fable-agent/SKILLS
artale 88d4f97397 complete Phase 15 Fusion: multi-model panel, OpenRouter Fusion, CineFable media, Fable 5 prompt engine
- FusionExecutor: draft -> critique -> fuse panel, PhaseExecutor-compatible
- OpenRouterFusionExecutor: direct OpenRouter Fusion API integration
- Fable 5 prompt engine: tone, citation, refusal patterns from leaked prompt
- Media generator: fal.ai image/video generation (nano-banana-pro, happy-horse)
- Fusion panel workflow in dynamic-workflows: dispatch -> judge -> verify
- Fusion CLI: \usion run\ (local panel or --openrouter), \usion panels\
- Media CLI: \generate image\, \generate video\
- Compound-stack: fusionPanel config flag and layer
- Model-router: OpenRouter Fusion route entry
- Build fixes: PhaseExecutor.refine signature, SafetyBoundary.route() args
- ESM fix: replaced require() with dynamic import() in panels command
2026-06-14 12:05:04 +02:00
..
agent-system-research fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
code-review fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
coding-best-practices fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
debugging-strategies fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
exa-research fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
fusion complete Phase 15 Fusion: multi-model panel, OpenRouter Fusion, CineFable media, Fable 5 prompt engine 2026-06-14 12:05:04 +02:00
ui-verification fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00
README.md fable-agent: self-improving agent system — 77 modules, 16 CLI commands, 43 tests 2026-06-13 09:50:45 +02:00

README.md

Skills

Procedural memory — lessons that compound across sessions. Every file here sharpens every run. Failure modes get added. Anti-patterns get documented. Versions get bumped.

Structure

Each skill lives in its own directory with a SKILL.md file:

SKILLS/<skill-name>/
├── SKILL.md       # The skill definition (required)
├── scripts/       # Helper scripts (optional)
├── references/    # Reference materials (optional)
└── assets/        # Images, examples (optional)

Skill Template

# Skill: <name>

> <one-line description>

---

## Interface
- Input contract
- Output contract
- API endpoints (if applicable)

## Steps
1. Step description with instructions
2. Expected outcome for each step

## Known Failure Modes (accumulated)
- Failure 1: symptom, root cause, fix
- Failure 2: symptom, root cause, fix

## Rules Discovered
- Rule 1
- Rule 2

## Anti-patterns Observed
- Anti-pattern 1
- Anti-pattern 2

## Version History
- v1 — initial (date): description
- v1.1 — (date): changes

Skill Sync

Skills defined here in markdown are synced to the TS skill registry:

npx tsx src/index.ts skills list          # List all registered skills
npx tsx src/index.ts skills inspect <id>  # View skill details
npx tsx src/index.ts skills evolve <id>   # Auto-improve skill
npx tsx src/index.ts skills sharpen       # Review all skills

To register a new skill from markdown, create the file here and run:

npx tsx src/index.ts skills create "<name>" --description "<desc>" --steps "<step1>,<step2>"

Current Skills

Skill Version Executions Status

(Run npx tsx src/index.ts skills list to populate)