research-scout — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited research-scout (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Maintain AZIMUTH's empirical integrity by tracking source publications and staging updates to references/base-rates.md.
This skill has two modes. Trigger is determined from the user's request.
SCOUT — Search for new publications. Stage validated findings. PROMOTE — Review research/staged-findings.md. Promote, hold, or discard each staged entry.
Trigger PROMOTE when user says "promote," "review staged findings," or "what's in the queue."
references/base-rates.md draws from 8 source families. Each has a day-of-week assignment. Do not search all groups every run.
| Day | Source Family | What constitutes a finding |
|---|---|---|
| Mon | Standish CHAOS + PMI Pulse of the Profession | New annual edition; changed on-time/overrun rates |
| Tue | CB Insights startup failure reports | New edition (last known: 2021); updated failure-cause percentages |
| Wed | BLS Business Employment Dynamics | New annual entrepreneurship/survival cohort tables |
| Thu | McKinsey large IT projects + transformation research | New survey citing overrun or value-delivery figures |
| Fri | Nielsen Breakthrough Innovation + consumer launch research | New consumer product failure rate data |
| Sat | Pendo Feature Adoption + B2B SaaS adoption research | New feature utilization dataset |
| Sun | Schmidt & Hunter follow-on meta-analyses + KPMG/Bain M&A + Flexera State of the Cloud | New predictive validity meta-analysis; M&A value-destruction update; cloud migration overrun data |
For today's source family, run:
WebSearch: "[source name]" [current year] reportWebSearch: site:[publisher domain] [topic] [current year]Maximum 3 searches per source family. Do not search topics outside the assigned family.
Keep only:
references/base-rates.md (direction reversal, or >15% change in a cited rate)Discard:
Append each surviving finding to research/staged-findings.md:
## [YYYY-MM-DD] [Source Name] — [Brief title]
**URL:** [direct link to primary source — not a search results page]
**Source family:** [which rotation day]
**Current base-rates.md claim:** "[quote the exact sentence this finding would update or replace]"
**New finding:** [what the source actually says — one sentence, no paraphrase that changes numerical meaning]
**Impact:** UPDATE existing claim / ADD new claim / CONTRADICTS cited source / IMPROVES attribution
**Confidence:** High / Medium / Low — [one reason]
**Status:** STAGEDMaximum 3 findings per SCOUT run.
If 0 findings survive filtering:
## [YYYY-MM-DD] NO NEW FINDINGS
Sources searched: [list]RESEARCH SCOUT — [date]
Source family searched: [which rotation]
Searches run: [N]
Findings staged: [N]
Discarded: [N] — [brief reason if any]
Staged:
[list titles]
Run 'promote staged findings' to review queue.Read research/staged-findings.md. For each entry with Status: STAGED:
references/base-rates.mdreferences/base-rates.md with the new or updated claim. Update entry status to PROMOTED — [date]DISCARDED — [reason]Do not edit references/base-rates.md without explicit user confirmation.
references/base-rates.md without PROMOTE mode and explicit confirmationreferences/startup-failures.md, software-failure-patterns.md, launch-risks.md) are out of scope — route any findings about those to a separate editorial session~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.