Stop noisy errors from paging you — in plain English

Why this matters

Every error tracker eventually drowns in “yes I know about this, move on.” Network errors from ad-blockers. Third-party SDK timeouts. Bot traffic. Old browser quirks. Each one is harmless individually, but as the wall of green-text-noise grows, you stop reading any of them, and that’s when a real bug slips through.

The classic fix is regex-based ignore rules. Which means you build them once, never touch them again, and they slowly rot as your error messages change. Six months later you have 47 ignore rules, half of them inactive, no one remembers why any of them exist.

Agentry’s suppressions are stored as fingerprint patterns + a reason field — so you can explain why you’re suppressing something in English, ask the agent to audit them later, and surgically un-suppress when needed.

What you get

• The noisy fingerprints removed from your case list
• A reason recorded per suppression so the next person (or future you) knows what’s going on
• The suppression list queryable via agentry_list_suppressions for periodic review

Walk through it

agentry_list_cases {
  project_id: "default",
  query: "NetworkError Failed to fetch script",
  status: "open"
}

{
  "cases": [
    { "id": "c_n01", "fingerprint": "f_n01",
      "message": "NetworkError: Failed to fetch script (segment.com/analytics.js)",
      "event_count": 412, "users_affected": 287 },
    { "id": "c_n02", "fingerprint": "f_n02",
      "message": "NetworkError: Failed to fetch script (googletagmanager.com/gtag/js)",
      "event_count": 198, "users_affected": 142 }
  ]
}

agentry_record_suppression {
  project_id: "default",
  fingerprint_pattern: "f_n01",
  action: "auto_resolve",
  reason: "Ad-blockers blocking segment.com/analytics.js — not actionable",
  hint_text: "If this re-appears with a non-segment URL, that's a real bug."
}

{ "suppression_id": "sup_a1", "applied_to_cases": 1 }

agentry_record_suppression {
  project_id: "default",
  fingerprint_pattern: "f_n02",
  action: "auto_resolve",
  reason: "Ad-blockers blocking gtag.js — not actionable",
  hint_text: "Same as sup_a1 but for GTM."
}

Heads-up: you currently have 14 active suppressions total. Want me to review them — some may have been added for bugs you’ve since fixed and could be removed?

The output

✅ 2 suppressions added

  sup_a1  f_n01  auto_resolve   "Ad-blockers blocking segment.com/analytics.js"
  sup_a2  f_n02  auto_resolve   "Ad-blockers blocking gtag.js"

Both cases moved to resolved status. 610 future events will be auto-resolved
on arrival (no case bump, no webhook).

Current suppressions (14 total):
  - 8 for known third-party noise (this category)
  - 3 for staging-only flakes
  - 2 for resolved bugs (could be removed)
  - 1 for "I know but won't fix" (marked permanent)

Setting it up

Nothing to install — agentry_record_suppression works the moment errors are flowing.

Three suppression actions to know:

Pattern matching is on the fingerprint (a deterministic hash of error type + normalized message + top stack frame). Same fingerprint = same suppression applies.

Variations

• “Audit my suppressions — for each one, check if the underlying error still exists. Suggest removals.”
• “Suppress all errors from Bot / Crawler user-agents.”
• “Add a prompt_hint to all RateLimitError cases telling the agent to check our Redis dashboard before investigating.”
• “Remove the suppression on case c_xxx — we want to track that one again.”