@alevental/cccp 0.1.0

package/examples/agents/diff-evaluator.md

@@ -0,0 +1,57 @@
+---
+name: diff-evaluator
+description: Compares an actual output against a ground truth expected output and produces a structured evaluation.
+---
+
+# Diff Evaluator Agent
+
+You are a precision evaluator. Your job is to compare an actual output against a known-correct expected output and determine whether they match in substance.
+
+## Instructions
+
+1. Read the actual output (provided as an input).
+2. Read the ground truth expected output at the path specified in the Ground Truth section.
+3. Compare them across all evaluation criteria.
+4. Write a structured evaluation to the output path.
+
+## Evaluation criteria
+
+Evaluate on these dimensions:
+
+| Criterion | What to check |
+|-----------|---------------|
+| **Completeness** | Does the actual output cover all points present in the expected output? |
+| **Accuracy** | Are the facts, claims, and details correct relative to the expected output? |
+| **Structure** | Does the actual output follow the same organizational structure? |
+| **Tone** | Does the actual output match the expected tone and voice? |
+| **Conciseness** | Is the actual output similar in length — no major omissions or padding? |
+
+## Output format
+
+Your evaluation MUST use this exact format:
+
+```markdown
+## Evaluation
+
+### Criterion Results
+
+| # | Criterion | Result | Evidence |
+|---|-----------|--------|----------|
+| 1 | Completeness | PASS/FAIL | [specific evidence] |
+| 2 | Accuracy | PASS/FAIL | [specific evidence] |
+| 3 | Structure | PASS/FAIL | [specific evidence] |
+| 4 | Tone | PASS/FAIL | [specific evidence] |
+| 5 | Conciseness | PASS/FAIL | [specific evidence] |
+
+### Overall: PASS or FAIL
+
+### Iteration Guidance
+
+[If FAIL: specific, actionable guidance on what to change in the prompt to fix each failing criterion]
+```
+
+## Rules
+
+- PASS means all criteria pass. Any single FAIL means Overall: FAIL.
+- Be specific in evidence — quote or reference exact passages.
+- Iteration guidance should tell the prompt tuner exactly what to add, remove, or change.

package/examples/agents/prompt-tuner.md

@@ -0,0 +1,30 @@
+---
+name: prompt-tuner
+description: Reads evaluation feedback and adjusts a prompt artifact to better produce the desired output.
+---
+
+# Prompt Tuner Agent
+
+You are a prompt optimization specialist. Your job is to read an evaluation of a prompt's output, understand where the output diverged from the expected result, and adjust the prompt to close that gap.
+
+## Instructions
+
+1. Read the current prompt artifact at the path specified in your task.
+2. Read the evaluation feedback from the previous iteration to understand what went wrong.
+3. Read the ground truth (expected output) to understand the target.
+4. Modify the prompt artifact to address the specific issues identified in the evaluation.
+5. Write the adjusted prompt back to the same output path.
+
+## Adjustment strategies
+
+- **Missing content**: Add explicit instructions for the content that was absent.
+- **Wrong structure**: Add formatting directives or structural constraints.
+- **Tone mismatch**: Add tone/voice guidance or examples.
+- **Extra content**: Add constraints like "Do not include..." or "Only cover...".
+- **Ordering issues**: Specify the expected order of sections or points.
+
+## Constraints
+
+- Preserve the core intent of the prompt — only adjust what the evaluation flagged.
+- Make minimal, targeted changes rather than rewriting from scratch.
+- Each adjustment should be traceable to a specific evaluation criterion.

package/examples/agents/summarizer.md

@@ -0,0 +1,14 @@
+---
+name: summarizer
+description: Reads a project directory and writes a short summary of what it contains.
+---
+
+# Summarizer Agent
+
+You are a project summarizer. Your job is to read the project directory and produce a concise summary of what the project is, what technologies it uses, and its overall structure.
+
+## Instructions
+
+1. Read the project's README, package.json, or other top-level files to understand what it is.
+2. List the main directories and their purposes.
+3. Write a clear, concise summary (under 500 words) to the output path specified in your task.

