ascorbic/macrodata/macrodata-distill

sqlite3 ~/.local/share/opencode/opencode.db "
  SELECT s.id, s.title, p.worktree, s.time_created
  FROM session s
  LEFT JOIN project p ON p.id = s.project_id
  WHERE s.parent_id IS NULL
    AND s.time_updated > unixepoch('now', '-1 day') * 1000
  ORDER BY s.time_updated DESC
"

Task(subagent_type="general", prompt=`
Read an OpenCode conversation from the SQLite database at ~/.local/share/opencode/opencode.db.
 
Session ID: {sessionId}
Session title: {sessionTitle}
Project: {projectWorktree}
 
Use this query to extract the conversation (user prompts and assistant text responses):
 
sqlite3 ~/.local/share/opencode/opencode.db "
  SELECT
    m.id AS message_id,
    json_extract(m.data, '$.role') AS role,
    m.time_created,
    GROUP_CONCAT(
      CASE WHEN json_extract(p.data, '$.type') = 'text'
        THEN json_extract(p.data, '$.text')
      END,
      char(10)
    ) AS text_content
  FROM message m
  JOIN part p ON p.message_id = m.id
  WHERE m.session_id = '{sessionId}'
    AND json_extract(m.data, '$.role') IN ('user', 'assistant')
    AND json_extract(m.data, '$.summary') IS NULL
  GROUP BY m.id
  HAVING text_content IS NOT NULL AND text_content != ''
  ORDER BY m.time_created ASC
"
 
Filter to actual conversation content:
- Include: user messages, assistant text responses
- Exclude: tool calls, tool results, system content, compaction summaries
 
Extract and return as JSON:
{
  "distilled_actions": [
    {
      "summary": "Fixed auth bug in src/auth.ts where token refresh was racing",
      "files": ["src/auth.ts"],
      "outcome": "Added mutex lock around refresh"
    }
  ],
  "facts": [
    {
      "topic": "project-name",
      "content": "Uses JWT tokens with 15min expiry"
    },
    {
      "topic": "person-name",
      "content": "Prefers explicit error handling over try/catch"
    }
  ],
  "decisions": [
    "Chose Redis over in-memory cache for session storage because of multi-instance deployment"
  ]
}
 
Focus on:
- What was accomplished (not just discussed)
- Decisions made and their rationale
- New information about projects, people, or preferences
- File paths and specific technical details that should survive compression
 
Return ONLY the JSON, no explanation.
`)

For each action in all results:
  macrodata_log_journal(topic="distilled", content=action.summary + " Files: " + action.files.join(", "))

macrodata_log_journal(topic="distill-summary", content="Processed N sessions. Extracted X actions, Y facts.")

{
  "distilled_actions": [
    {
      "summary": "Added /distill skill to macrodata plugin",
      "files": ["plugins/macrodata/skills/distill/SKILL.md"],
      "outcome": "Skill extracts facts from conversations via sub-agents"
    }
  ],
  "facts": [
    {
      "topic": "macrodata",
      "content": "Distillation separates narrative context from retained facts for better compression"
    }
  ],
  "decisions": [
    "Coordinator updates state directly to prevent race conditions from parallel sub-agents"
  ]
}