groundtruth: tighten messaging-performance-analyzer against Sent docs snapshot#20
Merged
Conversation
Update messaging-performance-analyzer references against references/_inputs/sent-docs-v3-2026-05-19.md so the skill keys diagnoses on real Sent error codes instead of generic buckets. - mdr-status-codes.md: rewrite around Sent's normalized error catalog (AUTH_*, VALIDATION_*, RESOURCE_*, BUSINESS_*, CONFLICT_001, SERVICE_001, INTERNAL_*), add the 3 send-time ERR_* codes that surface in the message description field, add the verified webhook payload envelope, confirm the QUEUED -> ROUTED -> SENT -> DELIVERED -> READ (+ FAILED, RECEIVED) lifecycle, and move provider-level codes to a clearly illustrative-only section. - performance-diagnosis-playbook.md: replace the generic decision tree with symptom -> Sent code mappings keyed on real codes (stuck QUEUED -> BUSINESS_002/003/008; sync batch failure -> AUTH_*/BUSINESS_*/VALIDATION_*; per-recipient FAILED -> ERR_CONSENT_BLOCKED/ROUTE_DENIED/TEMPLATE_PARAMS_INVALID; missing webhooks -> is_active/consecutive_failures/ last_successful_delivery_at). - SKILL.md: tighten the "Unverified claims" section, dropping the MDR-is-not-an-endpoint and provider-bucket disclaimers (snapshot now confirms both surface and normalized codes), and retain only the cohort-threshold heuristic and external provider IDs disclaimers. - analyze_mdr_funnel.py: add --show-errors flag that tallies ERR_* codes parsed from FAILED message descriptions. Existing behavior and exit codes unchanged; good.json still exits 0, bad.json still exits 3. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Updates
messaging-performance-analyzeragainstreferences/_inputs/sent-docs-v3-2026-05-19.mdso the skill's references and playbook are keyed on Sent's actual normalized error catalog (HTTP envelope + per-messagedescriptionfield) instead of generic provider-bucket guesses.Scope
skills/messaging-performance-analyzer/references/mdr-status-codes.md— rewritten around the full Sent error catalog (AUTH_*,VALIDATION_*,RESOURCE_*,BUSINESS_*incl. the operationally-interesting 002/003/004/005/006/007 codes,CONFLICT_001,SERVICE_001,INTERNAL_*) and the 3 send-time codes (ERR_CONSENT_BLOCKED,ERR_ROUTE_DENIED,ERR_TEMPLATE_PARAMS_INVALID). Documents the verified webhook payload envelope{field, sub_type, timestamp, payload}and the confirmedQUEUED -> ROUTED -> SENT -> DELIVERED -> READ(+FAILED,RECEIVED) lifecycle. Provider-level codes (WhatsApp131xxx, RBM, carrier) demoted to a clearly illustrative-only section.skills/messaging-performance-analyzer/references/performance-diagnosis-playbook.md— generic decision tree replaced with symptom-driven diagnosis keyed on real Sent codes:QUEUED -> ROUTED-> checkBUSINESS_002/003/008AUTH_*/BUSINESS_*/VALIDATION_*descriptionforERR_*codesis_active,consecutive_failures,last_successful_delivery_at,event_filtersskills/messaging-performance-analyzer/SKILL.md— tightened "Unverified claims" section: removed the "MDR isn't a v3 endpoint" and "exact normalized buckets unverified" disclaimers (the snapshot confirms both); kept the cohort-threshold heuristic and external provider IDs caveats.skills/messaging-performance-analyzer/scripts/analyze_mdr_funnel.py— new--show-errorsflag that talliesERR_*codes parsed from FAILED messagedescriptionfields. Existing behavior and exit codes unchanged.Test plan
bash scripts/validate-skills.shexits 0 (all 8 skills pass)python skills/messaging-performance-analyzer/scripts/analyze_mdr_funnel.py good.json-> exit 0python skills/messaging-performance-analyzer/scripts/analyze_mdr_funnel.py bad.json-> exit 3--show-errorscorrectly talliesERR_CONSENT_BLOCKED,ERR_ROUTE_DENIED,ERR_TEMPLATE_PARAMS_INVALIDon a synthetic fixture<!-- Grounded against references/_inputs/sent-docs-v3-2026-05-19.md ... -->