/nd-search lets an AI agent search the web, read the clean text of a page by URL, crawl a few
pages of one site and search Telegram channels through the
NeuralDeep Search API (api.neuraldeep.ru/v1/search/*), with the same sk-
key as the NeuralDeep chat API.
The skill gives the agent a workflow (search, then read the picked links, crawl only a whole section)
and a small client, scripts/nd_search.py, that turns each call into one command with compact output
and meaningful exit codes.
Part of the rpa-skills collection of agent skills.
| File | Purpose |
|---|---|
SKILL.md |
Agent instructions (frontmatter name, description, metadata.version). |
scripts/nd_search.py |
The client: web, read, crawl, tg, quota. Python 3.8+ standard library only. |
tests/test_nd_search.py |
Unit tests against a local stub of the API: python -m unittest discover -s tests -v. |
export NEURALDEEP_SEARCH_API_KEY=sk-... # or use NEURALDEEP_API_KEY / --api-key
python3 scripts/nd_search.py web "vllm speculative decoding" -n 5
python3 scripts/nd_search.py read https://docs.vllm.ai/en/latest/features/spec_decode.html --max-chars 4000
python3 scripts/nd_search.py crawl https://docs.vllm.ai/en/latest/ -n 3
python3 scripts/nd_search.py tg "RAG in production" -n 5 --sort date_desc
python3 scripts/nd_search.py quota
python3 scripts/nd_search.py read https://example.com --json # the raw API answerThe key is taken from --api-key when given, then from NEURALDEEP_SEARCH_API_KEY, then from
NEURALDEEP_API_KEY. An empty search-key variable falls back to the general key. The client ignores
proxy variables from the environment and talks to the API directly.
Exit codes: 0 success, 2 bad arguments or no key, 3 request rejected (400, 404, 413, 415, 422),
4 quota used up (429), 5 access denied (401, 402, 403), 6 platform or network failure.
Every NeuralDeep plan except the API-only Coder includes search, the free one too, with a quota of
its own that does not touch the LLM limits. Endpoints, fields, errors and quota rules are described
in SKILL.md and in the NeuralDeep docs.
This skill is distributed through the rpa-skills catalog (a Claude Code plugin marketplace), and also installs as a plain skill folder (Cursor, OpenAI Codex, Kimi Code CLI, and others).
As a plugin (Claude Code): add the catalog once, then install this skill from it:
/plugin marketplace add EvilFreelancer/rpa-skills
/plugin install nd-search@rpa-skills
In Coddy: the rpa-skills catalog is built into Coddy as a system source,
so there is nothing to connect. One sync installs this skill together with the rest of the collection
into ~/.coddy/skills/nd-search/ and keeps it up to date (or press Sync all in Settings → Skills):
coddy skills syncAs a plain skill folder: copy or symlink this repository into a skill root (its SKILL.md lives at the repo root):
| Tool | Path |
|---|---|
| Coddy | ~/.coddy/skills/nd-search/ |
| Claude Code | ~/.claude/skills/nd-search/ |
| Cursor | ~/.cursor/skills/nd-search/ |
| OpenAI Codex | ~/.codex/skills/nd-search/ |
| Kimi Code CLI | ~/.kimi/skills/nd-search/ |
The directory name must match the name field in SKILL.md.
- Slash command: type
/nd-searchin agent chat. @context: attach the skill folder orSKILL.md.- Automatic: the agent may load it when your request matches the
descriptioninSKILL.md("найди в интернете", "прочитай страницу", "search the web", "read this URL").
Part of rpa-skills, Pavel Rykov's agent-skills collection (see notes on vibe coding and the prompt collection cursor-vibe-prompts).
Licensed under the MIT License, see LICENSE.