package/examples/autoresearch-artifacts/expected-output.md

@@ -0,0 +1,17 @@
+# Acme Corp Q4 2025 Summary
+
+Acme Corp delivered strong Q4 results with $4.2B revenue (+18% YoY) and $712M net income ($3.56/share). Operating margin expanded to 21.2%.
+
+## Segment Highlights
+
+- **Cloud Services** led growth at $2.1B (+31%), crossing $8B ARR with 97% retention
+- **Enterprise Software** grew 8% to $1.4B as subscription mix reached 71%
+- **Professional Services** dipped 3% to $700M but maintains $1.8B backlog
+
+## Financial Health
+
+The company generated $1.1B free cash flow (26% margin) and added 1,800 net new customers, bringing the total to 14,200. R&D investment was $630M (15% of revenue).
+
+## Outlook
+
+FY2026 guidance calls for $18.5-19.0B revenue (16-19% growth) with 22-23% operating margin and $4.5-4.8B FCF. A $2B buyback was announced. CEO Jane Martinez highlighted accelerating enterprise AI adoption and expects cloud to exceed 60% of revenue by year-end.

package/examples/autoresearch-artifacts/prompt.md

@@ -0,0 +1,35 @@
+# Summarization Prompt
+
+Read the source material and write a concise summary targeting **~170 words** (never exceed 200 words).
+
+## Content requirements
+
+Include key financial metrics, segment KPIs, and forward guidance. Specifically preserve:
+- Revenue, net income, and EPS
+- Operating margin
+- Each segment's revenue, growth rate, and one defining KPI (e.g., ARR, retention rate, subscription mix, backlog)
+- Free cash flow and FCF margin
+- Total customer count and net new customers
+- R&D spend as a percentage of revenue
+- Full forward guidance (revenue range, margin target, FCF target)
+- Any capital return announcements and CEO forward-looking commentary
+
+## Compression rules
+
+- **Report current-period metrics only.** Do not include prior-period comparisons or baselines (e.g., omit prior-year margin figures).
+- **Omit sub-segment drivers and explanations.** Report each segment's topline number, growth rate, and one key metric — do not explain what drove the result.
+- **Paraphrase executive commentary.** Do not use direct quotes; summarize the CEO's remarks in your own words.
+- **One line per segment bullet.** Each segment highlight must fit on a single line.
+
+## Output structure
+
+Use this format:
+1. **Title** as a top-level heading (# Title)
+2. **Intro paragraph** — 1-2 sentences covering top-line results (revenue, growth, net income, EPS, operating margin)
+3. **Segment Highlights** section (## heading) — one bullet per business segment with revenue, growth, and key metric
+4. **Financial Health** section (## heading) — short paragraph covering FCF, customer count, and R&D
+5. **Outlook** section (## heading) — short paragraph covering FY guidance, buyback, and CEO forward-looking remarks (paraphrased)
+
+## Tone
+
+Use a professional, factual, report-style tone. Do not editorialize or add analysis beyond what the source states.

package/examples/autoresearch-artifacts/source-material.md

@@ -0,0 +1,28 @@
+# Acme Corp Q4 2025 Earnings Report
+
+## Financial Highlights
+
+Acme Corp reported revenue of $4.2 billion for Q4 2025, up 18% year-over-year. Operating income reached $890 million, representing a 21.2% operating margin, compared to 19.1% in Q4 2024. Net income was $712 million, or $3.56 per diluted share.
+
+## Segment Performance
+
+**Cloud Services** ($2.1B, +31% YoY): The largest segment, driven by enterprise migrations and AI workload demand. Annual recurring revenue crossed $8B for the first time. Customer retention rate held steady at 97%.
+
+**Enterprise Software** ($1.4B, +8% YoY): Steady growth from subscription transitions. Legacy perpetual license revenue declined 22%, but subscription revenue grew 34%, now representing 71% of segment revenue.
+
+**Professional Services** ($700M, -3% YoY): Slight decline due to completion of several large implementation contracts. Backlog remains healthy at $1.8B, with three new Fortune 100 engagements signed in the quarter.
+
+## Key Metrics
+
+- Free cash flow: $1.1B (26% FCF margin)
+- Total customers: 14,200 (+1,800 net new)
+- Employees: 28,500 (+2,100 YoY)
+- R&D spend: $630M (15% of revenue)
+
+## Guidance
+
+For FY2026, Acme expects revenue of $18.5-19.0B (16-19% growth), operating margin of 22-23%, and free cash flow of $4.5-4.8B. The company announced a $2B share repurchase program.
+
+## CEO Commentary
+
+"Our cloud business reached an inflection point this quarter," said CEO Jane Martinez. "Enterprise AI adoption is accelerating faster than our most optimistic projections. We're investing aggressively in GPU infrastructure and expect cloud to surpass 60% of total revenue by end of FY2026."

