fdp

Authentication

fdp login mints a SciToken via the pelican GitHub-OAuth flow and caches it under ~/.fdp/cache/<device>.token. fdp run ... auto-acquires a token when none is valid and the session is interactive; set FDP_NO_AUTO_LOGIN=1 to disable that (e.g. in batch jobs). fdp logout deletes the cached token. fdp env never launches the flow. Resolution order: -t/--bearer-token, then $BEARER_TOKEN, then the managed cache, then the legacy ~/.fdp/token.

Devices

fdp discovers devices (tokamaks) from installed packages: toksearch_d3d contributes d3d, toksearch_mast contributes mast.

Most commands need no device selection:

• fdp env / fdp run / fdp chat / fdp query compose the environments of all installed devices. Their variables are disjoint, so the union is well-defined. If two devices ever set the same variable to different values, fdp reports the conflicting variable and asks you to choose. chat and query differ in one way: with no device installed at all they warn and carry on, since an LLM session that never fetches shot data is still useful.
• fdp ls uses the device that has an origin server.
• fdp login / fdp logout use the device that requires a bearer token.

To choose explicitly, in increasing order of persistence:

fdp --device d3d ls /archives       # one invocation (also: fdp ls -D d3d)
export FDP_DEFAULT_DEVICE=d3d       # one shell
fdp device use d3d                  # recorded in ~/.fdp/config.toml
fdp device use --clear              # undo

fdp device list shows each installed device and its capabilities.

AI assistant

fdp fronts the TokSearch LLM agent (toksearch.llm), which turns plain-language requests into executable TokSearch pipelines and runs them against the FDP stack:

fdp query "Fetch ip for shot 200000 and report the peak current in MA."
fdp chat                # interactive terminal session
fdp chat --gui          # local Gradio GUI (--no-browser to skip the tab)
fdp backends            # list available LLM backends/presets

Both accept --backend, --model, and -n/--max-iterations. Backend resolution: --backend → $FDP_LLM_BACKEND → ~/.fdp/config.toml [llm].backend → built-in anthropic (needs ANTHROPIC_API_KEY). Other backends: openai (OPENAI_API_KEY), claude-max (Claude Max plan via claude login, no API key), and site presets like amsc (key in ~/amsc_api_key), contributed by device packages.

The agent's documentation library also serves your coding assistant:

fdp skills list                       # available skills + install status
fdp skills install                    # → ~/.claude/skills (Claude Code)
fdp skills install --backend cursor   # or codex; --force to overwrite

MCP-capable agents can consume the same skills from the standalone server (toksearch >= 2.8.2): claude mcp add toksearch-skills -- fdp run python -m toksearch.llm.mcp.

Full reference: https://ga-fdp.github.io/toksearch/latest/llm/ and the tutorial at https://ga-fdp.github.io/toksearch/latest/LLM_Tutorial/.