agentskills-specification
Agent Skills — open format (reference)
Educational routing only. This skill does not reproduce the specification. Use the canonical links below for authoritative text, schema details, and license terms.
Canonical URLs
| Resource | URL |
|---|---|
| Repository | github.com/agentskills/agentskills |
| Site | agentskills.io — documentation hub (per upstream README) |
The upstream project describes Agent Skills as a simple, open format: folders of instructions, scripts, and resources that agents can discover for task-specific expertise (“write once, use everywhere”). The repo holds the specification, documentation, reference SDK material, and pointers to example skills and community channels (e.g. Discord — see live README).
Documentation index (llms.txt)
Use agentskills.io/llms.txt as the machine-readable index of first-party docs (discover pages before deep-linking). It currently lists:
| Doc | URL |
|---|---|
| Overview | agentskills.io/home.md — open format for agent capabilities and expertise |
| What are skills? | agentskills.io/what-are-skills.md — lightweight format for specialized knowledge and workflows |
| Specification | agentskills.io/specification.md — complete format specification |
| Integrate skills into your agent | agentskills.io/integrate-skills.md — adding Agent Skills support to an agent or tool |
What to read there
| Area | Typical use |
|---|---|
| Specification | specification.md — exact format expectations for portable skills |
| Integration | integrate-skills.md — agent/tool implementers |
| Documentation | home.md, what-are-skills.md — concepts |
| Example skills | Patterns and breadth (see GitHub README + examples) |
| Contributing | CONTRIBUTING.md in-repo |
License (summary)
Upstream states: code in the repository is under Apache-2.0; documentation under CC-BY-4.0 — confirm in-repo LICENSE and per-directory notices. The format is community-open; Anthropic is named as a maintainer in the public README.
How to combine with blockint
| Need | Skill |
|---|---|
| This repo’s BI / chain-focused skills | blockchain-intelligence-playbook |
| Portable skill format vs domain content | agentskills-specification (upstream spec); topical skills here for blockchain intel |
Guardrails
- Upstream wins — If blockint layout ever diverges from the published spec, treat agentskills as the portability reference.
- No credential paste — Spec work does not require embedding secrets in skills.
Goal: stable pointers to agentskills/agentskills, agentskills.io/llms.txt, and agentskills.io/specification.md for Agent Skills format and documentation alignment.
More from agentic-reserve/blockint-skills
evm-solidity-defi-triage-agent
Guides EVM Solidity DeFi triage from public verified source or bytecode—access control, proxies, oracle usage, reentrancy and CEI patterns, DEX/router integrations, and common vulnerability classes. Use when the user asks for Ethereum or L2 smart contract security review, Solidity audit triage, OpenZeppelin proxy risks, or EVM-specific DeFi patterns—not for live exploits or private keys.
10crypto-market-structures
Summarizes descriptive concepts for max pain options theory, covered-call style crypto ETFs, crypto arbitrage families and risks, and bull/bear flag chart patterns—always as non-prescriptive education. Use when the user asks about max pain, premium income ETFs, arbitrage, funding rates, flash loans, or bull/bear flags in crypto trading context.
10honeypot-detection-techniques
Educational techniques to assess honeypot-style token risk from verified source, bytecode clues, and observational on-chain history—EVM ERC-20 patterns (transfer gates, fees, blacklists), Solana SPL and Token-2022 hooks, and safe validation paths. Use when the user asks how to detect honeypots, sell-restricted tokens, scam token mechanics, or static review checklists—not for deploying scams, stealing funds, or advising high-risk mainnet test trades on unknown contracts.
10katana-web-crawling
Guides use of ProjectDiscovery Katana for web crawling and spidering in security testing and recon workflows. Covers installation, standard vs headless mode, scope and rate limits, JSONL output, and piping from httpx or URL lists. Use when the user mentions Katana, projectdiscovery/katana, web crawling, spidering, endpoint discovery, attack surface mapping, or chaining crawlers in automation pipelines.
10solana-defi-vulnerability-analyst-agent
Guides discovery and documentation of Solana DeFi protocol risks from public code and chain state—Anchor/native programs, PDAs, CPIs, oracles, pools, SPL mechanics, and historical tx reconstruction. Use when the user asks for Solana program security review, DeFi vulnerability triage, PDA or CPI safety, oracle or liquidity-pool risk, launchpad/bonding-curve issues, or evidence-backed severity findings without exploits or private keys.
10solana-tracing-specialist
Guides Solana-specific on-chain forensics—ATA resolution, SPL instruction parsing, transaction history via RPC and indexers (e.g. Helius-style APIs), fund-flow graphs, Solana clustering heuristics, and program authority review. Use when the user investigates Solana wallets, SPL tokens, DEX/Jito flows, rug or phishing patterns on Solana, or needs evidence-structured tracing reports with public data only.
10