Claude Skill

research

Run an autonomous research lane against primary sources, ending in cited findings and, when the missing facts are human-held, a questionnaire. Use for investigations of vendor materials, regulations, upstream repos, recorded calls, or any fact-finding that should run fire-and-rep

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download timharris707-skills-skills_investigate_research-b6a29f5.zip · 3 KB

Install

skills CLI npx skills add https://github.com/timharris707/skills/tree/main/skills/investigate/research
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install timharris707-skills@llmmart
Git git clone https://github.com/timharris707/skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole timharris707/skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Research

A research lane runs autonomously (fire-and-report): it investigates against primary sources (vendor materials, regulations, upstream repos, recorded calls, live systems you have access to) and ends in a cited findings file linked from the driving ticket. It runs two ways: as a research ticket inside a decision map (see the decision-map skill), and standalone for any investigation worth a recorded answer.

Contract

  1. Primary sources only. Read the actual material: the vendor doc, the upstream code, the regulation text, the recording. Secondary summaries are leads, not evidence.
  2. The deliverable is a findings file, tracked in the repo's docs (default docs/research/<topic>.md, or the findings home the team-workflow binding doc names). Material that is PII-bearing or otherwise local-only goes in a git-ignored location instead; the findings file there is still linked from the ticket by path.
  3. Findings cite their sources inline. A claim without a source is a hypothesis and is labeled as one. Distinguish clearly: what the source says, what you infer, what remains unknown.
  4. Link the findings file from the ticket and summarize the feature-level answers in a ticket comment, so downstream decision rounds can cite it without opening the file.
  5. Multiple research lanes may run in parallel: they share no state beyond the tracker.

Standalone lanes still ride the tracker

A research lane outside a map follows the same discipline a map ticket would: claim the driving ticket per the repo's claim recipe before investigating, and title the session with the item (default #<N> — <short item name>, per the orchestrate skill's lane titling), so parallel lanes stay legible on the picker and the tracker alike. An investigation with no driving ticket gets one filed first: the findings need somewhere durable to land.

The skeptic pass

Findings that will drive a build decision face a skeptic before they are reported: dispatch an independent subagent whose brief is to kill each load-bearing claim: re-read the cited source, hunt the contradicting primary material, check versions and dates. A claim the skeptic kills is corrected or relabeled a hypothesis; the findings file records that the pass ran and what it changed. When nothing downstream builds on the findings, the pass is optional; say which way you called it in the ticket summary.

When the facts are human-held

When a research lane hits facts only an external human holds (a vendor contact, a partner engineer, a pilot user), it does not stall or guess: it ends with the questionnaire terminal move, references/questionnaire.md.

Done when (checkable: verify each line before reporting complete)

  • The findings file exists at the named location, every claim cited or labeled a hypothesis.
  • The driving ticket links the file and carries the feature-level summary.
  • Standalone lane: the discipline in "Standalone lanes still ride the tracker" held: check the claim and the title on the tracker and picker, not your memory of doing it.
  • The skeptic pass ran or was skipped per its section above, and the call is recorded in the ticket summary.
  • Any human-held gaps are covered by an emitted questionnaire (per the terminal move), not by silent assumptions.

Attribution

This skill is adapted from Matt Pocock's skills (MIT), and follows them closely: research, a background agent investigating against primary sources, following every claim back to the source that owns it, leaving a cited Markdown file in the repo, and to-questionnaire, which the questionnaire terminal move tracks structure for structure: "grill the send, not the subject" is his phrase, and the gap-targeted questions, most-important-first ordering, answer stubs, welcome for partial answers, and closing catch-all are his design. The tracker half of the lane framing is his wayfinder's: research tickets run fire-and-report, in parallel, linked from the driving issue.

What this repo adds: the standalone-lane discipline (the claim recipe and session titling, the findings home, the PII carve-out) and the skeptic pass.

