Skip to content

getting-started.md

Latest commit

 

History

History
72 lines (50 loc) · 2.47 KB

getting-started.md

File metadata and controls

72 lines (50 loc) · 2.47 KB
Raw

Getting Started

1. Clone and Validate

git clone https://github.com/RiaoMike/crypto-research-agent-skills.git
cd crypto-research-agent-skills
python scripts/validate.py skills

Python 3.10 or later is required. No third-party package is needed.

2. Preview an Installation

python scripts/install.py --target agents --dry-run

Choose one target:

Target Destination
codex ~/.codex/skills
claude ~/.claude/skills
agents ~/.agents/skills
custom The path supplied with --dest

Examples:

python scripts/install.py --target codex
python scripts/install.py --target claude
python scripts/install.py --target agents
python scripts/install.py --target custom --dest /path/to/harness/skills

Install only the orchestrator and literature skill:

python scripts/install.py --target agents \
  --skill crypto-research-pipeline \
  --skill crypto-literature-search

The installer validates every requested name before copying and refuses existing destinations. --force explicitly replaces selected existing skills.

3. Start a Fresh Session

Some clients discover skills only at startup. Start a fresh task or session after the first installation. Then ask the agent to use a named skill or give a request that clearly matches its frontmatter description.

Use crypto-idea-evaluator on this idea. Produce a claim map, feasibility risks,
kill criteria, and the next evidence-producing action. Do not claim novelty yet.

For the full workflow:

Use crypto-research-pipeline to manage this project from idea triage through
literature, bounded novelty assessment, theory/security review, parameters,
experiments, writing, and internal review. Preserve blockers and stop for human
approval before external submission.

4. Supply Evidence, Not Just a Topic

Better inputs include the exact idea, security model, intended theorem or construction, known papers, manuscript/repository paths, parameter sets, experiment environment, resource limits, disclosure constraints, and intended venue. Missing material should remain explicitly missing; the suite instructs agents not to invent it.

5. Interpret the Output

The pipeline uses pass, pass-with-limitations, fail, and blocked. A polished draft does not clear a proof or novelty blocker. “Ready for human submission review” means the documented internal gates are clear; it is not an acceptance prediction or authorization to submit.