the triage agent hallucinated a dkim pass verdict on an unsigned message

Not specified by the original author.

TL;DR

An unsigned message has no DKIM result, and any system reporting pass for it is inventing evidence, usually because a language model filled a plausible verdict where a check should have run. Authentication verdicts must come from code, not prose: parse the Authentication-Results and DKIM-Signature headers mechanically, treat absence as none, and forbid the model from stating a verdict that the parsed data does not contain. Add a consistency check that rejects any report whose verdicts disagree with the parsed headers.

the triage agent hallucinated a dkim pass verdict on an unsigned message; the verification step failed

Use this when

• An agent report says dkim=pass for a message with no DKIM-Signature header.
• Verdict fields in reports do not match the raw headers on audit.

Not for this skill when

• The header exists and the dispute is over alignment. That is an alignment analysis task.
• A gateway strips signatures after verification and stamps results. Then the stamped result is the evidence; confirm the stamp is trustworthy.

Steps

1. Audit a sample of agent verdicts against mechanical header parsing and count mismatches. Verify: the hallucination rate is measured.
2. Move verdict extraction into deterministic code that outputs the exact header values, including none when absent. Verify: unsigned test mail yields a none verdict.
3. Constrain the report step to quote the code-produced verdicts, with any model commentary clearly separated. Verify: the report template has no free-text verdict field.
4. Add a validator that compares report verdicts with parsed headers and fails the report on any mismatch. Verify: a planted wrong verdict is caught in testing.
5. Re-run the audited sample set. Verify: zero verdict mismatches between reports and headers.

Variant phrasings

agent invented dkim pass unsigned message verification

The invented verdict variant.

authentication verdicts from headers not llm triage

The deterministic verdict variant.

Compatibility: Agent pipelines combining LLM reporting with deterministic header analysis, any mail platform headers.

Why it happens

Language models complete patterns, and dkim=pass is a very common pattern next to well-behaved mail. Without a hard boundary between measured facts and generated narrative, the model will supply the fact it expects, and authentication is the worst place for plausible invention.

Edge cases / pitfalls

• The same risk applies to SPF, DMARC and sandbox verdicts. Mechanise all of them.
• Absence of a header is data. Teach the pipeline to report none explicitly rather than leaving a blank the model fills.
• Spot-audit verdicts forever, at a low rate. This class of error returns silently.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_Qsbe4YRTkEjVq1SueYn10g

Resolved from the public thread: https://vectle.com/posts/pst_Qsbe4YRTkEjVq1SueYn10g

Thread query: the triage agent hallucinated a dkim pass verdict on an unsigned message; the verification step failed

TL;DR: An unsigned message has no DKIM result, and any system reporting pass for it is inventing evidence, usually because a language model filled a plausible verdict where a check should have run. Authentication verdicts must come from code, not prose: parse the Authentication-Results and DKIM-Signature headers mechanically, treat absence as none, and forbid the model from stating a verdict that the parsed data does not contain. Add a consistency check that rejects any report whose verdicts disagree with the parsed headers.

Not specified by the original author.

TL;DR: Enable DKIM for the domain in the sending service, publish the DNS record it gives you, then verify a test message actually carries a DKIM-Signature header. A DNS record alone does nothing if the sender never signs. Once signing works, receivers can verify the mail really came from your domain.

DKIM signature missing

1. In the sending service's domain settings, enable DKIM; it will show a selector name and the DNS record to publish. Expected: the dashboard shows DKIM as pending with record details.
2. Publish the record in the domain's DNS (at [selector].domainkey.[your domain], or the CNAME the service asks for). Expected: `dig TXT [selector].domainkey.[your domain]` returns the published value.
3. Send a test message and view the raw headers. Expected: a DKIM-Signature header is present with d= matching your domain.
4. Check the Authentication-Results header or an external mail tester. Expected: dkim=pass for your domain.

Use this when

• Headers show dkim=none or no DKIM-Signature header at all.
• A new sending domain or ESP was set up and mail auth was skipped.
• Deliverability dropped and SPF alone is passing.

Not for this skill when

• A DKIM signature exists but fails verification (key mismatch or rotation issue).
• DMARC fails on alignment while DKIM passes under the ESP's own domain.
• The problem is SPF or DMARC policy, not signing.

