Skill Recordings Support Agent CLI

CLI tool for the Skill Recordings support platform. Named skill after Skill Recordings (the company), not Claude skills.

Quick start

skill front search "refund" --inbox inb_4bj7r  # Search conversations
skill front search --status unassigned          # Filter-only search
skill front conversation <id> -m                # Conversation with history
skill front assign cnv_xxx tea_xxx              # Assign to teammate
skill front reply cnv_xxx --body "On it"        # Draft reply (never auto-sends)
skill front api GET /me                         # Raw API escape hatch
skill inngest events --after 1h                 # Recent workflow events
skill inngest failures --after 2h               # Aggregate failure analysis

Inngest Commands

List events

skill inngest events                      # List recent events (default 20)
skill inngest events --after 2h           # Events from last 2 hours
skill inngest events --after 1d           # Events from last day
skill inngest events --name "support/inbound.received"  # Filter by name
skill inngest events --limit 50           # Limit results
skill inngest events --json               # JSON output
skill inngest events --dev                # Use dev server (localhost:8288)

Get event details

skill inngest event <id>                  # Event details + triggered runs
skill inngest event <id> --json           # JSON output

Get run details

skill inngest run <id>                    # Run status, output, timing
skill inngest run <id> --json             # JSON output

Investigation Commands (Agent-Optimized)

These commands output JSON by default, optimized for agent consumption with aggregate stats and anomaly detection.

skill inngest inspect <event-id>          # Deep dive: event + runs + results
skill inngest failures --after 2h         # Aggregate failure analysis
skill inngest stats --after 1d            # Stats with anomaly detection
skill inngest trace <run-id>              # Full workflow trace
skill inngest search "email@example"      # Search event data for patterns
skill inngest search "refund" --field result  # Search specific field

Inspect (Deep Dive)

skill inngest inspect 01KFxxx
# Returns: event details, all triggered runs with results/errors, durations

Failures (Aggregate Analysis)

skill inngest failures --after 2h --limit 20
# Returns: total failures, grouped by error message, failure details

Stats (Pattern Detection)

skill inngest stats --after 1d
# Returns: event counts by type, workflow outcomes, skip reasons, anomalies
# Anomalies detected: high failure rate, more skipped than completed, etc.

Trace (Workflow Debug)

skill inngest trace 01KFxxx
# Returns: run details, input, result, error, duration

Search (Pattern Matching)

skill inngest search "purchase_id" --after 1h
skill inngest search "error" --field result --limit 100
# Returns: matching events with data

Cancel run

skill inngest cancel <id>                 # Cancel running function

Send signal

skill inngest signal <name>               # Resume waiting function
skill inngest signal approval:draft_123   # Example: approval signal
skill inngest signal <name> --data '{"approved":true}'  # With payload

Front Commands

Search conversations

skill front search "payment failed"                              # Text search (subject + body)
skill front search "refund" --inbox inb_4bj7r --status unassigned  # With filters
skill front search "is:unreplied" --inbox inb_4bj7r --after 1706745600
skill front search --status unassigned --inbox inb_4bj7r         # Filters only, no text
skill front search "from:dale@a.com tag:tag_14nmdp" --limit 50  # Inline Front syntax
skill front search                                                # Shows full query syntax help

Filters (CLI flags): --inbox, --tag, --assignee, --status, --from, --after, --before, --limit, --json

Inline Front syntax (in the query arg): inbox:inb_xxx, tag:tag_xxx, from:email, to:email, cc:email, recipient:email, assignee:tea_xxx, participant:tea_xxx, author:tea_xxx, mention:tea_xxx, commenter:tea_xxx, is:<status>, before:<ts>, after:<ts>, during:<ts>, custom_field:"Key=Value"

Status values: open, archived, assigned, unassigned, unreplied, snoozed, trashed, waiting

Logic: Multiple filters = AND. Multiple from/to/cc/bcc = OR. Max 15 filters.

