agentleFS
dasein108dasein108/ai-crypto-scoring/.claude/skills/cpd-crypto-analysis/SKILL.md

cpd-crypto-analysis

Skill0 starsChanged 5 months agoMITView on GitHub

Perform deep due diligence and risk scoring for any crypto project using the CPD framework (Chain, Protocol, Dapp), sometimes referred to as CDP in user requests, plus stage, age, code quality, and incident history. Use when the user asks to evaluate crypto project reliability, compare projects by security/risk posture, produce an evidence-backed CPD report, or calculate CPD percentage/tier from sourced findings.

  • In this repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run evidence-first CPD analysis.

    line 11
  2. Very specific:

    All paths below are project-root-relative.

    line 13
  3. Very specific:

    Run project-state freshness gate (mainnet/testnet/devnet).

    line 18
  4. Distinctive:

    If user does not specify version or interface, mark incomplete.

    line 32
  5. Distinctive:

    Verify current network state from at least:

    line 36
  6. Distinctive:

    Use WebFetch / WebSearch tools.

    line 40
  7. Distinctive:

    Do not score until filled.

    line 45
  8. Very specific:

    Read .claude/skills/cpd-crypto-analysis/references/source_playbook.md to map each criterion to canonical sources.

    line 67

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

CPD Crypto Analysis

Overview

Run evidence-first CPD analysis. Produce reproducible scorecard. Use bundled scripts to generate research prompt, scaffold report, compute scores deterministically.

All paths below are project-root-relative. Skill assets live at .claude/skills/cpd-crypto-analysis/.

Workflow

  1. Collect target metadata.
  2. Run project-state freshness gate (mainnet/testnet/devnet).
  3. Generate project-specific research brief.
  4. Gather verifiable evidence from primary or canonical sources.
  5. Score each CPD block from evidence only.
  6. Compute totals. Produce final recommendation.

Step 1: Collect Target Metadata

Capture before researching:

  • Project name and ticker
  • Chain and deployment environment (mainnet/testnet)
  • Exact protocol/version under review
  • Exact dapp/interface under review (official UI, fork, aggregator route)

If user does not specify version or interface, mark incomplete. State assumptions in report.

Step 2: Project-State Freshness Gate (Mandatory)

Verify current network state from at least:

  • one official source (project docs/blog/announcements),
  • one independent source (major media, explorer, analytics platform).

Use WebFetch / WebSearch tools. Capture in scorecard:

  • critical_facts.network_status.value — mainnet | testnet | devnet | unknown
  • critical_facts.network_status.as_of — YYYY-MM-DD
  • critical_facts.network_status.source_url

Do not score until filled. If sources disagree, use more recent. Flag risk.

Step 3: Generate Research Brief

Generate reusable deep-research prompt:

python3 .claude/skills/cpd-crypto-analysis/scripts/build_cpd_prompt.py \
  --project "Aave" > /tmp/aave_cpd_prompt.md

Stablecoin mode (adds ADP anti-depeg block):

python3 .claude/skills/cpd-crypto-analysis/scripts/build_cpd_prompt.py \
  --project "USDC" --include-adp > /tmp/usdc_cpd_prompt.md

Script loads template at .claude/skills/cpd-crypto-analysis/references/cpd_prompt_template.md.

Step 4: Collect Evidence

Read .claude/skills/cpd-crypto-analysis/references/source_playbook.md to map each criterion to canonical sources. Record per finding:

  • Source name
  • URL
  • Access/check date
  • One-line evidence note

Prioritize direct sources (official docs, explorers, audit reports, repos). Cross-check secondary sources against ≥1 primary.

Step 5: Fill the Scorecard

Initialize structured scorecard JSON:

python3 .claude/skills/cpd-crypto-analysis/scripts/init_cpd_scorecard.py \
  --project "Aave" --out /tmp/aave_scorecard.json

Stablecoin variant:

python3 .claude/skills/cpd-crypto-analysis/scripts/init_cpd_scorecard.py \
  --project "USDC" --out /tmp/usdc_scorecard.json --enable-adp

Edit JSON with evidence-backed values. Allowed ranges:

  • cpd.chain_score: 1..3
  • cpd.protocol_score: 0..2
  • cpd.dapp_score: -1..1
  • stage_score: -1..3
  • age_score: -1..3
  • code_score: -1..3
  • incidents_score: -1..3

Step 6: Compute Final Score and Tier

python3 .claude/skills/cpd-crypto-analysis/scripts/score_cpd.py \
  --scorecard /tmp/aave_scorecard.json \
  --format markdown \
  --out /tmp/aave_report.md

Scorer enforces critical-fact freshness. Default: network_status must be present and ≤45 days from analysis_date.

Output:

  • Block subtotals
  • Total out of 18
  • Percentage
  • Final tier classification
  • Optional ADP score when present

Reporting Standard

Per criterion include:

  1. Assigned score with concise rationale
  2. Evidence references (URL + checked date)
  3. Critical risk flags when applicable

Conclude with:

  • Final CPD percentage and tier
  • Top risk drivers
  • Recommendation: proceed / proceed with caution / avoid

Final report layout: .claude/skills/cpd-crypto-analysis/assets/cpd_report_template.md.

Resources

  • references/cpd_prompt_template.md — base CPD prompt template
  • references/source_playbook.md — source map and evidence checklist
  • assets/cpd_report_template.md — final report layout
  • scripts/build_cpd_prompt.py — prompt generator
  • scripts/init_cpd_scorecard.py — scorecard initializer
  • scripts/score_cpd.py — score calculator and report generator

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.