ingest-health
Skill50 starsChanged 7 months agoMITView on GitHub
Use when the user says /ingest-health, asks to import, sync, ingest, or load Apple Health / Apple Watch / HealthKit data, has a fresh export.zip from the iPhone Health app, a Simple Health Export CSV folder, or a Health Auto Export TCP live feed, or when health-mcp queries return empty because no data was ever imported. Not for querying existing health data (health-mcp tools do that) or non-Apple wearables (Garmin, Whoop, Oura).
In this repository39 other files it gives its agents.
- docs/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- setup-brainSkill
- backfill-journal-body-contextSkill
- cierre-de-llamadaSkill
- coachingSkill
- coachSkill
- daily-journalSkill
- deconstructSkill
- diagnoseSkill
- doubt-driven-developmentSkill
- evolveSkill
- for-my-teamSkill
- graphifySkill
- health-contextSkill
- health-doctorSkill
- health-setupSkill
- ingest-githubSkill
- ingest-youtubeSkill
- insightsSkill
- instinct-exportSkill
- instinct-importSkill
- interview-meSkill
- longitudinalSkill
- meeting-todosSkill
- nano-bananaSkill
- note-todosSkill
- optimize-brainSkill
- patternsSkill
- repurpose-talkSkill
- resolver-querySkill
- riseSkill
- second-brain-mappingSkill
- secret-warnSkill
- security-snapshotSkill
- setup-vault-typesSkill
- skillify-meta-loopSkill
- sunday-reviewSkill
- vault-systemSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
After a fresh Apple Health export from iPhone (mode A) or after refreshing the Simple Health Export folder (mode B)
- Distinctive: line 19
Do NOT use for:
- Distinctive: line 33
If any of these is unclear, do not run the skill.
- Distinctive: line 61
No em dashes in any user-facing prose
- Distinctive: line 80
Call the appropriate health-mcp tool:
- Distinctive: line 87
After successful import (A or B), suggest a sanity check call:
- Distinctive: line 110
After ingestion, these skills can use the data:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
ingest-health, Apple Health to DuckDB connector
Loads Apple Health data into ~/.claude/health-mcp/health.duckdb so the rest of the substrate (daily-journal, coaching, advisory-panel, patterns, insights) can read biometrics through health-mcp tools.
This is one of the substrate's ingestion connectors (slack, gmail, linear, github, notion, whatsapp, health). Each owns a single source.
When to use
- User says
/ingest-health(with or without--mode xml|csv|tcpand a path) - User asks to import, sync, capture, ingest, or pull Apple Health data
- User wants to set up the health connector for the first time
- After a fresh Apple Health export from iPhone (mode A) or after refreshing the Simple Health Export folder (mode B)
Do NOT use for:
- Querying existing health data (use health-mcp's
health_query,health_metric_series,health_recovery_score, etc.) - Writing to the vault (this skill is read-only with respect to the vault)
- Non-Apple-Health sources (Garmin / Whoop / Oura get different connectors; for multi-wearable see the open-wearables project)
Privacy notes
Apple Health data is among the most sensitive content the user owns. Read this before running:
- Storage stays local. The DuckDB file lives at
~/.claude/health-mcp/health.duckdb. Treat it as a medical record. - Export files are sensitive too. Apple's
export.zipcontains every metric ever recorded on the device. Don't commit it to git, don't sync it through cloud storage you don't trust, and delete the staging copy after import if you don't need to re-import. - Live mode is local Wi-Fi only. The Health Auto Export TCP server is unauthenticated. Run it only on a trusted network.
- The MCP never writes the vault. This skill imports into DuckDB. Vault-aware tools READ frontmatter; they never write.
If any of these is unclear, do not run the skill. Ask first.
How it works
Three modes. Pick by argument or ask the user.
Mode A: XML export.zip (default; free, manual, universal)
- User has exported from iOS Health app: Profile → Export All Health Data →
export.zip - They've moved the zip to disk (AirDrop / Files / share)
- Skill calls
health_import_xml(zip_path) - The MCP streams-parses the XML, inserts into DuckDB, returns row counts
- Idempotent: re-running on the same SHA returns
skipped: trueunlessforce=True
Mode B: Simple Health Export CSV folder (free, manual)
- User has the folder of
HKQuantityTypeIdentifier*.csvfiles from the Simple Health Export iOS app - Skill calls
health_import_csv(folder_path) - Same idempotency as XML
Mode C: Health Auto Export TCP-live (paid iOS app, real-time)
- User has Health Auto Export installed on iPhone with TCP server enabled
- Skill calls
health_live_query(metric, host, port, start, end)for the metric the user asks about - v1 returns the raw response. v0.2 will normalize into the same DuckDB schema as A and B.
Voice rules
- No em dashes in any user-facing prose
- No exclamation marks
- Direct, no fluff
- Voice rules apply to the summary the skill returns, NOT to docstrings inside health-mcp source
Invocation
The skill is a thin orchestrator. Actual import runs in Python at ${SKILL_ROOT}/ingest.py. The skill assembles the health-mcp tool calls and hands the work to the script.
When invoked:
- Parse arguments. Accept any of:
<path>(positional): infer mode from path (.zipor.xml-> A, directory -> B)--mode xml|csv|tcp--force(re-import even if SHA matches)--host HOST --port PORT --metric METRIC --start DATE --end DATE(TCP mode)
- Resolve mode + paths. If ambiguous, ask the user.
- Call the appropriate health-mcp tool:
- Mode A:
health_import_xml(zip_or_xml_path, force=...) - Mode B:
health_import_csv(folder_path, force=...) - Mode C:
health_live_query(metric, host=..., port=..., start=..., end=...)
- Mode A:
- Surface the response: row counts, skipped/imported flag, top metric types.
- After successful import (A or B), suggest a sanity check call:
health_status()for statshealth_recovery_score("<recent date>")for end-to-end smoke
Output
The skill returns a brief summary. No file writes. Example:
Imported 482,193 records / 18 workouts / 1,247 sleep segments from
/Users/<you>/Downloads/export.zip in 6.4s.
Top types:
HKQuantityTypeIdentifierStepCount: 358,201 records (2018-09-01 to 2026-05-09)
HKQuantityTypeIdentifierHeartRate: 92,047 records (2019-12-15 to 2026-05-09)
HKQuantityTypeIdentifierHeartRateVariabilitySDNN: 4,182 records
...
Try: health_recovery_score("2026-05-09") for an end-to-end smoke check.Composition
After ingestion, these skills can use the data:
- daily-journal: calls
health_journal_context(date)to prompt with last-night sleep + HRV - patterns: calls
health_floor_correlation(metric, days, vault_root)to surface biometric-vs-Floor patterns - coaching: calls
health_coaching_context(start, end, vault_root)to surface recovery markers - advisory-panel: calls
health_panel_context(date, vault_root)to inform deferral suggestions - insights: calls
health_weekly_rollup(week_start)for the weekly review
The wrapper skill health-context auto-fires these when the relevant trigger keywords appear.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

