@the-agenticflow/openflows 0.1.6 → 0.1.8-dev.230.5aa03a0

package/bin/orchestration/plugin/skills/forge-coding.md

@@ -1,161 +0,0 @@
-# FORGE Coding Skill
-
-## Your role
-
-You are FORGE, the generator in a FORGE-SENTINEL pair.
-Your job is to produce correct, complete, well-tested implementations.
-You work in segments. After each segment you submit to SENTINEL.
-Quality comes from the pair loop, not from you alone.
-
-## Before writing any code
-
-1. Read `TICKET.md` from `${SPRINTLESS_SHARED}/TICKET.md` - understand what you are building
-2. Read `CONTRACT.md` from `${SPRINTLESS_SHARED}/CONTRACT.md` - this is your definition of done
-3. Search the codebase - find existing patterns before inventing new ones
-   - Use Glob and Grep tools to find relevant files
-   - Look for similar functionality in existing code
-
-## Coding standards
-
-- All standards are in `orchestration/agent/standards/CODING.md` (if it exists)
-- All architecture patterns are in `orchestration/agent/arch/patterns.md` (if it exists)
-- API contracts are in `orchestration/agent/arch/api-contracts.md` (if it exists)
-- **READ these before implementing. They are not optional.**
-
-## Testing discipline
-
-- Every new function needs a test
-- Every changed file needs updated tests
-- Run tests after every segment: `orchestration/agent/tooling/run-tests.sh`
-- Do not submit a segment with failing tests
-
-## Error handling
-
-- Never throw raw Error - use the project's error type (e.g., `AppError` from `src/errors/`)
-- Every async function must have explicit error handling
-- Network calls must have timeout and retry logic
-
-## Submitting a segment
-
-When you believe a segment is complete:
-
-1. Run tests - all must pass
-2. Run linter - zero warnings
-3. Use `/segment-done` command to commit and notify SENTINEL
-4. Wait for `segment-N-eval.md` to appear in `${SPRINTLESS_SHARED}/`
-
-## Handling SENTINEL feedback
-
-If SENTINEL returns `CHANGES_REQUESTED`:
-
-- Read the `## Specific feedback` section carefully
-- Each item has: `file:line:problem:fix`
-- Fix **only** the specific items listed - do not refactor beyond what's requested
-- Re-run tests and linter
-- Re-submit with `/segment-done`
-
-## File locking
-
-Before writing to any file, the `pre_write_check.sh` hook validates ownership.
-
-- If you get `BLOCKED: File locked by pair-X`, you must:
-  1. Find an alternative implementation that avoids this file, OR
-  2. Set STATUS.json to `BLOCKED` with reason `FILE_LOCK_CONFLICT`
-
-## Context reset
-
-If you receive a "CONTEXT RESET REQUIRED" message:
-
-1. Run `/handoff` command immediately
-2. This writes `HANDOFF.md` with your current state
-3. Exit cleanly - a fresh FORGE will continue from your handoff
-
-## Valid STATUS.json Status Values
-
-When writing `STATUS.json`, you MUST use one of these exact status strings.
-Any other value will be treated as `BLOCKED` and your work will be wasted.
-
-| Status | When to use |
-|---|---|
-| `PR_OPENED` | Work complete and PR created (include `pr_url`, `pr_number`, `branch`) |
-| `COMPLETE` | All work done but PR creation deferred to harness |
-| `BLOCKED` | Cannot proceed (include `reason` and `blockers`) |
-| `FUEL_EXHAUSTED` | Budget/tokens exhausted |
-| `PENDING_REVIEW` | Work paused, waiting for review |
-| `AWAITING_SENTINEL_REVIEW` | Segment done, waiting for SENTINEL evaluation |
-| `APPROVED_READY` | Changes requested by SENTINEL have been addressed |
-| `SEGMENT_N_DONE` | Segment N complete (e.g. `SEGMENT_1_DONE`) |
-
-Do NOT invent status values. If you need review use `PENDING_REVIEW`. If you need help use `BLOCKED`.
-
-## When work is complete
-
-When SENTINEL approves all segments and you're ready to finish:
-
-1. **Push the branch to remote:**
-   ```bash
-   git push -u origin forge-${SPRINTLESS_PAIR_ID}/${SPRINTLESS_TICKET_ID}
-   ```
-   
-   NOTE: Direct `git push` is blocked. Instead, use the GitHub MCP tool:
-   - Get the current commit SHA
-   - Create a new branch reference on the remote
-
-2. **Create a Pull Request using GitHub MCP tool:**
-   - Use `create_pull_request` from the GitHub MCP server
-   - Set title: `[T-{id}] Brief description of the change`
-   - Set body: Use the PR description from `final-review.md`
-     - MUST include `Closes #<issue_number>` to auto-close the issue on merge
-     - Extract issue number from `SPRINTLESS_TICKET_ID`: `T-004` → `Closes #4`
-     - DO NOT use `Closes: T-004` (invalid - will not close the issue)
-   - Set head: `forge-${SPRINTLESS_PAIR_ID}/${SPRINTLESS_TICKET_ID}`
-   - Set base: `main`
-
-3. **Write STATUS.json with PR_OPENED:**
-   ```json
-   {
-     "status": "PR_OPENED",
-     "pair": "${SPRINTLESS_PAIR_ID}",
-     "ticket_id": "${SPRINTLESS_TICKET_ID}",
-     "branch": "forge-${SPRINTLESS_PAIR_ID}/${SPRINTLESS_TICKET_ID}",
-     "pr_url": "https://github.com/owner/repo/pull/42",
-     "pr_number": 42,
-     "files_changed": ["list", "of", "files"],
-     "segments_completed": N,
-     "timestamp": "2025-03-24T10:00:00Z"
-   }
-   ```
-
-4. **Exit** - The harness will detect STATUS.json and complete the lifecycle.
-
-## If you cannot create a PR
-
-If you encounter issues pushing or creating a PR:
-
-1. Write STATUS.json with `BLOCKED` status:
-   ```json
-   {
-     "status": "BLOCKED",
-     "pair": "${SPRINTLESS_PAIR_ID}",
-     "ticket_id": "${SPRINTLESS_TICKET_ID}",
-     "branch": "forge-${SPRINTLESS_PAIR_ID}/${SPRINTLESS_TICKET_ID}",
-     "reason": "Could not push/create PR: <specific error>",
-     "blockers": [],
-     "files_changed": ["list", "of", "files"]
-   }
-   ```
-
-2. Exit - NEXUS will be alerted for human intervention.
-
-## Branch naming
-
-Your branch is: `forge-${SPRINTLESS_PAIR_ID}/${SPRINTLESS_TICKET_ID}`
-
-Example: `forge-pair-1/T-42`
-
-## Environment variables
-
-- `SPRINTLESS_PAIR_ID` - your pair identifier (e.g., "pair-1")
-- `SPRINTLESS_TICKET_ID` - the ticket you're working on (e.g., "T-42")
-- `SPRINTLESS_WORKTREE` - your working directory
-- `SPRINTLESS_SHARED` - the shared directory for FORGE-SENTINEL communication

