
    COMMENT_BODY="## Deliver Phase — Validation Report

${REVIEW_SUMMARY}

___
*Multi-AI validation by Claude Octopus (/octo:deliver)*
*Providers: available external providers + 🔵 Claude*"
    if ! "${CLAUDE_PLUGIN_ROOT:-${HOME}/.claude-octopus/plugin}/scripts/safe-gh-comment.sh" \
            --repo "$REPO_SLUG" pr-comment "$PR_NUM" - <<< "$COMMENT_BODY"; then
        echo "GitHub write state is unknown; check for the validation report before retrying:" >&2
        gh pr view "$PR_NUM" --repo "$REPO_SLUG" --comments || true
        return 1 2>/dev/null || exit 1
    fi

    echo "Validation report posted to PR #${PR_NUM}"

    # Update agent registry
    REGISTRY="${HOME}/.claude-octopus/plugin/scripts/agent-registry.sh"
    if [[ -x "$REGISTRY" ]]; then
        AGENT_ID="$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "")"
        "$REGISTRY" update "$AGENT_ID" --pr "$PR_NUM" 2>/dev/null || true
    fi
fi

Behavior:

• Auto-posts when running inside /octo:embrace or /octo:factory
• When running standalone via /octo:deliver, asks user first:

  "PR #N found. Post validation report as a PR comment?"
  Options: "Yes, post to PR", "No, terminal only"
  

• If no PR or gh CLI unavailable, skips silently

Deliver Workflow - Deliver Phase ✅

⚠️ MANDATORY: Context Detection & Visual Indicators

BEFORE executing ANY workflow actions, you MUST:

Step 1: Detect Work Context

Analyze the user's prompt and project to determine context:

Knowledge Context Indicators (in prompt):

• Document terms: "report", "presentation", "PRD", "proposal", "document", "brief"
• Quality terms: "argument", "evidence", "clarity", "completeness", "narrative"

Dev Context Indicators (in prompt):

• Code terms: "code", "implementation", "API", "endpoint", "function", "module"
• Quality terms: "security", "performance", "tests", "coverage", "bugs"

Also check: What is being reviewed? Code files -> Dev, Documents -> Knowledge

Step 1b: Detect Dev Subtype — see EXECUTION CONTRACT Step 1b above for subtype table and validation supplements. Append the matching supplement to the prompt before calling orchestrate.sh.

Step 2: Output Context-Aware Banner

For Dev Context:

🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation mode
✅ [Dev] Deliver Phase: [Brief description of code review]
📋 Session: ${CLAUDE_SESSION_ID}

Providers:
🔴 Codex CLI - Code quality analysis
🟡 Antigravity CLI - Security and edge cases
🔵 Claude - Synthesis and recommendations

For Knowledge Context:

🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation mode
✅ [Knowledge] Deliver Phase: [Brief description of document review]
📋 Session: ${CLAUDE_SESSION_ID}

Providers:
🔴 Codex CLI - Structure and logic analysis
🟡 Antigravity CLI - Content quality and completeness
🔵 Claude - Synthesis and recommendations

{{VISUAL_INDICATORS}}

Part of Double Diamond: DELIVER (convergent thinking)

        DELIVER (ink)

         \         /
          \       /
           \     /
            \   /
             \ /

          Converge to
           delivery

What This Workflow Does

The deliver phase validates and reviews implementations using external CLI providers:

1. 🔴 Codex CLI - Code quality, best practices, technical correctness
2. 🟡 Antigravity CLI - Security audit, edge cases, user experience
3. 🔵 Claude (You) - Synthesis and final validation report

This is the convergent phase for delivery - we ensure quality before shipping.

When to Use Deliver

Use deliver when you need:

Dev Context Examples

• Code Review: "Review the authentication implementation"
• Security Audit: "Check for security vulnerabilities in auth.ts"
• Quality Validation: "Validate the API endpoints are production-ready"
• Implementation Verification: "Verify the caching layer works correctly"
• Pre-Deployment Check: "Ensure the feature is ready to ship"

Knowledge Context Examples

• Document Review: "Review the PRD for completeness"
• Presentation Validation: "Check the executive presentation for clarity"
• Report Quality Check: "Validate the market analysis report"
• Proposal Review: "Review the business case for stakeholder readiness"
• Content Audit: "Ensure the strategy document is actionable"

Don't use deliver for:

• Building implementations (use develop-workflow)
• Research and exploration (use discover-workflow)
• Requirement definition (use define-workflow)
• Simple code/document reading (use Read tool)

Visual Indicators

Before execution, you'll see:

🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation
✅ Deliver Phase: Reviewing and validating implementation

Providers:
🔴 Codex CLI - Code quality and best practices
🟡 Antigravity CLI - Security and edge cases
🔵 Claude - Synthesis and validation report

How It Works

Step 1: Invoke Ink Phase

${HOME}/.claude-octopus/plugin/scripts/orchestrate.sh deliver "<user's validation request>"

Step 2: Multi-Provider Validation

The orchestrate.sh script will:

1. Call Codex CLI for code quality analysis
2. Call Antigravity CLI for security and edge case review
3. You (Claude) synthesize findings into validation report
4. Generate quality scores and recommendations

Step 3: Quality Gates (Automatic)

The ink phase includes automatic quality validation via PostToolUse hook:

• Code Quality: Complexity, maintainability, documentation
• Security: OWASP top 10, authentication, input validation
• Best Practices: Error handling, logging, testing
• Completeness: Missing functionality, edge cases

Step 4: Read Results

Results are saved to:

~/.claude-octopus/results/${SESSION_ID}/ink-validation-<timestamp>.md

Step 5: Present Validation Report

Read the synthesis and present findings with quality scores to the user.

