按命令族和 profile 选择合适的 OpenClaw CLI 命令,并校验执行结果。
文档
EODHD CLI
试用Plan, build, review, or extend a stateless Bash CLI wrapper for the publicly documented EODHD API, intended for OpenClaw agents and Clawhub.ai distribution....
它能做什么
Plan, build, review, or extend a stateless Bash CLI wrapper for the publicly documented EODHD API, intended for OpenClaw agents and Clawhub.ai distribution....
技能文档
EODHD
Overview
Build the wrapper as a small, stateless Bash CLI that translates stable subcommands into EODHD HTTPS requests and emits machine-friendly output. Keep secrets outside the repo and outside the CLI's own persistence model by aligning auth to OpenClaw-managed environment injection.
Workflow
-
Confirm the task shape.
- If the user is asking for a plan, load references/implementation-plan.md.
- If the user is asking how to wire credentials safely, load references/openclaw-secrets.md.
- If the user is asking for actual code changes, implement the narrowest useful slice first and preserve stateless behavior.
-
Hold the contract steady.
- Prefer
eodhdas the binary name. - Default to JSON on stdout and diagnostics on stderr.
- Exit non-zero on HTTP, auth, or argument errors.
- Do not create config files, caches, token stores, or hidden state.
- Prefer
-
Keep auth OpenClaw-native.
- Expect
EODHD_API_KEYto be injected by OpenClaw. - Allow a one-shot explicit override like
--api-keyonly if the user asks for it or the implementation already exposes it. - Do not read plaintext secrets from repo files, sample configs, shell history helpers, or generated artifacts.
- Keep the publish metadata aligned with that contract by declaring
EODHD_API_KEYas the primary required credential.
- Expect
-
Enforce redaction and logging hygiene.
- Never print the raw API key to stdout or stderr.
- Redact secrets in debug output, dry-run output, request traces, and test snapshots.
- Avoid
set -xunless xtrace output is redirected through a sanitizer. - Treat query strings, headers, and copied curl commands as sensitive if they can contain
api_token.
-
Prefer a narrow, agent-friendly surface.
- Start with stable commands such as
eod,fundamentals,real-time,search,exchanges, andraw. - Add generic
--query key=valuesupport before adding many bespoke flags. - Use a
rawescape hatch for unsupported endpoints instead of blocking new EODHD API coverage.
- Start with stable commands such as
-
Preserve public-distribution constraints.
- Assume the repo is public.
- Use only public EODHD docs and public sample symbols in examples.
- Avoid local-machine assumptions, personal data, or environment-specific artifacts.
- Verify the release bundle contains
scripts/eodhdand the metadata that declares it.
Implementation Rules
- Keep the CLI dependency-light: Bash plus
curlas the baseline. - Use deterministic argument parsing and stable exit codes.
- Separate URL construction, request execution, error translation, and output handling into small shell functions.
- Make debug behavior opt-in with
--verboseor--dry-run. - If tests are added, prefer smoke tests around URL generation, argument validation, and redaction.
Deliverables
scripts/eodhdfor the CLI entrypoint when implementation startsscripts/check-package.shfor package and manifest consistency checksreferences/implementation-plan.mdfor milestones and acceptance criteriareferences/openclaw-secrets.mdfor secret and logging constraintsagents/openai.yamlfor Clawhub/OpenAI-facing metadata
Validation
- Check that examples never expose a real token.
- Check that debug or dry-run output replaces token values with a fixed mask such as
***REDACTED***. - Check that auth failure, HTTP failure, and usage failure produce distinct exit codes or clearly distinct stderr messages.
- Check that the manifest declares
EODHD_API_KEYand includesscripts/eodhdin the packaged files. - Check that the repo contains no
.env, token snapshots, or generated local artifacts.
相关技能
Standalone advanced network diagnostics for OpenClaw to continuously test end-to-end connectivity from OpenClaw agent to Telegram Bot API and approximate del...
Audit an OpenClaw agent workspace and generate standardized evaluation reports, scores, and patches. Use when asked to review memory quality, retrieval effic...
通过多轮引导式问答,生成完整的 OpenClaw 个人 AI 助手配置文件。
Monitor public OpenClaw ecosystem sources with source-quality checks, metadata-only collection, and trust-first reporting.
Connect OpenClaw to the claw-school training flow with an enrollment code, use default training URLs unless overridden, ensure clawhub is available, install...