package/bin/orchestration/plugin/skills/forge-frontend-design.md

@@ -1,30 +0,0 @@
----
-name: frontend-design
-description: Create distinctive, production-grade frontend interfaces with high design quality. Focuses on premium aesthetics and avoiding "AI slop".
----
-
-# FORGE Frontend Design Skill
-
-Use this skill when tasked with building web components, pages, artifacts, or applications. Your goal is to create distinctive, production-grade frontend interfaces that avoid generic "AI slop" aesthetics.
-
-## Design Principles
-
-- **Commit to BOLD Aesthetics**: Pick a clear conceptual direction (e.g., minimal, maximalist, retro-futuristic, etc.) and execute it with precision.
-- **Memorable & Polished**: What is the one thing someone will remember? Ensure meticulous refinement in every detail.
-- **Technical Excellence**: Framework compatibility, performance, and accessibility are baseline requirements.
-
-## Guidelines
-
-- **Typography**: Pair a distinctive display font with a refined body font. Avoid Inter, Roboto, and Arial.
-- **Color & Theme**: Use CSS variables for a cohesive theme. Choose accents that pop.
-- **Motion**: Orchestrate high-impact animations. Use CSS transitions or library-specific motion (e.g., Framer Motion for React).
-- **Composition**: Break the grid. Use asymmetry, negative space, and overlapping elements for depth.
-- **Background Content**: Use meshes, noise, geometric patterns, or layered transparencies to create atmosphere.
-
-## What to Avoid
-
-- Purple gradients on white backgrounds.
--系統 default font stacks.
-- Predictive layouts and cookie-cutter component patterns.
-
-MATCH implementation complexity to the aesthetic vision. Elegance comes from executing the vision well.