Implementation Instructions

When this skill is invoked, follow the EXECUTION CONTRACT above exactly. The contract includes:

1. Blocking Step 1: Detect work context (Dev vs Knowledge)
2. Blocking Step 2: Check providers, display visual indicators
3. Blocking Step 3: Execute orchestrate.sh deliver via native shell command tool
4. Blocking Step 4: Verify validation file exists
5. Step 5: Present formatted validation report

Each step is mandatory and blocking - you cannot proceed to the next step until the current one completes successfully.

Task Management Integration

Create tasks to track execution progress:

// At start of skill execution
TaskCreate({
  subject: "Execute deliver workflow with multi-AI providers",
  description: "Run orchestrate.sh deliver for validation",
  activeForm: "Running multi-AI deliver workflow"
})

// Mark in_progress when calling orchestrate.sh
TaskUpdate({taskId: "...", status: "in_progress"})

// Mark completed ONLY after validation report presented
TaskUpdate({taskId: "...", status: "completed"})

Error Handling

If any step fails:

• Step 1 (Context): Default to Dev Context if ambiguous
• Step 2 (Providers): If all external providers are unavailable, suggest /octo:setup and STOP
• Step 3 (orchestrate.sh): Show bash error, check logs, report to user
• Step 4 (Validation): If validation file missing, show orchestrate.sh logs, DO NOT substitute with direct review

Never fall back to direct review if orchestrate.sh execution fails. Report the failure and let the user decide how to proceed.

Validation Report Format

After successful execution, present validation report with:

# Validation Report: <task>

## Overall Status: ✅ PASSED / ⚠️ PASSED WITH WARNINGS / ❌ FAILED

**Quality Score**: XX/100

## Summary
[Brief summary of validation findings]

## Critical Issues (Must Fix)
- [ ] Issue 1: [Description]
- [ ] Issue 2: [Description]

## Warnings (Should Fix)
- [ ] Warning 1: [Description]
- [ ] Warning 2: [Description]

## Recommendations (Nice to Have)
- [ ] Recommendation 1: [Description]
- [ ] Recommendation 2: [Description]

## Validation Details

### Code Quality (Codex Analysis)
**Score**: XX/100
- [Specific findings about code quality]
- [Best practice violations]
- [Improvement suggestions]

### Security Audit (Antigravity Analysis)
**Score**: XX/100
- [Security vulnerabilities found]
- [Authentication/authorization issues]
- [Input validation concerns]

### Final Validation (Claude Synthesis)
**Score**: XX/100
- [Integrated assessment]
- [Risk analysis]
- [Go/no-go recommendation]

## Quality Gates
- [X] Code Quality: XX/100 (PASSED/FAILED)
- [X] Security: XX/100 (PASSED/FAILED)
- [X] Best Practices: XX/100 (PASSED/FAILED)
- [X] Completeness: XX/100 (PASSED/FAILED)

## Next Steps
[Recommended actions based on findings]

Full validation report saved to: <validation file path>

Example Usage

Example 1: Review Authentication Implementation

User: Review the authentication implementation in src/auth/

Claude:
🐙 **CLAUDE OCTOPUS ACTIVATED** - Multi-provider validation
✅ Deliver Phase: Reviewing authentication implementation

[Executes: ${HOME}/.claude-octopus/plugin/scripts/orchestrate.sh deliver "Review authentication implementation in src/auth/"]

[After completion, reads validation and presents:]

# Validation Report: Authentication Implementation

## Overall Status: ⚠️ PASSED WITH WARNINGS

**Quality Score**: 78/100

## Summary

The authentication implementation is functional and follows most best practices, but has security concerns with token storage and missing rate limiting. Code quality is good with TypeScript types and error handling. Recommend addressing security warnings before production deployment.

## Critical Issues (Must Fix)
- [ ] **Security**: JWT secret is hardcoded in auth/config.ts (should use env variable)
- [ ] **Security**: No rate limiting on /login endpoint (vulnerability to brute force)

## Warnings (Should Fix)
- [ ] **Best Practice**: Password validation regex could be stronger (min 12 chars, special chars)
- [ ] **Testing**: Missing unit tests for token refresh logic
- [ ] **Error Handling**: Generic error messages leak whether email exists (account enumeration)

## Recommendations (Nice to Have)
- [ ] **UX**: Add "remember me" functionality for longer sessions
- [ ] **Monitoring**: Add logging for failed login attempts
- [ ] **Documentation**: Add JSDoc comments for public API methods

## Validation Details

### Code Quality (Codex Analysis)
**Score**: 85/100

**Strengths**:
- Clean TypeScript with proper types
- Good separation of concerns (controller/service/middleware)
- Proper async/await error handling
- Express middleware pattern used correctly

**Issues**:
- Missing JSDoc documentation
- Some functions exceed 50 lines (refactor recommended)
- No unit tests for edge cases

**Code Examples**:
\`\`\`typescript
// Good: Proper error handling
try {
  const token = jwt.sign(payload, JWT_SECRET, { expiresIn: '15m' });
  return token;
} catch (error) {
  logger.error('Token generation failed', error);
  throw new AuthError('Failed to generate token');
}

// Issue: Hardcoded secret (security risk)
const JWT_SECRET = 'my-secret-key-123'; // Should be process.env.JWT_SECRET
\`\`\`

### Security Audit (Antigravity Analysis)
**Score**: 68/100

**Critical Vulnerabilities**:
- ❌ Hardcoded JWT secret (HIGH severity)
- ❌ No rate limiting on auth endpoints (MEDIUM severity)
- ❌ Account enumeration via error messages (MEDIUM severity)