package/examples/business-case.yaml

@@ -0,0 +1,41 @@
+name: business-case
+description: Build a business case for a new initiative.
+
+stages:
+  - name: market-research
+    type: agent
+    task: "Research the market opportunity for the proposed initiative."
+    agent: researcher
+    output: "{artifact_dir}/market-research.md"
+
+  - name: financial-analysis
+    type: agent
+    task: "Analyze the financial implications and produce a cost-benefit summary."
+    agent: analyst
+    output: "{artifact_dir}/financial-analysis.md"
+
+  - name: business-case
+    type: pge
+    task: "Write the business case for the initiative."
+    inputs:
+      - "{artifact_dir}/market-research.md"
+      - "{artifact_dir}/financial-analysis.md"
+    planner:
+      agent: strategist
+      operation: business-case
+    generator:
+      agent: strategist
+      operation: business-case
+    evaluator:
+      agent: exec-reviewer
+    contract:
+      deliverable: "{artifact_dir}/business-case.md"
+      guidance: "Must include problem statement, market opportunity, proposed solution, financial projections, risks, and recommendation."
+      max_iterations: 3
+
+  - name: exec-approval
+    type: human_gate
+    prompt: "Review the business case. Approve to proceed with the initiative."
+    artifacts:
+      - "{artifact_dir}/business-case.md"
+      - "{artifact_dir}/financial-analysis.md"

package/examples/cccp.yaml

@@ -0,0 +1,48 @@
+# Example CCCP project configuration
+# Place this file at your project root as cccp.yaml
+
+# Directories to search for agent definitions (in priority order).
+# Paths are resolved relative to this file.
+agent_paths:
+  - ./agents
+  - ./.claude/agents
+
+# Named MCP server profiles.
+# Each profile defines which MCP servers are available to agents using it.
+# Use "extends" to inherit from another profile.
+mcp_profiles:
+  base:
+    servers:
+      qmd:
+        command: qmd
+        args: [serve, --stdio]
+
+  design:
+    extends: base
+    servers:
+      figma:
+        command: npx
+        args: [-y, figma-console-mcp]
+
+  qa:
+    extends: base
+    servers:
+      ios-simulator:
+        command: npx
+        args: [-y, "@anthropic/ios-simulator-mcp"]
+
+  research:
+    extends: base
+    servers:
+      google-workspace:
+        command: npx
+        args: [-y, google-workspace-mcp]
+
+# Claude config directory for agent subprocesses.
+claude_config_dir: /Users/alevental/.claude-iceman713
+
+# Default artifact output directory pattern.
+artifact_dir: docs/projects/{project}/{pipeline_name}
+
+# Default MCP profile when a stage doesn't specify one.
+default_mcp_profile: base

package/examples/content-calendar.yaml

@@ -0,0 +1,59 @@
+name: content-calendar
+description: Plan and produce a month of marketing content.
+
+stages:
+  - name: content-strategy
+    type: pge
+    task: "Create a content calendar for the upcoming month."
+    planner:
+      agent: marketer
+      operation: content
+    generator:
+      agent: marketer
+      operation: content
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/content-calendar.md"
+      guidance: "Must include 4 content pieces with topics, target audience, channel, and key messages."
+      max_iterations: 2
+
+  - name: calendar-approval
+    type: human_gate
+    prompt: "Review the content calendar. Approve to begin content production."
+    artifacts:
+      - "{artifact_dir}/content-calendar.md"
+
+  - name: article-1
+    type: pge
+    task: "Write the first article from the content calendar."
+    inputs:
+      - "{artifact_dir}/content-calendar.md"
+    planner:
+      agent: marketer
+      operation: content
+    generator:
+      agent: copywriter
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/article-1.md"
+      guidance: "Must align with content calendar brief. Engaging, clear, and SEO-aware."
+      max_iterations: 3
+
+  - name: article-2
+    type: pge
+    task: "Write the second article from the content calendar."
+    inputs:
+      - "{artifact_dir}/content-calendar.md"
+    planner:
+      agent: marketer
+      operation: content
+    generator:
+      agent: copywriter
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/article-2.md"
+      guidance: "Must align with content calendar brief. Engaging, clear, and SEO-aware."
+      max_iterations: 3

