[Bug] Overly aggressive safeguard flags on legitimate development tasks (DB reads, CI inspection, security hardening)

Bug Description
False-positive safeguard flags on routine, benign development work — 4 distinct cases in one session on a legitimate production
False-positive safeguard flags on routine, benign development work — 4 distinct cases in one session on a legitimate production e-commerce repo (gacha/TCG site, own codebase, owner-authorized):

1. PR REVIEW: gh pr view <n> --json statusCheckRollup in a for-loop to see which

CI checks failed across open PRs. Read-only GitHub metadata. Flagged.

1. DATABASE: read-only psql SELECT queries against our own Supabase instance

(row counts, inventory state) via the sanctioned infisical run secret boundary.
No writes, no secrets printed. Flagged.

1. SECURITY HARDENING: asking how to improve our own website's security posture

(edge protection / rate limiting in front of paid endpoints, SPF/DMARC email
auth). Defensive security on our own property. Flagged.

1. COMPETITOR/REFERENCE RESEARCH: looking at other public websites for UI/flow

reference — standard product design research. Flagged.

Impact: each flag interrupts the session and force-switches model mid-task,
breaking continuity on long engineering work. These are the exact categories the
notice admits are over-flagged ("legitimate coding, cybersecurity, and biology
tasks"), but hitting 4 in one session on ordinary work suggests the thresholds
for read-only DB queries, CI/PR status inspection, and defensive-security
questions are too tight.

Request: loosen classification for (a) read-only DB/SQL SELECT in a developer's
own repo, (b) CI/PR metadata inspection commands, (c) defensive security work
on user-owned infrastructure.

Environment Info

• Platform: darwin
• Terminal: ghostty
• Version: 2.1.233
• Feedback ID: 67d77869-b154-450f-9c3b-073f2609640f

Errors

[]