Parallel Deep Research

1---2name: parallel-deep-research3description: "ONLY use when user explicitly says 'deep research', 'exhaustive', 'comprehensive report', or 'thorough investigation'. Slower and more expensive than parallel-web-search. For normal research/lookup requests, use parallel-web-search instead. Supports multi-turn: pass --previous-interaction-id from a prior research or enrichment to continue with context."4user-invocable: true5argument-hint: <topic>6compatibility: Requires parallel-cli and internet access.7allowed-tools: Bash(parallel-cli:*)8metadata:9  author: parallel10---11 12# Deep Research13 14Research topic: $ARGUMENTS15 16## When to use (vs parallel-web-search)17 18ONLY use this skill when the user explicitly requests deep/exhaustive research. Deep research is 10-100x slower and more expensive than parallel-web-search. For normal "research X" requests, quick lookups, or fact-checking, use **parallel-web-search** instead.19 20## Step 1: Start the research21 22```bash23parallel-cli research run "$ARGUMENTS" --processor pro-fast --no-wait --json24```25 26If this is a **follow-up** to a previous research or enrichment task where you know the `interaction_id`, add context chaining:27 28```bash29parallel-cli research run "$ARGUMENTS" --processor lite --no-wait --json --previous-interaction-id "$INTERACTION_ID"30```31 32By chaining `interaction_id` values across requests, each follow-up question automatically has the full context of prior turns — so you can drill deeper into a topic without restating what was already researched. Use `--processor lite` for follow-ups since the heavy research was already done in the initial turn and the follow-up just needs to build on that context.33 34This returns instantly. Do NOT omit `--no-wait` — without it the command blocks for minutes and will time out.35 36Processor options (choose based on user request):37 38| Processor | Expected latency | Use when |39|-----------|-----------------|----------|40| `pro-fast` | 30s – 5 min | Default — good balance of depth and speed |41| `ultra-fast` | 1 – 10 min | Deeper analysis, more sources (~2x cost) |42| `ultra` | 5 – 25 min | Maximum depth, only when explicitly requested (~3x cost) |43 44Parse the JSON output to extract the `run_id`, `interaction_id`, and monitoring URL. Immediately tell the user:45- Deep research has been kicked off46- The expected latency for the processor tier chosen (from the table above)47- The monitoring URL where they can track progress48 49Tell them they can background the polling step to continue working while it runs.50 51## Step 2: Poll for results52 53Choose a descriptive filename based on the topic (e.g., `ai-chip-market-2026`, `react-vs-vue-comparison`). Use lowercase with hyphens, no spaces.54 55```bash56parallel-cli research poll "$RUN_ID" -o "$FILENAME" --timeout 54057```58 59Important:60- Use `--timeout 540` (9 minutes) to stay within tool execution limits61- Do NOT pass `--json` — the full output is large and will flood context. The `-o` flag writes results to files instead.62- The `-o` flag generates two output files:63  - `$FILENAME.json` — metadata and basis64  - `$FILENAME.md` — formatted markdown report65- The poll command prints an **executive summary** to stdout when the research completes. Share this executive summary with the user — it gives them a quick overview without having to open the files.66 67### If the poll times out68 69Higher processor tiers can take longer than 9 minutes. If the poll exits without completing:701. Tell the user the research is still running server-side712. Re-run the same `parallel-cli research poll` command to continue waiting72 73## Response format74 75**After step 1:** Share the monitoring URL (for tracking progress only — it is not the final report).76 77**After step 2:**781. Share the **executive summary** that the poll command printed to stdout792. Tell the user the two generated file paths:80   - `$FILENAME.md` — formatted markdown report81   - `$FILENAME.json` — metadata and basis823. Share the `interaction_id` and tell the user they can ask follow-up questions that build on this research (e.g., "drill deeper into X" or "compare that to Y")83 84Do NOT re-share the monitoring URL after completion — the results are in the files, not at that link.85 86Ask the user if they would like to read through the files for more detail. Do NOT read the file contents into context unless the user asks.87 88**Remember the `interaction_id`** — if the user asks a follow-up question that relates to this research, use it as `--previous-interaction-id` in the next research or enrichment command.89 90## Setup91 92If `parallel-cli` is not found, install and authenticate:93 94```bash95curl -fsSL https://parallel.ai/install.sh | bash96```97 98If unable to install that way, install via pipx instead:99 100```bash101pipx install "parallel-web-tools[cli]"102pipx ensurepath103```104 105Then authenticate:106 107```bash108parallel-cli login109```110 111Or set an API key: `export PARALLEL_API_KEY="your-key"`