Get message

skill front message <id>                  # Full message details (body, author, recipients)
skill front message msg_xxx --json        # JSON output

Get conversation

skill front conversation <id>             # Conversation details
skill front conversation cnv_xxx -m       # Include message history
skill front conversation <id> --json      # JSON output

Inbox & conversations

skill front inbox                         # List all inboxes
skill front inbox inb_xxx                 # List conversations in inbox
skill front inbox inb_xxx --status unassigned  # Filter by status
skill front inbox "AI Hero"               # Lookup by name
skill front inbox inb_xxx --tag "500 Error" --limit 100 --json

Assign / unassign conversations

skill front assign cnv_xxx tea_xxx        # Assign to teammate
skill front assign cnv_xxx --unassign     # Remove assignee
skill front assign cnv_xxx tea_xxx --json # JSON output

Tag / untag conversations

skill front tag cnv_xxx "bug report"      # Add tag by name (case-insensitive)
skill front tag cnv_xxx tag_xxx           # Add tag by ID
skill front untag cnv_xxx "bug report"    # Remove tag by name
skill front untag cnv_xxx tag_xxx --json  # JSON output

Reply (draft only — HITL)

skill front reply cnv_xxx --body "Thanks for reporting this"
skill front reply cnv_xxx --body "We're looking into it" --author tea_xxx
skill front reply cnv_xxx --body "..." --json
# NEVER auto-sends. Creates a draft. Review and send from Front.

Tag management

skill front tags list                     # All tags with conversation counts
skill front tags list --unused            # Tags with 0 conversations
skill front tags delete tag_xxx           # Delete a tag
skill front tags rename tag_xxx "New Name"
skill front tags cleanup                  # Interactive: dedup, fix variants, remove obsolete

Teammates

skill front teammates                     # List all teammates
skill front teammates --json
skill front teammate tea_xxx              # Get teammate details
skill front teammate tea_xxx --json

Archive & bulk operations

skill front archive cnv_xxx               # Archive single conversation
skill front bulk-archive --inbox inb_xxx --dry-run  # Preview bulk archive
skill front bulk-archive --inbox inb_xxx --tag "spam" --execute

Report & triage

skill front report --inbox inb_xxx        # Inbox forensics report
skill front triage --inbox inb_xxx        # AI-powered triage
skill front triage --inbox inb_xxx --auto-archive  # Auto-archive noise

Raw API passthrough

skill front api GET /me                   # Escape hatch for any endpoint
skill front api GET /conversations/cnv_xxx
skill front api PATCH /conversations/cnv_xxx --data '{"status":"archived"}'
skill front api POST /conversations/cnv_xxx/tags --data '{"tag_ids":["tag_xxx"]}'
# Always outputs JSON. Supports GET, POST, PATCH, PUT, DELETE.

Pull conversations (eval datasets)

skill front pull-conversations --inbox inb_xxx --output data.json

Linear Commands

⚠️ Personal API Key Required for Write Operations

All write operations (create, update, assign, close, comment, etc.) require a personal LINEAR_API_KEY.

# Set up your personal API key
skill config init
skill config set LINEAR_API_KEY=lin_api_xxx

# Without a personal key, write commands will fail with:
# "Write operations require a personal API key for LINEAR_API_KEY."

Read Operations (no personal key needed)

skill linear issues                           # List issues
skill linear issues --team ENG                # Filter by team
skill linear issues --state "In Progress"     # Filter by state
skill linear issues --assignee me             # My issues
skill linear issue ENG-123                    # Get issue details
skill linear search "bug"                     # Search issues
skill linear teams                            # List teams
skill linear users                            # List users
skill linear states ENG                       # Workflow states for team
skill linear labels ENG                       # Labels for team
skill linear projects                         # List projects
skill linear my                               # My assigned issues