Files (skills)
  • agents
    • openai.yaml 116 B
      Loading…
  • references
    • questionnaire.md 1.3 KB
      Loading…
  • SKILL.md 4.7 KB
    ---
    name: research
    description: Run an autonomous research lane against primary sources, ending in cited findings and, when the missing facts are human-held, a questionnaire. Use for investigations of vendor materials, regulations, upstream repos, recorded calls, or any fact-finding that should run fire-and-report.
    ---
    
    # Research
    
    A research lane runs **autonomously** (fire-and-report): it investigates against primary sources (vendor materials, regulations, upstream repos, recorded calls, live systems you have access to) and ends in a **cited findings file** linked from the driving ticket. It runs two ways: as a `research` ticket inside a decision map (see the decision-map skill), and standalone for any investigation worth a recorded answer.
    
    ## Contract
    
    1. **Primary sources only.** Read the actual material: the vendor doc, the upstream code, the regulation text, the recording. Secondary summaries are leads, not evidence.
    2. **The deliverable is a findings file**, tracked in the repo's docs (default `docs/research/<topic>.md`, or the findings home the team-workflow binding doc names). Material that is PII-bearing or otherwise local-only goes in a git-ignored location instead; the findings file there is still linked from the ticket by path.
    3. **Findings cite their sources inline.** A claim without a source is a hypothesis and is labeled as one. Distinguish clearly: what the source says, what you infer, what remains unknown.
    4. **Link the findings file from the ticket** and summarize the feature-level answers in a ticket comment, so downstream decision rounds can cite it without opening the file.
    5. Multiple research lanes may run in parallel: they share no state beyond the tracker.
    
    ## Standalone lanes still ride the tracker
    
    A research lane outside a map follows the same discipline a map ticket would: **claim the driving ticket per the repo's claim recipe before investigating**, and **title the session with the item** (default `#<N> — <short item name>`, per the [orchestrate skill's](../../run/orchestrate/SKILL.md) lane titling), so parallel lanes stay legible on the picker and the tracker alike. An investigation with no driving ticket gets one filed first: the findings need somewhere durable to land.
    
    ## The skeptic pass
    
    Findings that will drive a build decision face a skeptic before they are reported: dispatch an independent subagent whose brief is to **kill each load-bearing claim**: re-read the cited source, hunt the contradicting primary material, check versions and dates. A claim the skeptic kills is corrected or relabeled a hypothesis; the findings file records that the pass ran and what it changed. When nothing downstream builds on the findings, the pass is optional; say which way you called it in the ticket summary.
    
    ## When the facts are human-held
    
    When a research lane hits facts only an external human holds (a vendor contact, a partner engineer, a pilot user), it does not stall or guess: it ends with the **questionnaire terminal move**, [references/questionnaire.md](references/questionnaire.md).
    
    ## Done when (checkable: verify each line before reporting complete)
    
    - The findings file exists at the named location, every claim cited or labeled a hypothesis.
    - The driving ticket links the file and carries the feature-level summary.
    - Standalone lane: the discipline in "Standalone lanes still ride the tracker" held: check the claim and the title on the tracker and picker, not your memory of doing it.
    - The skeptic pass ran or was skipped per its section above, and the call is recorded in the ticket summary.
    - Any human-held gaps are covered by an emitted questionnaire (per the terminal move), not by silent assumptions.
    
    ## Attribution
    
    This skill is adapted from Matt Pocock's skills (MIT), and follows them closely: [`research`](https://github.com/mattpocock/skills/tree/main/skills/engineering/research), a background agent investigating against primary sources, following every claim back to the source that owns it, leaving a cited Markdown file in the repo, and [`to-questionnaire`](https://github.com/mattpocock/skills/tree/main/skills/productivity/to-questionnaire), which the questionnaire terminal move tracks structure for structure: **"grill the send, not the subject" is his phrase**, and the gap-targeted questions, most-important-first ordering, answer stubs, welcome for partial answers, and closing catch-all are his design. The tracker half of the lane framing is his [`wayfinder`](https://github.com/mattpocock/skills/tree/main/skills/engineering/wayfinder)'s: research tickets run fire-and-report, in parallel, linked from the driving issue.
    
    What this repo adds: the standalone-lane discipline (the claim recipe and session titling, the findings home, the PII carve-out) and the skeptic pass.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.