package/bin/orchestration/plugin/skills/forge-mcp-builder.md

@@ -1,37 +0,0 @@
----
-name: mcp-builder
-description: Comprehensive guide for developing and extending Model Context Protocol (MCP) servers.
----
-
-# FORGE MCP Builder Skill
-
-Use this skill when you need to build or extend MCP servers to provide new capabilities to the AgentFlow system.
-
-## High-Level Workflow
-
-### Phase 1: Research and Planning
-
-- Understand the external system or data source you are integrating.
-- Define the necessary tools and resources.
-- Choose the appropriate language/SDK (Rust, Python, Node.js).
-
-### Phase 2: Implementation
-
-- Implement the core protocol (Tools, Resources, Prompts).
-- Follow the official [MCP Specification](https://modelcontextprotocol.io/).
-- Ensure robust error handling and type safety.
-
-### Phase 3: Review and Test
-
-- Verify transport layers (Stdio, HTTP/SSE).
-- Test tool execution and resource retrieval.
-- Check for protocol compliance.
-
-### Phase 4: Integration
-
-- Create evaluations and documentation.
-- Add components to the `.claude-plugin/plugin.json` or `.mcp.json`.
-
-## Documentation Reference
-
-Consult the official SDK documentation for the language you choose. Prioritize security and minimal permission sets when exposing system capabilities via tools.

package/bin/orchestration/plugin/skills/forge-planning.md

@@ -1,102 +0,0 @@
-# FORGE Planning Skill
-
-## Writing PLAN.md
-
-Before any implementation, write a plan.
-Use the `/plan` command to structure it correctly.
-
-## What a good plan contains
-
-1. **Your understanding of the ticket** - in your own words
-2. **Technical approach** - follows `orchestration/agent/arch/patterns.md` (if it exists)
-3. **Explicit segment breakdown** - each segment is independently testable
-4. **Definition of done per segment** - specific and verifiable
-5. **List of files you will create or modify**
-6. **Risk areas** - things you are uncertain about
-7. **Questions for SENTINEL** - clarifications needed before starting
-
-## Segment sizing
-
-A good segment:
-- Touches 1-3 files
-- Has a single clear purpose
-- Can be tested in isolation
-- Takes roughly 20-40 minutes to implement
-
-A segment that is too large:
-- Touches more than 5 files
-- Has multiple unrelated concerns
-- Cannot be independently verified
-
-**Split it.**
-
-## Example PLAN.md structure
-
-```markdown
-# PLAN: [Ticket Title]
-
-## Understanding
-
-[Brief summary of what we're building and why]
-
-## Technical Approach
-
-[How we'll implement it, referencing existing patterns]
-
-## Segments
-
-### Segment 1: [Name]
-
-**Purpose:** [Single sentence]
-
-**Files:**
-- `src/path/file1.ts` (new)
-- `src/path/file2.ts` (modify)
-
-**Definition of Done:**
-- [ ] [Specific criterion 1]
-- [ ] [Specific criterion 2]
-- [ ] Tests pass
-
-### Segment 2: [Name]
-...
-
-## Files Changed
-
-- `src/auth/login.ts` (new)
-- `src/middleware/auth.ts` (modify)
-- `tests/auth/login.test.ts` (new)
-
-## Out of Scope
-
-- [Explicitly list what we're NOT building]
-- [This prevents scope creep]
-
-## Risks
-
-- [Risk 1]: [Mitigation strategy]
-- [Risk 2]: [Mitigation strategy]
-
-## Questions for SENTINEL
-
-- [Question 1]
-- [Question 2]
-```
-
-## Contract negotiation
-
-SENTINEL will review your plan.
-
-If SENTINEL objects:
-1. Read the objection carefully in `CONTRACT.md`
-2. Update `PLAN.md` addressing each specific objection
-3. Do not argue - either accept the feedback or ask a clarifying question
-
-Maximum 3 rounds of negotiation.
-After 3 rounds without agreement, the ticket is BLOCKED.
-
-## After CONTRACT is AGREED
-
-1. Read `CONTRACT.md` - these are your binding terms
-2. Begin implementation with Segment 1
-3. Do not deviate from the plan without updating `PLAN.md` and getting SENTINEL approval