Write Operations (require personal API key)

skill linear create "Fix the bug" --team ENG                    # Create issue
skill linear create "Task" --team ENG --priority 1 --state "Todo"
skill linear update ENG-123 --priority 0 --estimate 3           # Update issue
skill linear assign ENG-123 --to user@example.com               # Assign
skill linear assign ENG-123 --unassign                          # Unassign
skill linear state ENG-123 --state "In Review"                  # Change state
skill linear close ENG-123                                      # Close issue
skill linear close ENG-123 --reason "Fixed in PR #42"           # Close with reason
skill linear comment ENG-123 --body "Working on this"           # Add comment
skill linear label ENG-123 --add "bug"                          # Add label
skill linear label ENG-123 --remove "feature"                   # Remove label
skill linear link ENG-123 --blocks ENG-456                      # Link issues
skill linear link ENG-123 --blocked-by ENG-456
skill linear link ENG-123 --related ENG-456

JSON Output with HATEOAS

skill linear issue ENG-123 --json
# Response includes:
# - _meta.personal_key_hint: Warning about write operations
# - _actions[].requires_personal_key: true for write actions
# - _links: Related resources

Other Commands

Initialize integration

skill init                                # Interactive mode
skill init my-app                         # Quick mode with name
skill init --json                         # JSON output

Setup wizard

skill wizard                              # Interactive property setup
skill wizard --json                       # JSON output

Health check

skill health                              # Check default integration
skill health total-typescript             # Check by slug
skill health https://example.com -s secret  # Check by URL
skill health -l                           # List all registered apps
skill health --json                       # JSON output

Run evals

skill eval routing dataset.json           # Run routing eval
skill eval routing dataset.json --json    # JSON output
skill eval routing dataset.json --min-precision 0.95
skill eval routing dataset.json --min-recall 0.98
skill eval routing dataset.json --max-fp-rate 0.02
skill eval routing dataset.json --max-fn-rate 0.01

Response Analysis Commands

List recent responses

skill responses list                          # List recent responses (default 20)
skill responses list --app total-typescript   # Filter by app
skill responses list --limit 50               # More results
skill responses list --rating bad             # Only bad-rated responses
skill responses list --rating unrated         # Only unrated responses
skill responses list --since 2024-01-01       # Since date
skill responses list --json                   # JSON output

Get response details

skill responses get <actionId>                # Response details
skill responses get <actionId> --context      # Include conversation history
skill responses get <actionId> --json         # JSON output

Export for analysis

skill responses export                        # Export all with context
skill responses export --app total-typescript # Filter by app
skill responses export --rating bad           # Only bad responses
skill responses export -o bad-responses.json  # Output to file

Common Workflows

Triage an inbox

# 1. Search for unassigned conversations
skill front search --inbox inb_4bj7r --status unassigned --json

# 2. Get details on a specific conversation
skill front conversation cnv_xxx -m

# 3. Assign, tag, and reply
skill front assign cnv_xxx tea_xxx
skill front tag cnv_xxx "needs-investigation"
skill front reply cnv_xxx --body "Looking into this now"

# 4. Or bulk triage with AI
skill front triage --inbox inb_4bj7r

Find conversations from a specific sender

skill front search "from:user@example.com" --inbox inb_4bj7r
skill front search "from:user@example.com is:unreplied"

Debug empty message body

# 1. Find recent inbound events
skill inngest events --after 1h --name "support/inbound.received"

# 2. Get event details
skill inngest event 01KFxxx

# 3. Check if message has body via Front API
skill front message msg_xxx

# 4. Compare webhook data vs API data

Investigate failed workflow

# 1. Find failures
skill inngest failures --after 2h

# 2. Get run details
skill inngest run 01KFxxx

# 3. Check event that triggered it
skill inngest event <event_id>

Pipe search results to other commands

# Get all unassigned conversation IDs
skill front search --inbox inb_4bj7r --status unassigned --json | jq '.data.conversations[].id'

# Archive all snoozed conversations (careful!)
skill front search "is:snoozed" --inbox inb_4bj7r --json | jq -r '.data.conversations[].id' | xargs -I{} skill front archive {}

Environment Variables

Required in .env.local:

FRONT_API_TOKEN=           # For Front API commands
INNGEST_SIGNING_KEY=       # For Inngest API commands

JSON Output

All commands support --json for machine-readable output:

skill inngest events --json | jq '.[0].data'
skill front message msg_xxx --json | jq '.body'