X Bug Triage

Closed-loop bug triage from public X/Twitter complaints to clustered, evidence-backed GitHub issues.

Overview

Product teams learn about bugs from X/Twitter hours before internal monitoring catches them. This skill automates the pipeline: ingest complaints, classify and cluster them by bug family, scan repos for corroborating evidence, route to owners, and file issues — all with human confirmation gates. Results display directly in the terminal with optional Slack delivery for team review.

Prerequisites

• X API bearer token configured at ~/.claude/channels/x-triage/.env
• SQLite database initialized (bun run db:migrate)
• config/approved-accounts.json and config/approved-searches.json populated
• GitHub CLI (gh) for issue filing

Verify environment before starting:

!test -f data/triage.db && echo "DB ready" || echo "Run: bun run db:migrate"

!test -f config/approved-accounts.json && echo "Accounts configured" || echo "Missing: config/approved-accounts.json"

Instructions

Step 1: Intake

1. Resolve account username to ID: mcp__triage__resolve_username
2. Fetch mention timeline: mcp__triage__fetch_mentions
3. Run approved searches: mcp__triage__search_recent
4. Cross-reference mentions with search results for completeness
5. Hydrate conversation threads for posts with conversation_id: mcp__triage__fetch_conversation
6. Fetch quote tweets for high-engagement posts: mcp__triage__fetch_quote_tweets

After intake completes:

1. Call assessFreshness() from lib/freshness.ts with the combined post set and the requested window boundaries. If date_confidence is "low" or "medium", pass the warning string to the display step for rendering.
2. Collect all DegradationReport objects from intake tool responses. Call buildSourceStatusReport() from lib/source-status.ts to aggregate into a SourceStatusReport. Pass to the display step for rendering between the header and cluster list.

Step 2: Normalize

For each ingested post:

• Parse into BugCandidate (all 33 fields) using lib/parser.ts
• Classify into 12 categories using lib/classifier.ts
• Redact PII (6 types) using lib/redactor.ts
• Score reporter reliability (4 dimensions) using lib/reporter-scorer.ts
• Tag reporter_category from config/approved-accounts.json

Step 3: Match Existing Clusters

• Load active clusters from DB
• Load active overrides and suppression rules
• For each candidate, compute bug signature and match against existing clusters at >=70% overlap
• Family-first guard: different families NEVER cluster

Step 4: Create/Update Clusters

• New matches: create cluster with initial severity "low"
• Existing matches: update report_count, last_seen, sub_status
• Resolved matches: set state to "open", sub_status to "regression_reopened"
• Suppressed candidates: skip with audit log

Step 5: Repo Scan

For each cluster (top 3 repos per cluster):

• mcp__triage__search_issues — Match symptoms/errors
• mcp__triage__inspect_recent_commits — 7-day commit window
• mcp__triage__inspect_code_paths — Affected paths
• mcp__triage__check_recent_deploys — Recent releases

Assign evidence tiers (1-4) per evidence-policy.md.

Load evidence tier definitions:

!cat ${CLAUDE_SKILL_DIR}/references/evidence-policy.md

Step 6: Route Ownership

For each cluster, use strict 6-level precedence:

1. mcp__triage__lookup_service_owner
2. mcp__triage__lookup_oncall
3. mcp__triage__parse_codeowners
4. mcp__triage__lookup_recent_assignees
5. mcp__triage__lookup_recent_committers
6. Fallback mapping from config

Apply routing overrides from prior runs. Flag stale signals (>30 days).

Load routing precedence rules:

!cat ${CLAUDE_SKILL_DIR}/references/routing-rules.md

Step 7: Evaluate Severity + Escalation

Compute severity (low/medium/high/critical) based on:

• Report velocity, data loss signals, security/privacy, auth/billing lockout
• Cross-surface failure, enterprise impact, reproducibility quality
• Apply severity overrides from prior runs

Load escalation trigger definitions:

!cat ${CLAUDE_SKILL_DIR}/references/escalation-rules.md

Step 8: Display Results

Display triage results directly in the terminal as formatted markdown:

• Severity icons: red_circle critical/high, yellow_circle medium, green_circle low
• Top 5 clusters by severity (or all if <=5)
• Per cluster: report count, severity, status, assigned team, top evidence tier
• Available commands listed at the bottom

Step 9: Optional Slack Delivery

Check if claude-code-slack-channel plugin is available via mcp__slack__reply tool. If available, also deliver summary to Slack. If not, skip — terminal output is sufficient. Not an error.

Step 10: Interactive Review

Accept review commands from the user in the terminal. Parse via mcp__triage__parse_review_command.

After each command executes successfully, display the confirmation message from formatActionConfirmation() (in mcp/triage-server/lib.ts). This provides immediate user feedback for all review actions.

Load override and memory policy when processing review commands:

!cat ${CLAUDE_SKILL_DIR}/references/review-memory-policy.md

Step 11: Persist Learning

• All overrides stored in DB for future runs
• Audit log captures all actions (12 event types)
• Suppression rules created from dismiss commands
• Issue-family links created from file/merge commands

Output

Terminal markdown summary with severity-ranked clusters, evidence tiers, team assignments, and interactive command menu. Optionally mirrored to Slack.

Examples

/x-bug-triage @AnthropicAI --window 24h

Produces cluster summary, then user interacts:

> details 1
> file 2
> dismiss 3 noise
> confirm file 2

Error Handling

Resources

References: ${CLAUDE_SKILL_DIR}/references/

• schemas.md — Data model reference (BugCandidate, BugCluster, 9 DB tables)
• routing-rules.md — 6-level routing precedence
• escalation-rules.md — 6 escalation triggers
• evidence-policy.md — 4-tier evidence hierarchy
• review-memory-policy.md — Override types and application order