CatPane iOS console MCP

Use this skill when you need iOS logs from CatPane's MCP runtime. It is for live debugging, reproductions, and incremental monitoring of simulator or physical-device output.

Use it when

you need to discover whether a useful capture already exists
you need to start a scoped capture for an iOS simulator or physical device
you need focused log queries instead of dumping raw console output
you need to poll for new high-signal logs during an investigation

Tool surface

```
get_status
```
— inspect captures and optionally include all currently available capture devices
```
list_devices
```
— list all currently available capture devices; focus on iOS entries
```
start_capture
```
— start buffering console logs for an iOS capture target
```
get_logs
```
— read buffered logs with filters and cursor pagination
```
clear_logs
```
— reset the buffered window for a capture without stopping it
```
stop_capture
```
— stop and remove a capture
```
get_crashes
```
— detect structured crash reports from buffered logs
```
create_watch
```
— pin high-signal matches so they survive main-buffer overflow
```
list_watches
```
— inspect active pinned watches
```
get_watch_matches
```
— poll retained watch matches incrementally

Supporting docs:

tools/status.md —
```
get_status
```
tools/list-devices.md —
```
list_devices
```
tools/capture.md —
```
start_capture
```
,
```
clear_logs
```
,
```
stop_capture
```
tools/get-logs.md —
```
get_logs
```

Recommended workflow

Check runtime state first.
- Call
```
get_status
```
  with
```
{"includeDevices": true}
```
  .
- If a suitable capture already exists and is
```
running
```
  , reuse it.
- If you only need device identifiers,
```
list_devices
```
  is the lighter call.
- Both calls can include Android devices too, so filter to
```
platform: "iOS"
```
  for this skill.
Start capture only when needed, and scope it early.
- Call
```
start_capture
```
  with
```
device
```
  when more than one capture target is available.
- For iOS, prefer
```
process
```
  ,
```
text
```
  , and simulator
```
predicate
```
  scope so irrelevant lines never enter the main buffer.
- Use
```
quiet: true
```
  on physical iOS captures when you need a lighter default stream.
- Use
```
restart: true
```
  only when replacing an existing capture on the same target.
Clear logs before a fresh reproduction.
- Call
```
clear_logs
```
  to reset the main buffer and any retained watch matches.
Create a pinned watch for the important signal.
- Call
```
create_watch
```
  for the app process, subsystem, or key error text before reproducing.
- Poll
```
get_watch_matches
```
  during the session; retained matches survive main-buffer overflow.
Check crashes early.
- Call
```
get_crashes
```
  after reproduction to surface structured crash reports without paging raw logs first.
Query with focused filters.
- Keep
```
get_logs
```
  pulls small.
- Start with
```
limit
```
  ,
```
minLevel
```
  , and
```
text
```
  .
- Use
```
process
```
  ,
```
subsystem
```
  , and
```
category
```
  to narrow results to specific system components.
- Prefer targeted queries over large unfiltered pulls.
Page with cursors.
- ```
get_logs
```
  is cursor-based.
- Reuse the same
```
order
```
  , pass
```
page.nextCursor
```
  as the next
```
cursor
```
  , and stop when
```
page.hasMore
```
  is false.
Use
since
for incremental polling.
- Keep the last processed timestamp.
- Pass it back as
```
since
```
  on the next call.
- ```
since
```
  is inclusive, so expect one boundary overlap and dedupe by
```
seq
```
  if needed.
Stop capture only when you are done.
- Use
```
stop_capture
```
  when the capture is no longer needed.

Query rules

Always prefer
```
captureId
```
or
```
device
```
once more than one capture exists. Unqualified calls only auto-resolve when exactly one capture is registered.
```
get_logs
```
reads the main in-memory ring buffer for a capture. Older entries can age out when the buffer reaches capacity.
```
create_watch
```
+
```
get_watch_matches
```
give you a second retained path for relevant lines; use that for long or noisy iOS sessions.
```
cursor
```
is exclusive:
- ```
order: "desc"
```
  returns older entries with
```
seq < cursor
```
- ```
order: "asc"
```
  returns newer entries with
```
seq > cursor
```
Use
```
page.hasMore
```
to decide whether to continue paging.
```
page.nextCursor
```
is still the correct next anchor.
```
minLevel
```
is a threshold, not an exact match. Example:
```
warn
```
returns
```
warn
```
,
```
error
```
, and
```
fatal
```
.
```
text
```
is a case-insensitive substring filter over process name, subsystem, category, and message.
```
process
```
is a case-insensitive substring filter on the originating process name.
```
subsystem
```
is a case-insensitive substring filter on the logging subsystem.
```
category
```
is a case-insensitive substring filter on the logging category.
```
since
```
must use the format:
```
MM-DD HH:MM:SS.mmm
```

Quick start

get_status

json

{
  "includeDevices": true
}

start_capture

json

{
  "device": "My iPhone",
  "process": "MyApp",
  "text": "timeout",
  "quiet": true
}

create_watch

json

{
  "device": "My iPhone",
  "name": "app-errors",
  "pattern": "timeout",
  "minLevel": "error",
  "tag": "MyApp"
}

get_watch_matches

json

{
  "device": "My iPhone",
  "watchId": "w1",
  "sinceSeq": 0,
  "limit": 100
}

get_logs

json

{
  "device": "My iPhone",
  "order": "desc",
  "limit": 50,
  "minLevel": "error",
  "process": "MyApp",
  "subsystem": "com.example.app"
}

catpane-ios-console

NPX Install

Tags

SKILL.md Content

CatPane iOS console MCP

Use it when

Tool surface

Recommended workflow

Query rules

Quick start