MCP Servers · can4hou6joeng4

boss-agent-cli

boss-agent-cli is a Python CLI tool for AI agents to search and filter job listings on Chinese job boards (BOSS Zhipin, Zhilian, 51job), with local candidate tracking, welfare filtering, and JSON output. It enforces compliance by default, blocking sensitive operations like bulk messaging and candidate data scraping.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/can4hou6joeng4/boss-agent-cli

1.3k

GitHub stars

106

Forks

Python

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	can4hou6joeng4/boss-agent-cli
Owner	can4hou6joeng4
Primary language	Python
License	MIT — OSI-approved
Stars	1.3k
Forks	106
Open issues	2
Latest release	v1.14.0 (2026-06-25)
Last updated	2026-06-26
Source	https://github.com/can4hou6joeng4/boss-agent-cli

What boss-agent-cli is

Python 3.10+ CLI (Click-based) that wraps job board APIs with schema-driven command exposure, SQLite-backed local caching, optional OpenAI-compatible AI analysis (Ollama/vLLM support), and MCP server mode. Uses httpx with request throttling, Fernet encryption for auth tokens, and patchright/CDP for browser-based login when needed.

Quickstart

Get the boss-agent-cli source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/can4hou6joeng4/boss-agent-cli.git2cd boss-agent-cli3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

AI Agent Job Search Automation

Integrate via MCP or Python SDK into Claude, Cursor, or custom agents for keyword search + welfare-based filtering (e.g., 'dual weekends + five insurances') with structured JSON output suitable for downstream LLM processing and candidate scoring.

Local Job Candidate Pool Management

Build a personal shortlist with local tagging, offline comparison, funnel statistics, and AI-driven JD analysis without uploading data to cloud. Useful for coordinated job search across multiple platforms with local resume polishing and interview prep.

Recruiter Workflow Automation (Read-Heavy)

For recruiters, expose job listing & online/offline status via CLI; candidate search and messaging default to COMPLIANCE_BLOCKED, forcing manual completion on the platform website to stay within terms of service.

Implementation considerations

Authentication is user-initiated via browser login (patchright/CDP) and stored encrypted locally (~/.boss-agent/config.json, Fernet + PBKDF2). Requires managing browser bridge daemon for CDP access; Cookie/auth token drift is a known risk.
Request throttling uses Gaussian-distributed delays to avoid platform rate limits and detection; tuning `--request-interval` is critical if self-hosting or running concurrent agents against the same account.
Welfare filtering (core feature) requires parsing job card HTML to extract and match benefits; platform HTML changes break the adapter. Monitor release notes and error codes carefully for API drift.
Schema-driven design means all capability introspection flows through `boss schema` command; agent hosts (MCP, Python SDK, subprocess) must parse and respect `COMPLIANCE_BLOCKED` error responses.
Local caching (SQLite WAL) can grow; consider periodic cleanup of search results and logs using `boss clean` or manual `~/.boss-agent/` purge if storing months of data.

When to avoid it — and what to weigh

Need Full Candidate Data Pipeline — Project requires bulk scraping candidate profiles, resume downloads, or automated messaging. The tool explicitly blocks these operations by design and returns COMPLIANCE_BLOCKED; workarounds are discouraged.
Unsupported Job Boards — If targeting 51job (Qiancheng) heavily, note it is registered as a placeholder adapter returning NOT_SUPPORTED until 'read-only research threshold is met.' Only Zhipin and Zhilian have production support.
No Local Development Environment — Tool requires Python 3.10+, SQLite, patchright (Chromium binary), and local auth token storage. Not suitable for serverless, container-only, or air-gapped cloud deployments without adaptation.
Real-Time Bidirectional Syncing — The CLI is pull-oriented (search, detail, cache locally). If you need push-to-platform workflows (bulk apply, bulk chat, live candidate updates), manual steps are mandatory by compliance design.

License & commercial use

MIT License (permissive OSI-approved). Permits commercial use, modification, and redistribution with no warranty. See LICENSE file for full text.

MIT license permits commercial use. However, review the project's disclaimer and compliance warnings: the tool enforces default 'low-risk mode' that blocks sensitive operations (bulk messaging, candidate data scraping). Commercial deployments must respect BOSS Zhipin and other platform terms of service. Verify that your use case (e.g., recruiter workflow, job search agent) aligns with platform ToS and local regulations. Atlas Cloud and Doloffer sponsors are mentioned but appear to be third-party service recommendations, not part of the license.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Good
Assessment confidence	High

Security considerations

User login credentials are encrypted locally (Fernet + PBKDF2, machine-bound) and never transmitted to the CLI vendor. Auth tokens stored in ~/.boss-agent/ are encrypted at rest. The tool does not upload job data, candidate info, or resumes by default unless explicitly directed (e.g., to OpenAI API for AI features). Platform wind-control risk is acknowledged: tool stops auto-access if blocked, does not retry or circumvent. No exploit details in public docs. Validate credential handling and ensure secure local filesystem access before deploying to shared systems.

Alternatives to consider

Direct BOSS Zhipin API / WebScraping Libraries (BeautifulSoup, Selenium)

Lower-level, no compliance guardrails; requires your own auth, caching, and error handling. Suitable if you need custom filtering or multi-board pipelines but accept higher maintenance and platform ToS risk.

Recruitment SaaS Platforms (Lever, Greenhouse, Workday)

If you are building a recruiter product, use official recruitment APIs instead. boss-agent-cli is job-seeker/agent-focused; recruiter features are read-only by design.

LLM-Native Job Search (e.g., Custom GPT, Claude w/ Web Browsing)

Less structured, no local control, higher API costs, less compliance oversight. boss-agent-cli trades some LLM flexibility for deterministic, local, and compliant job search workflows.

Software development agency

Build on boss-agent-cli with DEV.co software developers

Explore boss-agent-cli's MCP integration, Python SDK, or CLI mode to enable your agents with structured, local-first job discovery and candidate management—no data uploads, compliance by default.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

mcp-client-cli agents-cli mcp2cli

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI Development API Development Talk to DEV.co

boss-agent-cli FAQ

Can I use boss-agent-cli to send bulk messages or auto-apply to jobs?

No. These operations return COMPLIANCE_BLOCKED by default. The tool intentionally enforces low-risk mode to comply with platform ToS. You must complete messaging, applying, and contact exchange manually on the BOSS Zhipin website.

What happens if the platform changes its HTML or API?

Adapters may break. Monitor releases and error codes. If a welfare-filter or job-detail scraper breaks, file an issue or wait for a fix. The tool includes diagnostics (boss doctor) to help identify drift.

Can I run multiple agents concurrently on the same BOSS Zhipin account?

Technically yes, but risky. The tool uses local SQLite caching and request throttling; concurrent writes to ~/.boss-agent/ may conflict. Throttling may also trigger platform rate-limiting. Isolate per-agent data directories or use sequential scheduling.

Does the tool upload my resume or candidate data?

No, by default. Local resume storage and AI features use local models or optional OpenAI-compatible backends (Ollama, vLLM, Claude). Only explicit API calls send data (e.g., if you configure a remote AI service). Check your config before use.

Software development & web development with DEV.co

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If boss-agent-cli is part of your mcp servers roadmap, our team can implement, customize, migrate, and maintain it.

Ready to Add Compliant Job Search to Your AI Agent?

Explore boss-agent-cli's MCP integration, Python SDK, or CLI mode to enable your agents with structured, local-first job discovery and candidate management—no data uploads, compliance by default.

Talk to DEV.co Schedule a Consultation