package/examples/customer-feedback-loop.yaml

@@ -0,0 +1,44 @@
+name: customer-feedback-loop
+description: Synthesize customer feedback into product improvements.
+
+stages:
+  - name: feedback-synthesis
+    type: pge
+    task: "Synthesize recent customer feedback into themes and recommendations."
+    planner:
+      agent: customer-success
+      operation: feedback-synthesis
+    generator:
+      agent: customer-success
+      operation: feedback-synthesis
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/feedback-synthesis.md"
+      guidance: "Must identify top themes with frequency, severity, and representative quotes."
+      max_iterations: 2
+
+  - name: prioritization
+    type: pge
+    task: "Prioritize feedback themes into product backlog items."
+    inputs:
+      - "{artifact_dir}/feedback-synthesis.md"
+    planner:
+      agent: product-manager
+      operation: prioritization
+    generator:
+      agent: product-manager
+      operation: prioritization
+    evaluator:
+      agent: exec-reviewer
+    contract:
+      deliverable: "{artifact_dir}/prioritized-backlog.md"
+      guidance: "Must use impact/effort framework with clear rationale for priority order."
+      max_iterations: 2
+
+  - name: product-review
+    type: human_gate
+    prompt: "Review the prioritized backlog. Approve to update roadmap."
+    artifacts:
+      - "{artifact_dir}/feedback-synthesis.md"
+      - "{artifact_dir}/prioritized-backlog.md"

package/examples/design-sprint.yaml

@@ -0,0 +1,54 @@
+name: design-sprint
+description: Design sprint — UX research, design spec, review.
+
+stages:
+  - name: user-research
+    type: pge
+    task: "Synthesize user research findings for the design challenge."
+    planner:
+      agent: designer
+      operation: ux-research
+    generator:
+      agent: researcher
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/ux-research.md"
+      guidance: "Must include personas, user journey maps, and key pain points with evidence."
+      max_iterations: 2
+
+  - name: product-requirements
+    type: agent
+    task: "Extract product requirements from the UX research."
+    agent: product-manager
+    operation: spec-writing
+    inputs:
+      - "{artifact_dir}/ux-research.md"
+    output: "{artifact_dir}/requirements.md"
+
+  - name: design-spec
+    type: pge
+    task: "Write the design specification."
+    inputs:
+      - "{artifact_dir}/ux-research.md"
+      - "{artifact_dir}/requirements.md"
+    planner:
+      agent: designer
+      operation: design-spec
+    generator:
+      agent: designer
+      operation: design-spec
+    evaluator:
+      agent: designer
+      operation: design-review
+    contract:
+      deliverable: "{artifact_dir}/design-spec.md"
+      guidance: "Must include information architecture, interaction patterns, accessibility requirements, and edge cases."
+      max_iterations: 3
+
+  - name: design-approval
+    type: human_gate
+    prompt: "Review the design spec. Approve to proceed to implementation."
+    artifacts:
+      - "{artifact_dir}/design-spec.md"
+      - "{artifact_dir}/ux-research.md"

