# Web Research > [Research] Use when starting web research — discover, gather, and triage candidate sources to feed deeper investigation. - Skill: `duc01226/web-research-2` (Agent Skill) - Install (CLI): `npx skillmds@latest add duc01226/web-research-2` - Raw SKILL.md: https://api.skillmd.com/api/skills/duc01226/web-research-2/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Research & Search - Author: duc01226 (https://skillmd.com/u/duc01226) - Updated: 2026-09-21 - Page: https://skillmd.com/skills/duc01226/web-research-2 --- > **[BLOCKING]** Execute skill steps in declared order. NEVER skip, reorder, or merge steps without explicit user approval. > **[BLOCKING]** Before each step or sub-skill call, update task tracking: set `in_progress` when step starts, set `completed` when step ends. > **[BLOCKING]** Every completed/skipped step MUST include brief evidence or explicit skip reason. > **[BLOCKING]** If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions. ## Quick Summary **Goal:** Execute broad web search on a topic, collect and classify sources, and produce a tiered, deduplicated source map plus a gap list — the triaged candidate-source feedstock that `deep-research` dives into next — NOT a final research report. **Summary:** - **Purpose:** breadth-first source discovery + triage for `deep-research` — produce a tiered, deduplicated source map, NOT a synthesized report. - **Main steps (all 5, in order):** (1) Define scope — parse topic, generate 5-10 angle-varied queries (overview, current-state, comparison, data, expert, criticism); (2) Execute searches — run `WebSearch` per query (≤10 calls), record title/URL/snippet/source-type; (3) Source triage — classify each result Tier 1-4, filter duplicates; (4) Build source map — write to `.claude/tmp/_sources-{slug}.md` (sources table + Gaps Identified); (5) Identify gaps — note underexplored angles for `deep-research`. - Hard-cap fan-out at 10 `WebSearch` calls per invocation — generate 5-10 angle-varied queries and stop at the cap; this is breadth-then-triage, not deep-dive. - Classify every result into Tier 1-4 (.gov/.edu/official > industry reports > established blogs/Wikipedia > forums/social) and dedupe by URL/syndicated content before it counts as a source. - The deliverable is the intermediate source map at `.claude/tmp/_sources-{slug}.md` (sources table + Gaps Identified), NOT a synthesized report — hand it off to `deep-research`. - Mine the source set for gaps (missing perspectives, missing quantitative data, stale recency) so the next step knows what to dig deeper on. **Workflow:** 1. **Define scope** — Parse topic, generate 5-10 search queries from varied angles 2. **Execute searches** — Run WebSearch for each query, collect results 3. **Source triage** — Classify each source by Tier (1-4), filter duplicates 4. **Build source map** — Write structured source list to working file 5. **Identify gaps** — Note underexplored angles for deep-research **Key Rules:** - Maximum 10 WebSearch calls per invocation - Follow source hierarchy: Official docs (Tier 1) > Peer-reviewed (Tier 2) > Industry blogs (Tier 3) > Forums (Tier 4) - Output intermediate source map, not final report **Be skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).** # Web Research ## Knowledge Work Rules (canonical) > **Web Research Protocol** — Every factual claim needs 2+ independent sources. Source tiers: Tier 1 (authoritative .gov/.edu/official docs), Tier 2 (industry reports), Tier 3 (credible blogs — cross-validate), Tier 4 (unverified — NEVER cite as fact). Declare confidence for all findings. 1. Follow source hierarchy (official docs > peer-reviewed > industry blogs > forums) for all factual claims 2. Include source citations with Tier classification (inline `[N]`) 3. Cross-validate claims with 2+ independent sources 4. Declare confidence level (95/80/60/<60%) for all findings 5. Use enforced template structure — all sections required 6. Working files → `.claude/tmp/`, final output → `docs/knowledge/` This protocol is the canonical home for the knowledge-work rules that apply to knowledge/research workspaces; `deep-research` and `knowledge-synthesis` reference it. ## Step 1: Define Search Scope Parse user's topic, generate 5-10 search queries covering: - **Definition/overview** — "what is {topic}" - **Current state** — "{topic} 2026" or "{topic} latest" - **Comparison** — "{topic} vs