You MUST use a todo list to complete these steps in order. Never move on to one step if you haven't completed the previous step. If you have multiple read steps in a row, read them all at once (in parallel).

Add all steps to your todo list now and begin executing.

## Steps

1. Check if Slack is connected. If not, help the user connect via the Slack workspace
discovery skill. Once connected, proceed with the fetch.

Tell the user: "Fetching Slack engagement data... This takes about 30 seconds."


2. [Gather Arguments: Fetch Workspace Messages] The next step has the following requirements for arguments, do not proceed until you have all the required information:
- `outputPath`: session/manager-slack-raw.json
- `daysBack` (default: "14"): 14

3. [Run Code: Fetch Workspace Messages]: Call `run_script` with:

```json
{
  "file": {
    "path": https://sk.ills.app/code/sentiment.slack.fetch/preview,
    "args": [
      "outputPath",
      "daysBack"
    ]
  },
  "packages": null
}
```

4. Verify the fetch succeeded by checking that `session/manager-slack-raw.json` exists
and contains thisWeek and lastWeek message arrays. If the file is missing or empty,
tell the user the fetch failed and stop.


5. [Read Slack Engagement Signals]: Read the documentation in: `skills/sauna/[skill_id]/references/manager.slack.signals.md` (Reference for signal extraction and cache format)

6. Read `session/manager-slack-raw.json` and aggregate per-person engagement signals.

**For each unique user in the message data, calculate:**
1. Message count this week
2. Message count last week
3. Frequency delta as percentage: ((thisWeek - lastWeek) / lastWeek) * 100
4. After-hours message percentage (messages where isAfterHours is true)
5. Weekend message count (messages where isWeekend is true)
6. Channel count (distinct channels they posted in)

**Also calculate team averages:**
- Average messages per person
- Average after-hours percentage
- Average channel count

Handle edge cases:
- If a person has 0 messages last week, frequency delta is "new" not infinity
- Skip bot users or system messages
- Note users with very few messages (< 5) as "low activity"


7. [Read People Directory]: Discover relevant information in the user's filesystem at `documents/knowledge/people/*.md`. (Check existing team member profiles to merge with)

8. Write the aggregated signals to `documents/manager/signals.json` (session file for this analysis).
Format:

```json
{
  "fetchedAt": "[current ISO timestamp]",
  "workspace": "[from raw data]",
  "period": "[from raw data period field]",
  "teamSignals": [
    {
      "name": "[user name]",
      "thisWeekMessages": [count],
      "lastWeekMessages": [count],
      "frequencyDelta": [percent or "new"],
      "afterHoursPercent": [percent],
      "weekendMessages": [count],
      "channelCount": [count]
    }
  ],
  "teamAverages": {
    "avgMessages": [number],
    "avgAfterHoursPercent": [number],
    "avgChannelCount": [number]
  }
}
```

Sort teamSignals by name for consistent ordering.


9. For team members with significant activity (10+ messages), update their entity profiles.
For each person, run `skills/sauna/[skill_id]/references/recipes/context.entity.resolve_person.md` with:
- name: their display name
- source: "Slack: Team analytics"
- relationship context: "Direct report" or "Team member" based on existing knowledge
- Add engagement signals as metadata (frequency delta, after-hours %)

The resolver handles merging with existing profiles.


10. Report to the user:
"✓ Slack signals cached for [N] team members.
 Period: [date range]
 Signals available: message frequency, after-hours activity, weekend work, channel breadth"

Note any limitations observed (e.g., "Only seeing [N] channels - signals reflect
visible workspace activity").