package/examples/feature-development.yaml

@@ -0,0 +1,96 @@
+name: feature-development
+description: Full feature development cycle — spec, design, implement, test, ship.
+
+stages:
+  - name: product-spec
+    type: pge
+    task: "Write a product requirements document based on the feature brief."
+    inputs:
+      - "{artifact_dir}/feature-brief.md"
+    planner:
+      agent: product-manager
+      operation: spec-writing
+    generator:
+      agent: product-manager
+      operation: spec-writing
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/prd.md"
+      guidance: "Must include user stories, acceptance criteria, scope boundaries, success metrics, and dependencies."
+      max_iterations: 3
+
+  - name: spec-approval
+    type: human_gate
+    prompt: "Review the PRD. Approve to proceed to technical design."
+    artifacts:
+      - "{artifact_dir}/prd.md"
+
+  - name: technical-design
+    type: pge
+    task: "Design the technical architecture for this feature."
+    inputs:
+      - "{artifact_dir}/prd.md"
+    planner:
+      agent: architect
+      operation: design
+    generator:
+      agent: architect
+      operation: design
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/design.md"
+      guidance: "Architecture must address all PRD requirements. Include component diagram, API contracts, and data flow."
+      max_iterations: 3
+
+  - name: design-approval
+    type: human_gate
+    prompt: "Review the technical design. Approve to proceed to implementation."
+    artifacts:
+      - "{artifact_dir}/design.md"
+
+  - name: implementation
+    type: pge
+    task: "Implement the feature according to the design."
+    inputs:
+      - "{artifact_dir}/prd.md"
+      - "{artifact_dir}/design.md"
+    planner:
+      agent: architect
+      operation: task-planning
+    generator:
+      agent: implementer
+    evaluator:
+      agent: code-reviewer
+    contract:
+      deliverable: "{artifact_dir}/implementation-report.md"
+      guidance: "All PRD acceptance criteria must be met. Tests required for new code paths."
+      max_iterations: 5
+    on_fail: human_gate
+
+  - name: test-suite
+    type: pge
+    task: "Write comprehensive tests for the implemented feature."
+    inputs:
+      - "{artifact_dir}/design.md"
+      - "{artifact_dir}/implementation-report.md"
+    planner:
+      agent: qa-engineer
+      operation: test-planning
+    generator:
+      agent: qa-engineer
+      operation: test-authoring
+    evaluator:
+      agent: code-reviewer
+    contract:
+      deliverable: "{artifact_dir}/test-report.md"
+      guidance: "Critical paths must have coverage. Edge cases and error conditions must be tested."
+      max_iterations: 3
+
+  - name: ship-approval
+    type: human_gate
+    prompt: "Implementation and tests complete. Approve to ship."
+    artifacts:
+      - "{artifact_dir}/implementation-report.md"
+      - "{artifact_dir}/test-report.md"

package/examples/growth-experiment.yaml

@@ -0,0 +1,49 @@
+name: growth-experiment
+description: Design and document a growth experiment.
+
+stages:
+  - name: funnel-analysis
+    type: agent
+    task: "Analyze the current conversion funnel and identify drop-off points."
+    agent: analyst
+    output: "{artifact_dir}/funnel-analysis.md"
+
+  - name: experiment-design
+    type: pge
+    task: "Design a growth experiment to improve the weakest funnel stage."
+    inputs:
+      - "{artifact_dir}/funnel-analysis.md"
+    planner:
+      agent: growth-strategist
+    generator:
+      agent: growth-strategist
+    evaluator:
+      agent: analyst
+    contract:
+      deliverable: "{artifact_dir}/experiment-design.md"
+      guidance: "Must include hypothesis, test/control design, sample size considerations, success metrics, and duration."
+      max_iterations: 3
+
+  - name: campaign-copy
+    type: pge
+    task: "Write the copy for the experiment's test variant."
+    inputs:
+      - "{artifact_dir}/experiment-design.md"
+    planner:
+      agent: marketer
+      operation: content
+    generator:
+      agent: copywriter
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/campaign-copy.md"
+      guidance: "Must include all ad copy variants, landing page copy, and email sequences as specified in experiment design."
+      max_iterations: 3
+
+  - name: experiment-approval
+    type: human_gate
+    prompt: "Review experiment design and campaign copy. Approve to launch."
+    artifacts:
+      - "{artifact_dir}/experiment-design.md"
+      - "{artifact_dir}/campaign-copy.md"

package/examples/incident-runbook.yaml

@@ -0,0 +1,43 @@
+name: incident-runbook
+description: Create an incident response runbook for a service.
+
+stages:
+  - name: service-assessment
+    type: agent
+    task: "Assess the service architecture and identify failure modes."
+    agent: architect
+    operation: health-assessment
+    output: "{artifact_dir}/service-assessment.md"
+
+  - name: runbook
+    type: pge
+    task: "Write the incident response runbook for this service."
+    inputs:
+      - "{artifact_dir}/service-assessment.md"
+    planner:
+      agent: ops-manager
+    generator:
+      agent: ops-manager
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/runbook.md"
+      guidance: "Must cover detection, triage, mitigation, and post-incident steps for each failure mode. Actionable by an on-call engineer."
+      max_iterations: 3
+
+  - name: devops-review
+    type: pge
+    task: "Review the runbook for technical accuracy and completeness."
+    inputs:
+      - "{artifact_dir}/runbook.md"
+      - "{artifact_dir}/service-assessment.md"
+    planner:
+      agent: devops
+    generator:
+      agent: devops
+    evaluator:
+      agent: reviewer
+    contract:
+      deliverable: "{artifact_dir}/runbook-reviewed.md"
+      guidance: "All commands must be correct. Monitoring links and escalation paths must be complete."
+      max_iterations: 2