package/bin/orchestration/plugin/skills/forge-skill-creator.md

@@ -1,25 +0,0 @@
----
-name: skill-creator
-description: Meta-skill for creating, evaluating, and improving AgentFlow skills.
----
-
-# FORGE Skill Creator Skill
-
-Use this skill when you need to formalize a new set of instructions or common workflows into an AgentFlow `SKILL.md`.
-
-## Workflow
-
-1. **Capture Intent**: What problem does this skill solve?
-2. **Research**: Identify best practices and common pitfalls for the targeted task.
-3. **Draft SKILL.md**:
-   - Use the frontmatter format: `name`, `description`.
-   - Include a clear `# Title`.
-   - Define a `# Workflow` or `# Guidelines`.
-4. **Iterative Improvement**: Use feedback from tests or user reviews to refine the skill instructions.
-
-## Skill Writing Tips
-
-- **Be Actionable**: Use imperative verbs.
-- **Be Structured**: Use headers and lists for scannability.
-- **Be Contextual**: Mention specific AgentFlow components if relevant.
-- **Be Minimal**: Only include instructions that provide high value.

package/bin/orchestration/plugin/skills/forge-web-artifacts-builder.md

@@ -1,29 +0,0 @@
----
-name: web-artifacts-builder
-description: Guide for creating elaborate, multi-component React artifacts using Tailwind CSS and shadcn/ui.
----
-
-# FORGE Web Artifacts Builder Skill
-
-Use this skill when building complex, stateful web artifacts that require more than a single-file approach.
-
-## Stack Requirements
-
-- **Framework**: React 18 / TypeScript.
-- **Styling**: Tailwind CSS + shadcn/ui.
-- **Components**: Utilize Radix UI primitives as provided by shadcn/ui.
-
-## Workflow
-
-1. **Plan Architecture**: Define state management and component hierarchy.
-2. **Implementation**: Build interactive components with clear prop interfaces.
-3. **Bundling**: Ensure all assets are optimized for presentation.
-4. **Testing**: Verify interactive flows (buttons, forms, state transitions).
-
-## Design Excellence
-
-Avoid "AI slop" by:
-
-- Using custom typography and color themes.
-- Implementing sophisticated layout patterns (not just centered stacks).
-- Adding subtle micro-animations and transitions.

package/bin/orchestration/plugin/skills/lore-brand-guidelines.md

@@ -1,33 +0,0 @@
----
-name: brand-guidelines
-description: Maintains project-specific visual and tonal identity for AgentFlow documentation.
----
-
-# AgentFlow Brand & Style Guidelines
-
-Use this skill to ensure all documentation and UI copy adheres to the AgentFlow project identity.
-
-## Visual Identity
-
-### Colors
-
-- **Primary**: Agentic Blue (#0047AB) - Represents focus and intelligence.
-- **Secondary**: Flow Silver (#C0C0C0) - Represents agility and connection.
-- **Accent**: Gravity Dark (#1A1A1A) - Represents stability and core logic.
-
-### Typography
-
-- **Headings**: Use strong, geometric sans-serif fonts (e.g., Montserat, Inter).
-- **Body**: Use clean, highly readable serif or sans-serif fonts (e.g., Lora, Open Sans).
-
-## Tonal Guidelines
-
-- **Professional & Precise**: Avoid jargon where possible, but maintain technical accuracy.
-- **Encouraging & Agentic**: Empower the user to interact with the system.
-- **Concise**: Value the user's time.
-
-## Formatting Standards
-
-- Use GitHub-style alerts (Note, Tip, Important, Warning, Caution) for emphasis.
-- Use consistent header hierarchies.
-- Ensure all code snippets are syntax-highlighted.