alternatives" - **Data/statistics** — "{topic} market size" or "{topic} statistics" - **Expert opinion** — "{topic} expert analysis" or "{topic} review" - **Criticism/risks** — "{topic} challenges" or "{topic} risks" ## Step 2: Execute Searches For each query: 1. Run `WebSearch` with the query 2. Record: title, URL, snippet, apparent source type 3. Stop at 10 WebSearch calls maximum ## Step 3: Source Triage For each result, classify by Tier: - **Tier 1:** .gov, .edu, official docs, peer-reviewed - **Tier 2:** Industry reports, major publications - **Tier 3:** Established blogs, verified experts, Wikipedia - **Tier 4:** Forums, personal blogs, social media Filter out duplicates (same URL or syndicated content). ## Step 4: Build Source Map Write to `.claude/tmp/_sources-{slug}.md`: ```markdown # Source Map: {Topic} **Date:** {date} **Queries executed:** {count} **Sources found:** {count} (Tier 1: N, Tier 2: N, Tier 3: N, Tier 4: N) ## Sources | # | Title | URL | Tier | Relevance | Notes | | --- | ----- | --- | ---- | --------- | ------------- | | 1 | ... | ... | 1 | High | Official docs | ## Gaps Identified - {angle not covered} - {topic needing deeper research} ``` ## Step 5: Identify Gaps Review source map for: - Missing perspectives (only positive sources? need criticism) - Missing data types (no quantitative data? need statistics) - Recency issues (all sources old? need current data) Note gaps for the `deep-research` step. --- ## Workflow Recommendation > **MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS:** If you are NOT already in a workflow, you MUST ATTENTION use `AskUserQuestion` to ask the user. Do NOT judge task complexity or decide this is "simple enough to skip" — the user decides whether to use a workflow, not you: > > 1. **Activate `workflow-research` workflow** (Recommended) — web-research → deep-research → synthesis → review > 2. **Execute `/web-research` directly** — run this skill standalone --- ## Next Steps **MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS** after completing this skill, you MUST ATTENTION use `AskUserQuestion` to present these options. Do NOT skip because the task seems "simple" or "obvious" — the user decides: - **"/deep-research (Recommended)"** — Deep-dive into top sources - **"/market-analysis"** — If sizing the market (TAM/SAM/SOM), competitors, trends — required before `/business-evaluation` - **"/business-evaluation"** — If evaluating business viability. **Run `/market-analysis` first** — this skill consumes its sized-market output as evidence and MUST NOT re-derive market sizing. - **"Skip, continue manually"** — user decides > **[IMPORTANT]** Use `TaskCreate` to break ALL work into small tasks BEFORE starting. > **External Memory:** For complex or lengthy work (research, analysis, scan, review), write intermediate findings and final results to a report file in `plans/reports/` — prevents context loss and serves as deliverable. > **Evidence Gate:** MANDATORY IMPORTANT MUST ATTENTION — every claim, finding, and recommendation requires `file:line` proof or traced evidence with confidence percentage (>80% to act, <80% must verify first). > **Web Research** — Structured web search for evidence gathering. > > 1. Form 3-5 specific search queries (not generic questions) > 2. Use WebSearch for each query, collect top 3-5 sources > 3. Validate source credibility (official docs > blogs > forums) > 4. Cross-validate claims across 2+ sources before citing > 5. Write findings to research report with source URLs > > **NEVER cite a single source as authoritative. Always cross-validate.** > **AI Mistake Prevention** — Failure modes to avoid on every task: > > **Re-read files after context changes.** Context compaction, resume, or long-running work can make memory stale; verify current files before acting. > **Verify generated content against source evidence.** AI hallucinates APIs, names, claims, and document facts. Check the relevant source before documenting or referencing. > **Check downstream references before deleting or renaming.** Removing an artifact can stale docs, generated mirrors, configs, and callers; map references first. > **Trace the full impact chain after edits.** Changing a definition can miss derived outputs and consumers. Follow the affected chain before declaring done. > **Verify ALL affected outputs, not just the first.