wiki-research-personal

name	wiki-research-personal
description	Weekly research agent for

Wiki Research: Personal

Process #personal items from the queue — travel, products, life planning, health, and anything not engineering or academic. Uses the 5-step loop tuned for practical decisions.

Vault location

/app/data/vault/

When to skip the loop

If the topic is a single recommendation you need (a hotel, a specific product), do one web_search (engine brave) for recent reviews, run intelli_extract on the top result, write a one-paragraph page, stop. The full loop is for decisions with trade-offs.

The 5-step loop

Each step has a gate. Do not advance until the gate is met.

1 — Define

Before searching, write three sentences:

• Assumption: what you currently believe is true about this topic
• Decision shape: the actual decision you're trying to make (buy / choose / plan / schedule / diagnose)
• Criteria: what "good" looks like — the 3–5 factors that matter most (price, time, quality, location, durability, etc.)

Gate: a reader could list the criteria the decision will be judged on.

2 — Triage

Identify the 2–3 source types most likely to answer. For personal topics, priority order:

1. Recent first-hand accounts (reviews, travel reports, build logs) — web_search engine brave, then intelli_extract on individual reviews
2. Comparison / round-up articles from the last 12 months
3. Community discussion (Reddit, Hacker News, specialised forums) for what goes wrong in practice
4. Official specs / datasheets / pricing pages
5. Adjacent vault pages (wiki/index.md)

Run 2–3 Brave queries with the criteria from Step 1 in mind. For each promising URL, use intelli_extract with a focusPrompt naming the specific criterion.

Gate: you can name at least one strong source per criterion.

3 — Query

Pull the answer from the triaged sources. For personal topics:

• Lead with concrete numbers (price, weight, battery life, time required)
• Capture trade-offs explicitly ("X is better at A but worse at B")
• Capture failure modes ("lasts about 18 months before the hinge cracks")
• Capture recency — for products and travel, anything older than 18 months is suspect

Gate: every claim has at least one source URL attached.

4 — Verify

For every claim, assign one tier:

• FACT — directly stated in a source, with citation
• LIKELY — consistent with sources but inferred
• SPECULATIVE — your reasoning beyond what sources support
• UNKNOWN — neither you nor the sources say

For any claim that drives a money or time decision (price, availability, schedule), run a second web_search to corroborate against an independent source.

Gate: every claim has a tier.

5 — Report

Write the wiki page. Lead with a recommendation or decision matrix if the topic is a choice. Append a Limitations section.

Wiki page format

---
tags: [#personal]
last_researched: YYYY-MM-DD
---

# <Title>

> One-sentence summary.

## Define
- **Decision shape:** ...
- **Criteria:** ...

## Findings

### <Criterion 1>
- **FACT:** <claim> [^1]

### <Criterion 2>
- **FACT:** <claim> [^2]
- **LIKELY:** <claim>

### <Criterion 3>
...

## Recommendation / Trade-offs

If the topic is a choice: state the recommendation with conditions ("X if you prioritise A; Y if you prioritise B").

## Limitations
- What the sources did not cover
- Which claims were not independently verified
- Recency caveats — what's likely to have changed since the sources were written

## Sources
[^1]: <url> — accessed YYYY-MM-DD
[^2]: <url> — accessed YYYY-MM-DD

Per-run workflow

1. Read RESEARCH_QUEUE.md in the vault root.
2. Find all unchecked items (- [ ]) under ## #personal. If none, stop.
3. For each real unchecked item, run the 5-step loop.
4. Check wiki/index.md for an existing page. Update if found; otherwise create wiki/personal/<slug>.md.
5. Update wiki/index.md.
6. Mark done in RESEARCH_QUEUE.md: - [x] + — researched YYYY-MM-DD.
7. Move [x] items to ## Archive.
8. Append entry to wiki/log.md:

   ## [YYYY-MM-DD] ingest | research-queue #personal | <topics> → <pages>
   

Rules

• Never edit raw/, Openclaw articles/, or personal stuff/.
• Only write to wiki/ and RESEARCH_QUEUE.md.
• Lead with concrete numbers and trade-offs, not prose.
• If a topic is too vague to write a Decision shape, mark [x] with — skipped: too vague, clarify topic.
• Today's date comes from system context.