package/bin/orchestration/plugin/skills/lore-changelog.md

@@ -1,69 +0,0 @@
----
-name: changelog
-description: Skill for maintaining CHANGELOG.md
----
-
-# Changelog Skill
-
-## CHANGELOG Format
-
-Follow [Keep a Changelog](https://keepachangelog.com/):
-
-```markdown
-# Changelog
-
-## [Unreleased]
-
-### Added
-- New feature X
-
-### Changed
-- Modified behavior Y
-
-### Fixed
-- Bug fix Z
-
-## [1.2.0] - 2024-01-15
-
-### Added
-- Feature A
-```
-
-## Categories
-
-| Category | Use For |
-|----------|---------|
-| `Added` | New features |
-| `Changed` | Modified behavior |
-| `Deprecated` | Soon-to-be removed |
-| `Removed` | Removed features |
-| `Fixed` | Bug fixes |
-| `Security` | Security fixes |
-
-## Entry Style
-
-### Good Entry
-```
-- Add OAuth2 authentication support (#42)
-```
-
-### Bad Entry
-```
-- Fixed stuff
-- Updated code
-```
-
-## Process
-
-1. Read PR description from SENTINEL's final-review.md
-2. Categorize changes
-3. Write concise, user-focused entry
-4. Place under `[Unreleased]` section
-5. Commit with message: `docs: update CHANGELOG for T-{id}`
-
-## Release
-
-When version is released:
-1. Rename `[Unreleased]` to `[version] - date`
-2. Add new `[Unreleased]` section
-3. Tag release in git

package/bin/orchestration/plugin/skills/lore-doc-coauthoring.md

@@ -1,33 +0,0 @@
----
-name: doc-coauthoring
-description: Advanced workflow for iterative documentation refinement and reader testing.
----
-
-# LORE Doc Co-Authoring Skill
-
-Use this skill when tasked with creating or refining complex documentation. This workflow ensures that the final output is contextually accurate and reader-friendly.
-
-## Workflow Stages
-
-### Stage 1: Context Gathering
-
-- **Identify Goals**: What is the purpose of the document? Who is the audience?
-- **Gather Info**: Interview the user or research the codebase to capture all necessary details.
-
-### Stage 2: Refinement & Structure
-
-- **Iterative Drafting**: Start with a structure and refine it based on feedback.
-- **Gap Check**: Identify missing information and resolve contradictions.
-- **Drafting**: Write clear, professional content.
-
-### Stage 3: Reader Testing
-
-- **Predict Reader Questions**: What will the reader be confused about?
-- **Sub-Agent Testing**: If possible, ask a sub-agent to "read" the doc and provide feedback.
-- **Iterate**: Fix issues identified during testing.
-
-## Quality Standards
-
-- Clarity, conciseness, and accuracy are paramount.
-- Maintain a consistent tone throughout the document.
-- Use visual hierarchy (headers, lists, alerts) to improve readability.

package/bin/orchestration/plugin/skills/lore-documentation.md

@@ -1,57 +0,0 @@
----
-name: documentation
-description: Skill for maintaining project documentation
----
-
-# Documentation Skill
-
-## Documentation Types
-
-### User Documentation
-- `README.md`: Project overview, setup, usage
-- `docs/usage/`: How-to guides
-- `docs/api/`: API reference
-
-### Developer Documentation
-- `CONTRIBUTING.md`: Contribution guidelines
-- `docs/architecture/`: System design
-- `.agent/standards/`: Coding standards
-
-### Changelog
-- `CHANGELOG.md`: User-facing changes
-
-## When to Update
-
-Update documentation when:
-- New feature added
-- Breaking change introduced
-- API modified
-- Configuration changed
-- Dependencies updated
-
-## Documentation Style
-
-### README Updates
-- Keep concise
-- Update examples
-- Reflect current API
-
-### API Documentation
-- Document all public interfaces
-- Include examples
-- Document error conditions
-
-### Architecture Docs
-- Update diagrams if structure changed
-- Document new patterns
-- Update decision records
-
-## Commit Format
-
-```
-docs: update README for new auth feature
-
-- Add OAuth2 setup instructions
-- Update configuration example
-- Add troubleshooting section
-```

package/bin/orchestration/plugin/skills/lore-docx.md

@@ -1,20 +0,0 @@
----
-name: docx
-description: Advanced document generation, analysis, and refinement for team reports and external docs.
----
-
-# LORE Advanced Document Skill
-
-Use this skill when processing Word documents (.docx) or generating formal reports in document formats.
-
-## Capabilities
-
-- **Structure Analysis**: Verify document hierarchy and style consistency.
-- **New Doc Generation**: Create professional reports from markdown context.
-- **XML Refinement**: Perform low-level XML edits for tracking changes or complex formatting (Table of Contents, Headers).
-
-## Critical Rules
-
-- **No Unicode Bullets**: Always use standard list formatting.
-- **Style Overrides**: Ensure headers match the Brand Style Guidelines.
-- **Table Formatting**: Use consistent borders and cell padding for readability.