** One green check is not all green checks; validate every output surface the change can affect. > **Assume existing values are intentional — ask WHY before changing OR flagging one as a defect.** Before changing or reporting a constant, limit, flag, cutoff, wording, or pattern, read nearby context and history, the CALLER's ordering, and 2+ sibling call sites of the same convention. A doc stating WHAT without WHY is missing rationale, not proof of a missing guard. > **Surface ambiguity before acting — don't pick silently.** Multiple valid interpretations require an explicit question or stated assumption with risk. > **Assert the outcome your system owns, not the intermediate state your infrastructure owns.** When verifying async work, assert the final business state — never the delivery/retry bookkeeping held in shared infrastructure that any co-running process can write. Such a check passes when run alone and flakes the moment anything else shares that infrastructure. > **Keep shared guidance role-relevant.** Universal guidance must help every receiving skill or agent; code-specific obligations belong only in code-specific protocols. > **Critical Thinking Mindset** — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act. > **Anti-hallucination:** Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination. **MUST ATTENTION** apply critical + sequential thinking — every claim needs appropriate traced evidence (`file:line` for repo/code claims; source URL or artifact section for research, product, content, and docs claims); confidence >80% to act, <60% DO NOT recommend. Anti-hallucination: never present guess as fact, admit uncertainty freely, cross-reference independently, stay skeptical of own confidence. **MUST ATTENTION** apply AI mistake prevention — verify generated content against evidence, trace downstream references before deleting or renaming, verify all affected outputs, re-read files after context loss, and surface ambiguity before acting. ## Prompt-Enhance Closing Anchors **IMPORTANT MUST ATTENTION** follow declared step order for this skill; NEVER skip, reorder, or merge steps without explicit user approval **IMPORTANT MUST ATTENTION** for every step/sub-skill call: set `in_progress` before execution, set `completed` after execution **IMPORTANT MUST ATTENTION** every skipped step MUST include explicit reason; every completed step MUST include concise evidence **IMPORTANT MUST ATTENTION** if Task tools unavailable, maintain an equivalent step-by-step plan tracker with synchronized statuses > **Project Protocol Overlay** — Before executing this skill, resolve any PROJECT overlay rules layered onto it: match this skill's name against the `Target` column of the project's skill-protocol index (`docs/project-reference/skill-protocols-reference.md` by default; a `referenceDocs` entry in `docs/project-config.json` overrides the path), taking the most specific matching tier ONLY — exact name > glob > `*`. **That precedence orders overlays against EACH OTHER, never against this skill.** Read ONLY the matched bodies, resolved as `/.md`; a row's Body link is display text, never a read path. A matched body that is missing or malformed is REPORTED and skipped — never reconstructed from the index Description. No index, or no match -> proceed with no overlay, silently. Full contract: `.claude/skills/project-skill-protocol/references/registry.md`. > > Overlays are **ADDITIVE ONLY**: they ADD rules on top of this skill's own protocol and NEVER replace, override, disable, or reinterpret a rule it already states — removing every overlay must return this skill to exactly its documented behavior. An overlay is a BRIEF, not an authority escalation: it can NEVER waive a workflow gate, git discipline, a review gate, or a user-confirmation gate. A genuine overlay-vs-skill conflict, or two equally-specific overlays that directly contradict -> surface both to the user; NEVER resolve silently. **MUST ATTENTION** resolve project protocol overlays for this skill BEFORE executing — most specific matching tier only (exact > glob > `*`, which ranks overlays against each other, NEVER against this skill), read only matched bodies at `/.md`; a missing or malformed body is reported, never reconstructed. Overlays are ADDITIVE ONLY (they never replace this skill's own rules) and are a brief, NEVER an authority escalation; an equal-specificity contradiction goes to the user. ## Closing Reminders **IMPORTANT