Bifrost Slpx skills

Bifrost SLPx liquid staking via @bifrostio/slpx-cli: exchange rates, APY (with optional DeFiLlama LP pools), TVL, holders, protocol info; vETH balances, rede...

Security Scan

The skill’s requests and instructions match its stated purpose (wrapping the @bifrostio/slpx-cli) and its handling of private keys and on‑chain flows is consistent with that goal.

Purpose & Capability

Name/description match the runtime instructions: the SKILL.md consistently instructs the agent to call the @bifrostio/slpx-cli (via npx) for queries (rate/apy/info) and vETH on‑chain flows (balance/mint/redeem/claim). The tokens, chains, and CLI examples align with the claimed coverage.

Instruction Scope

The instructions are narrowly scoped to running the CLI with --json and interpreting JSON output, with an explicit pre‑tx checklist and inversion prompts before any broadcast. The skill also includes guidance for configuring a private key (BIFROST_SKILL_PRIVATEKEY) and warning not to paste raw keys into chat; that guidance is reasonable but slightly ambiguous about whether the agent should run shell detection commands itself or instruct the user to run them. Ensure the agent does not attempt to collect raw keys or run arbitrary commands in the user's environment without explicit consent.

Install Mechanism

The skill is instruction-only (no install spec), but it repeatedly instructs agents to run npx -y @bifrostio/slpx-cli. Using npx means code is fetched and executed from npm at runtime — expected for a CLI wrapper but higher risk than pure local tooling. The SKILL.md does advise pinning a dist‑tag if needed; verify package provenance and consider pinning the package/version before running.

Credentials

No required env vars are declared by the skill; the only sensitive env referenced is BIFROST_SKILL_PRIVATEKEY for signing on‑chain transactions. That single credential is proportional to the claimed on‑chain write capabilities. The skill explicitly warns not to paste keys into chat and includes a pre‑tx checklist forbidding private key leakage.

Persistence & Privilege

The skill does not request always:true and does not ask to modify other skills or global agent settings. Autonomous invocation is allowed (platform default) but is not combined with broad unexplained privileges or credential requests.

Assessment

This skill is a coherent wrapper around the @bifrostio/slpx-cli and behaves as expected for a liquid‑staking CLI. Before installing or letting an agent run it: (1) confirm the npm package is the official @bifrostio/slpx-cli (check the package page and source repo), (2) prefer pinning an explicit version or dist‑tag rather than always using unpinned npx -y to reduce supply‑chain risk, (3) run npx commands yourself in a controlled terminal when possible instead of allowing the agent to execute them remotely, (4) never paste private keys into chat — follow the skill’s private‑key guidance and only set BIFROST_SKILL_PRIVATEKEY locally in your shell, and (5) only approve broadcasting transactions after the pre‑tx checklist is cleared and you explicitly confirm the action. If you want extra assurance, review the upstream CLI code/repo and the package release integrity before use.

Bifrost SLPx CLI

You operate the Bifrost liquid-staking CLI. On-chain execution is handled inside the tool; you run commands and interpret JSON output.

Grounding — no extra narrative

• Answer only from fields present in the CLI --json you just received. Do not add filler like time periods ("since inception", "YTD", "over time"), historic performance, or why the numbers moved unless that text is literally in the payload.
• rate (and the paired amounts in rate / info) is a spot conversion ratio at query time—not total return, not APR, and not evidence of a particular timeline. If (rate − 1) as a percent is mentioned, frame it strictly as “per 1 outputToken, you get this multiple of inputToken right now”, not as appreciation since launch.
• If the user wants interpretation beyond what JSON provides, say the CLI does not expose that dimension; offer another --json run for updated numbers, not invented context.

Progressive disclosure — what to load

Task	Read first
rate / apy / info only	references/commands.md (query sections)
balance / status (read-only on-chain)	references/commands.md + references/tokens-and-chains.md if chain/token unclear
mint, redeem, or claim	references/pre-tx-checklist.md, then references/commands.md
signing key unset / how to configure it	references/private-key-env.md

Do not skip pre-tx-checklist.md before any transaction that could broadcast.

Pre-flight

npx -y @bifrostio/slpx-cli --version

If the published package uses a non-latest dist-tag (e.g. prerelease on next), pin it: npx -y @bifrostio/slpx-cli@next --version.

On-chain pipeline (do not skip steps)

For mint, redeem, and claim:

1. Complete the inversion prompts below if anything is ambiguous.
2. Read references/pre-tx-checklist.md.
3. Run through the checklist with the user and report severity-grouped findings.
4. Stop until the user explicitly approves broadcasting.
5. Only then run the signing path per CLI docs with --json (assume the CLI signing key is already configured).

For redeem, always state that settlement is not instant (queued, often 1–3 days) before any real execution.

Inversion — gather before acting

Before mint/redeem/claim, resolve (ask the user if missing):

• Chain (ethereum | base | optimism | arbitrum for vETH on-chain).
• Amount and whether mint uses native ETH or --weth.
• Address / wallet context (you must not ask for raw key material):
  • mint / redeem / claim: use the CLI broadcast path with the configured signing wallet.
  • balance / status: pass an address argument, or omit it to query the default signing wallet derived by the CLI. Batch balance still needs explicit comma-separated addresses.

Do not broadcast until the above are clear and the pipeline steps are satisfied.

Signing key not configured

Read references/private-key-env.md and follow it end-to-end.

Commands, options, and examples

See references/commands.md.

Tokens, chains, contracts

See references/tokens-and-chains.md.

Errors

CLI returns JSON with error, code, message. Full table: references/errors.md.

Notes

• Always prefer --json for agent-driven calls.
• Substrate-only tokens: use query commands; there is no CLI on-chain path for them in this tool.