package/bin/orchestration/plugin/skills/lore-pdf.md

@@ -1,20 +0,0 @@
----
-name: pdf
-description: Advanced PDF processing, including creation, analysis, text extraction, and watermarking.
----
-
-# LORE Advanced PDF Skill
-
-Use this skill to process incoming PDF reference material or to generate formal, secure PDF outputs.
-
-## Processing Tasks
-
-- **Extraction**: Extract text and tables from complex PDF layouts using `pdfplumber`.
-- **Generation**: Create formal reports with consistent branding using `reportlab`.
-- **Security**: Apply watermarking or password protection to sensitive documents.
-
-## Quality Standards
-
-- Ensure all generated PDFs are searchable (OCR/embedded text).
-- Maintain high-resolution assets for images and charts within the PDF.
-- Follow the visual identity from `Brand & Style Guidelines`.

package/bin/orchestration/plugin/skills/lore-pptx.md

@@ -1,23 +0,0 @@
----
-name: pptx
-description: Professional presentation design and generation using a theme-first approach.
----
-
-# LORE Presentation Design Skill
-
-Use this skill to create high-impact, professional slide decks (.pptx) from project data or documentation.
-
-## Design Workflow
-
-1. **Choose a Theme**: Select an established goal (see `theme-factory`) or define a new one.
-2. **Slide Architecture**:
-   - One primary idea per slide.
-   - Use high-quality typography and consistent color palettes.
-   - Maintain a 16:9 aspect ratio.
-3. **Execution**: Use `python-pptx` or equivalent libraries to generate the deck.
-
-## Quality Assurance
-
-- **Visual QA**: Check for text overflows, color contrast, and font consistency.
-- **Content QA**: Ensure the message flows logically across the deck.
-- **Brand Alignment**: Follow the `Brand & Style Guidelines` strictly.

package/bin/orchestration/plugin/skills/lore-theme-factory.md

@@ -1,20 +0,0 @@
----
-name: theme-factory
-description: Professional visual styling and theming for AgentFlow documentation and reports.
----
-
-# LORE Theme & Styling Skill
-
-Use this skill to apply consistent, professional themes to all documentation, changelogs, and generated reports.
-
-## Themes & Atmosphere
-
-- **The "Agentic" Theme**: Blue/Silver/Dark Grey. High-tech, professional.
-- **The "Focus" Theme**: Monochrome/High-Contrast. Minimalist, objective.
-- **The "Warning" Theme**: Amber/Black. Used for critical alerts or post-mortem reports.
-
-## Styling Rules
-
-- Apply font pairings consistently (Display for headers, Serif/Sans for body).
-- Use color accents for data visualization and call-to-action blocks.
-- Ensure all styling enhances readability and doesn't distract from the content.

package/bin/orchestration/plugin/skills/lore-xlsx.md

@@ -1,20 +0,0 @@
----
-name: xlsx
-description: Advanced Excel document creation, editing, and technical data analysis.
----
-
-# LORE Advanced Excel Skill
-
-Use this skill for generating complex reports that involve budgeting, technical metrics, or financial modeling.
-
-## Core Requirements
-
-- **Use Formulas**: NEVER hardcode values that can be calculated.
-- **Data Integrity**: Use `pandas` for large dataset analysis before writing to Excel.
-- **Formatting**: Apply professional formatting (headers, borders, number formats) for all outputs.
-
-## Analysis Patterns
-
-- **Summary Sheets**: Create high-level dashboards at the beginning of the workbook.
-- **Validation**: Include formula testing strategies to ensure zero erroneous calculations.
-- **Documentation**: Comment complex formulas or data sources within the sheet.