LLM CLI for bug bounty
LLM CLI adds an OpenAI-compatible model through one YAML file, then takes piped files from the shell. It suits sorting saved notes and checking drafts in a script.
Setup
- model_id: kimi model_name: kimi-k3.1-pro-uncensored api_base: "https://api.silkcompute.com/v1" api_key_name: silkcompute- model_id: flash model_name: glm-5.3-flash-uncensored api_base: "https://api.silkcompute.com/v1" api_key_name: silkcomputellm keys set silkcomputecat notes.md | llm -m kimi 'Draft a HackerOne report from these notes. Mark gaps with [CHECK].'Source: LLM CLI docs. Read November 18, 2026.
Steps
- Create the YAML file
It sits in LLM's config folder; the docs show where.
- Store the key
api_key_name points to a key saved with llm keys set.
- Pipe a file in
The prompt in quotes applies to the piped text.
- Use the cheap alias for bulk
flash runs GLM-5.3 Flash for long note files.
Use it in a hunt
| Job | Ask |
|---|---|
| Group hosts | cat scope.md hosts.txt, then ask for hosts grouped by scope rule. |
| Check a draft | cat draft.md notes.md, then ask which claims the notes do not support. |
| Caption evidence | Ask for a one-line caption per file name in evidence.txt. |
Provider guide: Silk Compute LLM CLI setup.