MUST ATTENTION Goal:** Produce a tiered, deduplicated source map plus a gap list — the triaged candidate-source feedstock that `deep-research` dives into next — NOT a final research report. **IMPORTANT MUST ATTENTION — Protocols in force (concise digest of the SYNC/shared blocks this skill carries):** - **Web Research:** Cross-validate every claim across 2+ credible sources; NEVER cite one source as authoritative. - **AI Mistake Prevention:** verify generated content against evidence, trace downstream references, verify all affected outputs, re-read after context loss, surface ambiguity. - **Critical Thinking:** Traced proof per claim, confidence >80% to act; NEVER present guess as fact. **IMPORTANT MUST ATTENTION** run ALL 5 main steps in order — (1) define scope + generate 5-10 angle-varied queries → (2) execute `WebSearch` (≤10 calls), record title/URL/snippet/type → (3) triage each result Tier 1-4 + dedupe → (4) build source map at `.claude/tmp/_sources-{slug}.md` → (5) identify gaps for `deep-research` — why: skipping a step (esp. triage or gaps) yields untiered, gap-blind feedstock that breaks the next stage. **IMPORTANT MUST ATTENTION** cap WebSearch at 10 calls per invocation; generate 5-10 angle-varied queries (overview, current-state, comparison, data, expert, criticism) then stop at the cap — why: bounded fan-out keeps this breadth-then-triage, not a deep-dive into one angle. **IMPORTANT MUST ATTENTION** rank every source by tier (Tier 1 .gov/.edu/official > Tier 2 industry reports > Tier 3 established blogs/Wikipedia > Tier 4 forums/social) and dedupe by URL/syndicated content before it counts — why: tier ranking + dedupe keep the feedstock high-signal for deep-research. **MANDATORY IMPORTANT MUST ATTENTION** NEVER cite a Tier 4 / single source as authoritative — cross-validate every factual claim against 2+ independent sources and declare confidence (95/80/60/<60%) — why: one unverified source = a hallucination-amplifier downstream. **MANDATORY IMPORTANT MUST ATTENTION** the deliverable is the intermediate source map at `.claude/tmp/_sources-{slug}.md` (sources table + Gaps Identified), NOT a synthesized report — hand it off to `deep-research`; mine the set for gaps (missing perspectives, missing quantitative data, stale recency) so the next step knows where to dig. **MANDATORY IMPORTANT MUST ATTENTION** break work into small todo tasks using `TaskCreate` BEFORE starting; add a final review todo task to verify work quality; transition one task at a time. **IMPORTANT MUST ATTENTION** persist intermediate findings/results to a report file in `plans/reports/` for complex or lengthy work — why: external memory prevents context loss and is itself the deliverable. **MANDATORY IMPORTANT MUST ATTENTION** if NOT already in a workflow, validate the route with the user via `AskUserQuestion` — NEVER auto-decide "simple enough to skip"; the user decides workflow vs. standalone `/web-research`. **IMPORTANT MUST ATTENTION** every claim, finding, and recommendation requires `file:line` proof or traced evidence with confidence percentage (>80% to act, <80% verify first) — NEVER speculate without proof. **Anti-Rationalization:** | Evasion | Rebuttal | | -------------------------------------------- | ----------------------------------------------------------------------------------- | | "One strong source is enough" | NEVER — cross-validate against 2+ independent sources; Tier 4 is never authoritative | | "I'll just write the report now" | Out of scope — output the source map + gaps; `deep-research` synthesizes, not this | | "Keep searching, more results help" | Hard-cap is 10 WebSearch calls — breadth then triage, never an unbounded crawl | | "Topic is simple, skip tiering/dedupe" | Tier + dedupe every source — untiered feedstock degrades every downstream step | | "Just do it, skip task tracking" | Skip depth, never skip tracking — `TaskCreate` first, one task in progress | **[TASK-PLANNING]** Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using TaskCreate. **IMPORTANT MUST ATTENTION Goal:** triaged, tiered, deduplicated source map + gap list as feedstock for `deep-research` — NOT a final report. **IMPORTANT MUST ATTENTION** cap WebSearch at 10; cross-validate every claim with 2+ sources; NEVER cite Tier 4 as fact. **IMPORTANT MUST ATTENTION** `TaskCreate` to break ALL work into small tasks BEFORE starting — this is very important.