UniFuncs Deep Search
ReviewAudited by ClawScan on May 10, 2026.
Overview
This mostly looks like a UniFuncs Deep Search API wrapper, but review is warranted because it uses a credentialed external API and includes an undocumented hidden background-worker mode.
Install only if you trust UniFuncs with your search queries and are comfortable providing an API key. Avoid sensitive queries, do not publish or share results unless explicitly intended, clean up stream files when needed, and ask for clarification on the hidden background-worker behavior before relying on this in sensitive environments.
Findings (5)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A Deep Search run could potentially continue API work or file writing in a way the user did not see documented.
The report client defines a background-worker mode hidden from help and absent from SKILL.md's documented options. In a network client that uses an API key, an undocumented background mode makes runtime lifecycle, stopping behavior, and cleanup unclear.
import subprocess ... parser.add_argument("--background-worker", action="store_true", help=argparse.SUPPRESS)Ask the maintainer to document the background-worker behavior, its timeout, how it is started and stopped, and where it writes data; prefer explicit user opt-in for any background execution.
The agent can run Python commands for this skill, and results could be shared publicly if those flags are selected.
The skill needs Python execution for its included clients, but the tool permission is broadly expressed as python3 via Bash, and optional flags can publish or share results if used.
allowed-tools: Bash(python3:*) ... --push-to-share Push result to share space. ... --set-public Set result as public.
Use only the included scripts for this skill and do not use --push-to-share or --set-public unless the user explicitly asks to publish or share the result.
Anyone running the skill must provide an API key that may have billing or account implications.
The script reads a UniFuncs API key from the environment and sends it as a bearer credential to the UniFuncs API. This is expected for the service, but it is a credentialed integration.
api_key = os.environ.get("UNIFUNCS_API_KEY") ... "Authorization": f"Bearer {api_key}"Use a revocable, least-privilege UniFuncs key if available, avoid exposing it in logs or shell history, and revoke it if the environment is compromised.
Search topics, prompts, and any sensitive details included in the query may be sent to UniFuncs.
The user's query is transmitted to an external UniFuncs API endpoint for processing, which is central to the skill's purpose but affects data boundaries.
CHAT_COMPLETIONS_ENDPOINT = "https://api.unifuncs.com/deepsearch/v1/chat/completions" ... "messages": [{"role": "user", "content": args.query}]Do not include secrets, private documents, or confidential details in queries unless you trust UniFuncs' handling of that data.
Deep-search output may remain on disk after the run, including any sensitive information returned in the report.
The skill can persist streamed report chunks to a local file and later re-read them. This is useful for long searches but can leave local copies of query results.
--stream-file STREAM_FILE
Path to persist/read stream chunks. If omitted,
temp file is auto-created when writable.Use a known safe stream-file location when needed and delete stream files after use if the report contains sensitive information.