How to Use the Deep Company Research Mode for 6-Axis Analysis in career-ops

The deep company research mode is a built-in CLI prompt that generates a structured six-axis questionnaire for any target company and role, designed to be pasted directly into an LLM for interview preparation.

The santifer/career-ops repository provides this static prompt via the deep command to help candidates transform raw company data into actionable interview briefs. Activating the deep company research mode for 6-axis analysis requires no external keys—because the prompt lives entirely in the codebase and performs no external API calls, it keeps token usage predictable and your data under your control. This guide walks through the invocation flow, the six axes, and the key source files that make the mode work.

What Is the Deep Company Research Mode?

The deep company research mode is a research prompt defined in modes/deep.md that structures company intelligence into a concise, interview-ready brief. Its purpose is to take raw web-search results or a company-page scrape and turn them into targeted questions across six critical dimensions. The output is a markdown block that you can paste into any LLM—such as Claude, ChatGPT, or Perplexity—to generate a customized preparation document.

Unlike autonomous research agents, this mode is intentionally static. As implemented in santifer/career-ops, it performs no network calls and spawns no sub-agents. The only external interaction occurs when you manually feed the emitted prompt into your preferred LLM.

The Six Axes of Analysis

The prompt defined in modes/deep.md organizes research into six axes. When you run the deep mode, the CLI interpolates your [Company] and [Role] context—usually pulled from cv.md and config/profile.yml—into these sections:

• AI Strategy – What products or features use AI/ML, what stack they run on, whether they maintain an engineering blog, and what papers or talks they have published.
• Recent Moves – Relevant hires, acquisitions, partnerships, product launches, pivots, funding rounds, or leadership changes within the last six months.
• Engineering Culture – How teams are structured, their development practices, and cultural signals that affect day-to-day work.
• Likely Challenges – Current technical or business obstacles the team likely faces.
• Competitors – Key competitive threats and market positioning.
• Candidate Angle – How to position your specific background for the role based on the above intelligence.

How to Invoke the Deep Mode from the CLI

You trigger the prompt from the command line using the career-ops deep subcommand. The tool reads modes/deep.md, fills in the placeholders, and prints the result to stdout.

# Basic invocation – prints the 6-axis research prompt

career-ops deep --company "Acme Corp" --role "Senior ML Engineer"

Typical output begins with a context header and proceeds through the six axes:

## Deep Research: Acme Corp — Senior ML Engineer

Context: I am evaluating a candidacy for Senior ML Engineer at Acme Corp. I need actionable information for the interview.

### 1. AI Strategy

- What products/features use AI/ML?
- What is their AI stack? (models, infrastructure, tools)
- Do they have an engineering blog? What do they publish?
- What papers or talks have they presented on AI?

### 2. Recent moves (last 6 months)

- Relevant hires in AI/ML/product?
- Acquisitions or partnerships?
- Product launches or pivots?
- Funding rounds or leadership changes?

You can copy this block directly into your LLM chat interface.

Architecture and Key Source Files

The deep mode is not a monolithic script; it is a composition of static markdown files and skill manifests. According to the santifer/career-ops source code, the following files define how the mode behaves:

• modes/deep.md – Contains the core six-axis research prompt template. This is the file that the CLI loads when you run career-ops deep.
• modes/_shared.md – Enforces safety rules across all modes, including deep. It guarantees that no sub-agents are spawned and that only the static prompt is emitted.
• .agents/skills/career-ops/SKILL.md – The skill manifest that maps the CLI command career-ops deep to the modes/deep.md asset.
• AGENTS.md – High-level intent mapping that describes deep as a "structured 6-axis research prompt" for agent discovery.
• README.md – Lists deep among the available modes and provides a one-line description for end-users.

Safety Guardrails and Design Philosophy

The safety model for the deep mode is explicitly conservative. The shared rules in modes/_shared.md ensure that:

• No sub-agents are spawned during execution.
• Only static text is emitted; there is no dynamic web scraping or API calling from within career-ops.
• Token usage remains predictable because the prompt length is bounded by the template in modes/deep.md.

This design keeps you in full control of any external requests. The tool emits the prompt; you decide whether and where to send it.

Automating LLM Integration (Optional)

Because the output is plain markdown sent to stdout, you can pipe it directly into an HTTP request if you want to automate the LLM call. The example below sends the generated prompt to the Anthropic API via curl, though career-ops itself never performs this request:

career-ops deep --company "Acme Corp" --role "Senior ML Engineer" \
  | curl -X POST https://api.anthropic.com/v1/complete \
        -H "x-api-key: $ANTHROPIC_API_KEY" \
        -d @-

You must supply a valid API key and respect the provider's usage policies. The repository contains no logic for managing secrets or rate limits; that responsibility remains entirely with you.

Summary

• The deep company research mode is a static prompt in modes/deep.md that structures company research into six axes.
• Run career-ops deep --company "<name>" --role "<title>" to generate the markdown questionnaire on stdout.
• The six axes cover AI strategy, recent moves, engineering culture, likely challenges, competitors, and candidate angle.
• Safety rules in modes/_shared.md prevent sub-agent spawning and keep all external LLM interactions under your control.
• You can copy the output manually into any LLM, or pipe it into an API call for automated workflows.

Frequently Asked Questions

How do I enable the deep company research mode in career-ops?

The mode is built into the repository and requires no additional installation. As long as you have the santifer/career-ops CLI configured, the deep command is available immediately. The system discovers the prompt through the skill manifest at .agents/skills/career-ops/SKILL.md and the intent mapping in AGENTS.md.

Does career-ops make external API calls when running deep mode?

No. The career-ops deep command reads the static template from modes/deep.md, interpolates your [Company] and [Role] values from the local context, and prints the result to stdout. There is no network call performed by the tool itself. Any LLM request must be initiated manually by you.

What are the six axes in the deep mode prompt?

The six axes defined in modes/deep.md are: AI strategy, recent moves (last six months), engineering culture, likely challenges, competitors, and candidate angle. Each axis contains targeted questions designed to surface intelligence relevant to your target role.

Can I customize the deep mode prompt for a specific industry?

Yes. Because the prompt is a plain markdown file located at modes/deep.md, you can edit the template directly in your local clone of the repository. The CLI will emit your modified version on the next run. Changes remain local to your environment and do not affect the upstream safety guardrails enforced by modes/_shared.md.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}