Skip to content

TODO.md

Latest commit

 

History

History
325 lines (244 loc) · 8.28 KB

TODO.md

File metadata and controls

325 lines (244 loc) · 8.28 KB
Raw

MCP Feature Parity Tracking

This document tracks feature parity between appsignal-cli and the AppSignal MCP server.

Status legend:

  • Not started
  • [~] Partial parity
  • Complete
  • [n/a] Not a CLI parity target

Key implementation difference:

  • The MCP server has access to internal APIs and direct application data.
  • The CLI currently uses the public GraphQL API, and may need REST endpoints for metrics and other missing features.

REST migration tracking:

  • See rest-todo.md for the current GraphQL-to-REST migration inventory and implementation order.

Current Summary

Implemented in the CLI today:

  • Applications and org discovery
  • App resources: users, notifiers, namespaces, dashboards
  • Incident listing, incident detail, incident update, incident notes
  • Log search, log tail, log views, log sources
  • Trigger listing, creation, update, and archiving

Still missing or incomplete:

  • Metrics discovery and querying
  • Dashboard visual management
  • Trace inspection
  • Log line action management
  • Deploy marker access
  • Full performance overview parity
  • Some incident filter parity

Applications

get_applications

  • apps list
  • apps find --name <name> [--environment <env>]
  • apps info --app-id <id>

Notes:

  • The CLI requires org resolution because it works through the public API.
  • The org slug is persisted after apps list or apps set-org --org ....

get_app_resources

  • [~] apps resources <resource>
  • apps resources users
  • apps resources notifiers
  • apps resources namespaces
  • apps resources dashboards
  • apps resources deploy-markers covers deploy_markers
  • apps resources all provides the combined view
  • logs sources covers log_sources
  • logs views covers log_views
  • [n/a] Do not expose log_line_actions directly; use logs metrics and logs triggers

Notes:

  • The MCP tool returns more resource sections than the current CLI does.
  • The CLI now prefers one resource per call, with all as the explicit batch mode.
  • The remaining work is extending coverage.

Incidents

get_exception_incidents

  • [~] incidents list-exceptions

Covered today:

  • --app / --environment or --app-id
  • --namespaces
  • --action
  • --state (single state)
  • --query
  • --limit / --offset

Still missing:

  • start
  • end
  • revision
  • multiple states in a single request

Notes:

  • The current GraphQL query only supports a single incident state enum.
  • Time-range and deploy-scoped filters may require different API support.

get_anomaly_incidents

  • [~] incidents list-anomalies

Covered today:

  • app selection
  • pagination via --limit / --offset
  • --state

Still missing:

  • trigger_id
  • anomaly-specific states such as warmup, cooldown, and archived

get_incident

  • incidents show --number <N>

Notes:

  • The CLI incident detail query already handles multiple incident union types.

update_incidents

  • [~] incidents update --number <N>

Covered today:

  • --state
  • --severity
  • --assign
  • --assign-me
  • --unassign
  • --description
  • bulk state updates by explicit incident number

Still missing:

  • bulk severity, assignee, or description updates

Notes:

  • The CLI resolves assignee names case-insensitively and merges them with the current incident assignee list.

create_incident_note

  • incidents add-note --number <N> --content "..."

Performance And Traces

get_performance

  • [~] incidents list-performance

Covered today:

  • app selection
  • --namespaces
  • --action
  • --state
  • --query
  • pagination via --limit / --offset

Still missing:

  • start
  • end
  • revision
  • OTel action ranking output
  • richer performance overview beyond incident listing

get_traces

  • [~] samples / traces

Covered today:

  • list samples/traces for a performance action with samples list
  • fetch samples/traces from a performance incident with samples incident
  • fetch error traces from an exception incident with samples incident
  • list error traces for an exception digest with samples errors
  • inspect a trace tree by trace ID with samples show
  • inspect a span by span ID with samples show --span-id
  • inspect an error trace tree or span with samples show-error
  • inspect performance or exception traces from an incident number with samples show-incident
  • cursor pagination with --page-all

Logs

get_log_lines

  • logs search

Covered today:

  • query
  • source_ids
  • start / end
  • limit
  • order
  • JSON output for machine-readable use

CLI extras:

  • logs tail
  • --view support that resolves saved log views by name or ID
  • --page-all to auto-paginate beyond the 100-line API limit

Triggers

get_triggers

  • triggers list

manage_trigger

  • triggers create
  • triggers update

archive_trigger

  • triggers archive

Metrics

discover_metrics

  • Not implemented

get_metric_names

  • Not implemented

get_metric_tags

  • Not implemented

get_metrics_list

  • Not implemented

get_metrics_timeseries

  • Not implemented

Likely implementation work:

  • Add metrics discover
  • Add metrics names
  • Add metrics tags
  • Add metrics list
  • Add metrics timeseries

Notes:

  • Metrics parity likely requires REST endpoints under /api/v2/metrics/....
  • The MCP server also has category-level discovery behavior that may rely on server-side config unavailable to the CLI.

Dashboards

manage_dashboard

  • dashboards create
  • dashboards update

create_dashboard_visual

  • Not implemented

update_dashboard_visual

  • Not implemented

Likely implementation work:

  • Add dashboards create
  • Add dashboards update
  • Add dashboards add-visual
  • Add dashboards update-visual

Log Line Actions

manage_log_line_action

  • [~] Public CLI UX is split by product concept instead of exposing raw actions

Implemented today:

  • logs metrics list
  • logs metrics create
  • logs metrics update
  • logs metrics delete
  • logs triggers list
  • logs triggers create
  • logs triggers update
  • logs triggers delete

Still missing / undecided:

  • Decide whether log filters need a public CLI surface

delete_log_line_action

  • [~] Covered indirectly by logs metrics delete and logs triggers delete

reorder_log_line_actions

  • [n/a] Internal ingestion-order concept, not part of the public CLI UX for now

Likely implementation work:

  • Add logs metrics list
  • Add logs metrics create
  • Add logs metrics update
  • Add logs metrics delete
  • Add logs triggers list
  • Add logs triggers create
  • Add logs triggers update
  • Add logs triggers delete
  • Decide whether to add public logs filters ... commands

Not Applicable

get_more_tools

  • [n/a] Internal MCP capability-discovery tool, not a CLI feature target

Suggested Implementation Order

Phase 1: Finish Core Read/Write Gaps

  1. Add bulk incident updates
  2. Add missing incident filters where the public API allows them

Phase 2: Triggers And Metrics

  1. Add triggers list
  2. Add metrics name and tag discovery
  3. Add metrics timeseries and aggregated list queries
  4. Add broader metrics discovery if the underlying API supports it cleanly
  5. Evaluate whether log filters need a public CLI surface

Phase 3: Dashboards And Log Ingestion Controls

  1. Add dashboard create/update
  2. Add dashboard visual create/update
  3. Add log filter CLI support if product requirements justify it

Phase 4: Traces And Full Performance Parity

  1. Add richer performance overview behavior
  2. Add exception/performance timeframe and deploy filters if an API path exists

Research Notes

  • The public GraphQL API already covers more than the previous TODO suggested; this file was stale.
  • Metrics likely require REST support in the CLI in addition to the existing GraphQL client.
  • Trigger, dashboard, and log rule parity depends on whether the public API exposes the needed queries and mutations.
  • Anomaly state handling may differ from exception/performance incident state enums, so state support should be verified before designing CLI flags.