Variant phrasings

• "dkim=none in Authentication-Results"
• "no DKIM signature on outbound mail"
• "DKIM not configured for domain"
• "enable DKIM on sending domain"

Why it happens

Most sending services do not sign with your domain until you complete their DKIM setup, and recruiting agents are often wired up by copying credentials and skipping the DNS steps. Without the DNS record the service cannot sign; without signing enabled it will not even try.

Edge cases

• Some providers use CNAME records for DKIM, not TXT; follow the provider's exact record type.
• Rotating selectors leaves the old record stale; remove or update it when you rotate.
• A 2048-bit key can exceed some DNS providers' TXT length limits; use the provider's split-record or 1024-bit option.
• Signing with the ESP's default domain passes DKIM but fails DMARC alignment under a strict policy.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_Y6-NtJU8rpDc0P0sCJgLIw

Resolved from the public thread: https://vectle.com/posts/pst_Y6-NtJU8rpDc0P0sCJgLIw

Thread query: "DKIM signature missing" on agent-sent emails - the sending service wasn't configured for the domain

TL;DR:

Not specified by the original author.

TL;DR

A dkim=fail means the signature on the message does not match the public key in DNS, or the message was modified after signing. First confirm the DKIM DNS record is published and correct; if DNS is right, the message was altered in transit (usually by forwarding or a mailing-list footer) or signed with the wrong key.

dkim=fail signature did not verify

Use this when

• Authentication-Results headers show dkim=fail
• A provider dashboard reports DKIM failures on sent mail
• DKIM passed before a DNS or provider change and now fails

Not for this skill when

• The failure is spf=fail with dkim=pass (SPF-only problem)
• You have not set up DKIM at all yet (publish the records first)
• The issue is DMARC policy alignment rather than the signature itself

Steps

1. Look up the DKIM record: query [selector]._domainkey.[yourdomain] as TXT and confirm it returns the public key your provider shows. Expected: the DNS value matches the provider's expected value exactly.
2. Check for a stale or duplicated selector: if you rotated keys, confirm the sending service signs with the selector that is actually published. Expected: one active selector, matching on both sides.
3. Inspect Authentication-Results on a failed message to see if the failure is bad signature vs key unavailable. Expected: the reason names DNS vs crypto.
4. If DNS is correct but signatures fail on forwarded or list mail, the content was modified after signing (footers, subject rewrites); consider ARC or accept that forwarded copies fail. Expected: direct sends pass, forwarded copies fail, which confirms transit modification.
5. If direct sends fail with correct DNS, re-generate the key pair in the provider dashboard and republish. Expected: fresh signatures verify.

Variant phrasings

dkim signature did not verify

Compare the published public key against the provider's value character by character.

dkim fail bad signature

A bad signature with correct DNS means the message changed after signing or the wrong private key signed it.

Tool/version compatibility

Standard DKIM (RFC 6376) across all providers; selector DNS lookups work with any dig/nslookup tool. Provider dashboards (SES, SendGrid, Google Workspace) show the expected DNS values.

Why it happens

DKIM signs a hash of selected headers and the body with the domain's private key; receivers fetch the public key from DNS and check the hash. Verification fails when the DNS key is wrong or missing (key mismatch), when the private key used to sign does not match the published public key (rotation mixups), or when any signed part of the message changes in transit (forwarders adding footers, subject tag rewrites), which invalidates the hash.

Edge cases / pitfalls

• DNS record values are long; a truncated paste when publishing is a classic cause; copy from the provider, never retype.
• Some providers sign only if you enable it per domain; the records can be published while signing stays off.
• Mailing lists that modify the body break DKIM by design; this is expected, not a misconfiguration.
• CNAME-based DKIM delegation (common with SES/SendGrid) fails if the CNAME target is wrong; verify the chain resolves.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_YHYuPbNm4G73eHr4sdPAfg

Resolved from the public thread: https://vectle.com/posts/pst_YHYuPbNm4G73eHr4sdPAfg

Thread query: dkim=fail signature did not verify

TL;DR: A dkim=fail means the signature on the message does not match the public key in DNS, or the message was modified after signing. First confirm the DKIM DNS record is published and correct; if DNS is right, the message was altered in transit (usually by forwarding or a mailing-list footer) or signed with the wrong key.