Anti-hallucination research mode. Toggle on to enforce citation requirements, source grounding, and "I don't know" behavior. Toggle off for creative work.
How to install research-mode?
assafkip/research-mode
$npx -y skills add assafkip/research-mode --skill research-modeInstalls into the current project.
›Prefer a prompt? Paste this to your agent
Run `npx skills use "https://github.com/assafkip/research-mode" --skill "assafkip/research-mode"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/assafkip/research-mode" that are relevant to the current task. Run `npx skills add "https://github.com/assafkip/research-mode"` and select the relevant skills, then follow their instructions.
SKILL.md
| 1 | # Research Mode for Claude Code |
| 2 | |
| 3 | Anti-hallucination toggle for Claude Code. Activates citation constraints from [Anthropic's documentation](https://docs.anthropic.com/en/docs/test-and-evaluate/strengthen-guardrails/reduce-hallucinations) that force Claude to cite sources, say "I don't know" when unsure, and ground responses in direct quotes. |
| 4 | |
| 5 | > **Using [Kipi Founder OS](https://github.com/assafkip/kipi-system)?** Research mode is already built into kipi-core. Just run `/q-research <topic>`. No extra install needed. |
| 6 | |
| 7 | ## Install |
| 8 | |
| 9 | Works in Claude Code CLI, the macOS desktop app, and the VS Code extension. Open Claude Code and run these two commands: |
| 10 | |
| 11 | **Step 1 -- Add the marketplace:** |
| 12 | ``` |
| 13 | /plugin marketplace add assafkip/research-mode |
| 14 | ``` |
| 15 | |
| 16 | **Step 2 -- Install the plugin:** |
| 17 | ``` |
| 18 | /plugin install research-mode@assafkip-research-mode |
| 19 | ``` |
| 20 | |
| 21 | That's it. To confirm it installed, run `/plugin` and check that `research-mode` appears in your list. |
| 22 | |
| 23 | **Alternative -- install as a standalone skill:** |
| 24 | If you prefer not to use the plugin system, clone this repo and copy the `SKILL.md` file into your project's `.claude/skills/research-mode/` directory. |
| 25 | |
| 26 | ## Use |
| 27 | |
| 28 | ``` |
| 29 | /research-mode:research |
| 30 | ``` |
| 31 | |
| 32 | Or with a topic: |
| 33 | |
| 34 | ``` |
| 35 | /research-mode:research what caused the Change Healthcare breach |
| 36 | ``` |
| 37 | |
| 38 | Say "exit research mode" to turn it off. |
| 39 | |
| 40 | ## What it does |
| 41 | |
| 42 | Three constraints activate simultaneously: |
| 43 | |
| 44 | 1. **Say "I don't know"** -- no guessing, no inferring. If there's no credible source, Claude says so. |
| 45 | 2. **Cite everything** -- every claim must reference a file, URL, paper, or named source. Unsourced claims get retracted. |
| 46 | 3. **Quote first, then analyze** -- responses are grounded in word-for-word quotes from source material, not paraphrased summaries. |
| 47 | |
| 48 | ## Source cascade (keeps token costs low) |
| 49 | |
| 50 | Sources are checked in order. Claude stops at the first level that answers the question: |
| 51 | |
| 52 | 1. **Local files** (Grep + Read) -- zero cost, most reliable |
| 53 | 2. **WebSearch snippets** -- cite the snippet directly, skip downloading full pages |
| 54 | 3. **WebFetch** -- only when the snippet is ambiguous or user needs full quotes |
| 55 | 4. **Scholar Gateway** -- for academic papers, if available |
| 56 | |
| 57 | Token budget: 5 WebSearch, 3 WebFetch max per question. If the limit is hit, Claude summarizes what it found and asks before going deeper. |
| 58 | |
| 59 | ## What it doesn't do |
| 60 | |
| 61 | - Not always-on. It's a toggle. Turn it on for research, off for creative work. |
| 62 | - Not slow. Claude still uses tools in parallel and works efficiently. |
| 63 | - Not restrictive on new ideas. You can synthesize across sources, but inputs must be grounded. |
| 64 | |
| 65 | ## Why |
| 66 | |
| 67 | LLMs hallucinate. When you're doing research that matters, you need guardrails that force citation discipline. This plugin packages Anthropic's own recommendations into a one-command toggle. |
| 68 | |
| 69 | ## Troubleshooting |
| 70 | |
| 71 | **"Plugin not found" after Step 1:** Make sure you typed the marketplace command exactly as shown. The marketplace name is `assafkip/research-mode` (GitHub username/repo). |
| 72 | |
| 73 | **Command doesn't appear after install:** Run `/reload-plugins` to refresh, or restart Claude Code. |
| 74 | |
| 75 | **Want to scope it to one project only:** Add `--scope project` to the install command in Step 2. |
| 76 | |
| 77 | ## Built by |
| 78 | |
| 79 | [Assaf Kipnis](https://github.com/assafkip) -- built while running GTM, investor outreach, and content ops for [KTLYST](https://ktlystlabs.com) entirely through Claude Code. When your AI assistant is writing your pitch decks, researching competitors, and drafting investor briefs, hallucinated facts aren't a minor annoyance. They're a credibility risk. This toggle exists because I needed it. |
| 80 | |
| 81 | Questions or feedback: [REDACTED] |
Related
6 picksType
- handoffCompact the current conversation into a handoff document for another agent to pick up.Jul 2026
- lark-sharedUse for lark-cli setup/auth tasks: auth login/status/logout, user vs bot identity, business-domain permissions (--domain, including all/docs/drive), missing…Jul 2026
- lark-eventLark/Feishu real-time event listening / subscribing / consuming: stream events as NDJSON via `lark-cli event consume <EventKey>` (covers IM…Jul 2026
- lark-vc-agent飞书视频会议会中能力：用于让应用机器人真实加入或离开正在进行的会议，并读取当前身份可见的会中事件、发送会中文本消息或会中表情。适用于用户询问正在开的会议发生了什么、谁在发言、是否共享内容，或需要发现当前可读的进行中会议 ID。不负责已结束会议搜索、参会人快照、纪要、逐字稿或录制查询，这些使用 lark-vc 技能。Jul 2026
- ask-mattAsk which skill or flow fits your situation. A router over the skills in this repo.Jul 2026
- paperclip-create-agentCreate new agents in Paperclip with governance-aware hiring. Use when you need to inspect adapter configuration options, compare existing agent configs, draft…Jul 2026
