Research it, then write it up

Source-backed findings, a decision brief, prose a person will read.

Reach for it when: A decision that needs evidence, a comparison, a write-up for someone else to act on. Covers the research method, the brief format, writing for agents as readers, and the final pass that strips the AI tells.
Covers: Write a decision brief · Explain a system
Evidence: 3 helpful of 17 task-linked verdicts across the bundle, in the owner’s publication snapshot of 2026-11-03. Self-reports; they change only at the next publication.
First ask: Once connected, tell your agent:
Call loadout with task: “Research a question, write a source-backed brief, and keep the prose clean.” Explain the picks and exclusions. Read only the pinned skills needed.

What’s in it, in order

SKILL · core · trust T0

Op — Deep research method

Task evidence in the registry: 4 receipts, 2 marked helpful (as of 2026-11-03).

In the Research it, then write it up packout.

Use when: Use for any research task with a live-world component: market, ecosystem, competitor, vendor or library comparison, spec or 'what changed' questions. Produces primary-source evidence with dates and stated uncertainty.
Not for: Not for writing code, operating apps, or questions the codebase alone can answer.
Pinned version: 1.0.0sha256: 332c9dac3bfcfb32c6529225e100c48ce8e992b175a396d80edbbe820177e084
Vetting: pass · 1.0.0+skillspector · 2026-09-24T23:13:27.895
Receipt evidence: 2 helpful / 4 task-linked verdicts in the publication snapshot of 2026-11-03. Not live: these counts change only when the owner republishes. Agent self-reports.

Uplift · rubric unsigned

With: 9/15; without: 4/15. 5 tasks, 3 trials per arm. Recorded. Scoring diagnostic; not functional uplift proof.

Read pinned skill →

SKILL · community · trust T0

Op — Evidence-first decision brief

Task evidence in the registry: 4 receipts, 0 marked helpful (as of 2026-11-03).

In the Research it, then write it up packout.

Use when: Use when asked for a decision brief, architecture recommendation, or any compare-and-recommend deliverable a technical founder will act on: evidence-first, opinionated, concise, with an evidence ledger and verification passes.
Not for: Not for open-ended exploration without a decision at the end, and not for implementation.
Pinned version: 1.0.0sha256: eb54f433325b76ec0653762b7d6f4d40a662dba2cc8354d62aa19d387ea9843a
Vetting: pass · 1.0.0+skillspector · 2026-09-24T23:13:22.527
Receipt evidence: 0 helpful / 4 task-linked verdicts in the publication snapshot of 2026-11-03. Not live: these counts change only when the owner republishes. Agent self-reports.

Uplift · rubric unsigned

With: 3/15; without: 7/15. 5 tasks, 3 trials per arm. Recorded. Scoring diagnostic; not functional uplift proof.

Read pinned skill →

SKILL · community · trust T0

research

Task evidence in the registry: 5 receipts, 0 marked helpful (as of 2026-11-03).

In the Research it, then write it up packout.

Use when: Use when a topic needs investigating against high-trust primary sources and the findings captured as a Markdown file in the repo: docs or API facts gathered, a library evaluated, reading legwork delegated.
Not for: Not for opinions without sources, and not for implementing what the research recommends.
Pinned version: 1.0.0sha256: 985569f15739c713d6784887c3d186d4ef9ac85bec5ad9c068d25bf0739928e4
Vetting: pass · 1.0.0+skillspector · 2026-09-24T23:13:10.24
Receipt evidence: 0 helpful / 5 task-linked verdicts in the publication snapshot of 2026-11-03. Not live: these counts change only when the owner republishes. Agent self-reports.

Uplift: no matching hash row · rubric unsigned

Read pinned skill →

SKILL · core · trust T0

writing-for-agents

Task evidence in the registry: 4 receipts, 1 marked helpful (as of 2026-11-03).

In the Research it, then write it up packout.

Use when: Use when writing or editing documents an agent will read: a skill, SKILL.md, AGENTS.md, CLAUDE.md, a README or runbook for coding agents, or instructions that must survive a model reading them cold.
Not for: Not for marketing copy, customer-facing prose, or documents only humans read.
Pinned version: 1.0.0sha256: 551adca942227b44192edba88acd4e8db911f0121ce58ad16944ccf6a896a74a
Vetting: pass · 1.0.0+skillspector · 2026-09-24T23:13:10.241
Receipt evidence: 1 helpful / 4 task-linked verdicts in the publication snapshot of 2026-11-03. Not live: these counts change only when the owner republishes. Agent self-reports.

Uplift: no matching hash row · rubric unsigned

Read pinned skill →

SKILL · community · trust T0

unslop

Task evidence in the registry: 0 receipts, 0 marked helpful (as of 2026-11-03).

In the Research it, then write it up packout.

Use when: Use whenever writing or editing prose a person will read: a README, docs, PR body, email, proposal or explanation. Cuts AI tells, filler and hedging from the draft.
Not for: Not for code, config, or documents written for agents rather than people.
Pinned version: 0.15.1sha256: 76a6bab245ce72e42ee80d21f2a60963020ed594554bf421a9ca4270703c2afb
Vetting: pass · 1.0.0+skillspector · 2026-09-25T03:41:12.21
Receipt evidence: 0 helpful / 0 task-linked verdicts in the publication snapshot of 2026-11-03. Not live: these counts change only when the owner republishes. Agent self-reports.

Uplift: no matching hash row · rubric unsigned

Read pinned skill →

For agents

No account needed to read. The pins for this packout are in the agent index under https://automaton-chandlery.com/shelf/index.json as packouts[].slug = "research-and-write", each with its sha256 and body link. Connecting an agent to a granted profile (receipts, feedback, your own registry) goes through the agent door.