Research scout

Skill talgacapri/pm-os/.claude/skills/research-scout

Find new or contradictory external signals (web search, Reddit, Hacker News, Quora) and stage only validated learnings in memory/long-term-memory.md. Use when the user asks for new research, fresh strategies/tools announcements, or contradiction checks against existing docs.From its SKILL.md

Install

npx -y skills add talgacapri/pm-os --skill research-scout

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.3 KB, 476 tokens by cl100k_base, as published. Nobody here has run it

Research Scout

Find new information that challenges or extends current project knowledge.

Scope

This skill does one job:

Gather fresh signals from web search, Reddit, Hacker News, and Quora.
Cross-check against local docs so only novel or contradictory items survive.
Stage validated findings in memory/long-term-memory.md under ## new_learnings.

Commands

/research-scout run
/research-scout weekly-review
/research-scout status

run: Execute the scout pipeline and append validated findings.
weekly-review: Promote confirmed patterns and clear staging.
status: Read current staged learnings and promoted patterns.

Execution Rules

Use these source classes every run:
- Web search (Google News RSS or equivalent search feed)
- Reddit (search RSS)
- Hacker News (Algolia API)
- Quora (search constrained to quora.com)
Keep only findings that are either:
- New: not already represented in local docs
- Contradictory: challenge an existing assumption or prior guidance
Discard duplicates and low-signal items.
Each accepted finding must include:
- Timestamp (UTC ISO)
- Source URL
- One-line note on what changed or was added

Storage Contract

Write staged findings to:

memory/long-term-memory.md → ## new_learnings

Entry format:

- [ ] 2026-04-25T18:00:00Z | source: https://example.com | note: Adds a new workflow for AI eval gating before rollout.

Weekly Promotion Contract

Weekly review promotes repeated/confirmed patterns into:

memory/long-term-memory.md → ## promoted_patterns
context-library/pm-os-learning-log.md → ## Research Scout Promoted Patterns

After promotion, clear ## new_learnings so it is a fresh staging queue.

Automation

Run through GitHub workflows:

.github/workflows/research-scout-nightly.yml (3x nightly)
.github/workflows/research-scout-weekly.yml (weekly promotion + clear)

Core script:

scripts/research-scout/research_scout.py

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.