agentleFS
afain12afain12/coo-twin/.claude/skills/notion-map/SKILL.md

notion-map

Skill0 starsChanged 6 months agoMITView on GitHub

Property-role mapper for Notion. Consumes /notion-probe output, picks the best-guess DB per canonical role, and derives a draft `sources.yaml`-shaped mapping (status_field, date_field, assignee_field, workspace_field, etc.) using deterministic rules. Writes the proposal to state/notion-mapping-{date}.yaml for review. No Notion writes.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read-only with respect to Notion.

    line 24
  2. Very specific:

    If ASSIGNMENTS is empty, surface a warning and exit with status no_candidates — there's nothing to map.

    line 118
  3. Very specific:

    Write $OUT in the same shape as config/sources.yaml :: notion_databases, with one block per role that landed a high-confidence assignment:

    line 145
  4. Distinctive:

    No further interaction. The file IS the trust-gate artifact.

    line 203
  5. Very specific:

    Keep $OUT. Don't delete prior mappings.

    line 233
  6. Very specific:

    Run against a workspace containing the four canonical DBs.

    line 237
  7. Very specific:

    If any field deviates, fix the rule in notion_field_map.py — never hand-edit the proposal to make it match (that defeats the calibration).

    line 249

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/notion-map — Property-Role Mapper

You are deriving the field mapping layer of generic onboarding. The probe told us "this DB looks like Master Tasks at 0.92 confidence"; this skill says "and its Status field is called Status, its Due date is called Due, its Assignee is Owner, its Workspace bucket is Category."

Read-only with respect to Notion. Writes a single local YAML proposal.

Constants

REPO_DIR = <your repo root>
STATE    = "${REPO_DIR}/state"
TMP_DIR  = "${STATE}/tmp"
TODAY    = <current date in YYYY-MM-DD format>
RUN_ID   = "nm-${TODAY//-/}-$(date +%H%M)"
OUT      = "${STATE}/notion-mapping-${TODAY}.yaml"

Step 0: Mode Check

MODE=$(scripts/check_mode.sh) || { echo "🛑 Locked"; exit 0; }

Runs in any mode except locked (read-only with respect to Notion). Capture RUN_START_MS for telemetry.

Step 1: Locate Probe Output

This skill consumes /notion-probe's output. Look for the most recent ${STATE}/notion-probe-*.md report and its sibling scored JSON. If none exists or the latest is > 24h old, run /notion-probe first, then continue.

The scorer also needs the normalized schema JSONL. The probe leaves it at ${TMP_DIR}/np-*-schemas.jsonl but cleans up after itself in Step 7. So re-pull schemas now (cheaper than re-running the whole probe):

mkdir -p "$TMP_DIR"
SCHEMA_FILE="${TMP_DIR}/${RUN_ID}-schemas.jsonl"
: > "$SCHEMA_FILE"

NEXT=""
while : ; do
  if [ -z "$NEXT" ]; then
    BODY='{"page_size":100,"filter":{"value":"data_source","property":"object"}}'
  else
    BODY=$(printf '{"page_size":100,"filter":{"value":"data_source","property":"object"},"start_cursor":"%s"}' "$NEXT")
  fi
  RESP=$(curl -s --max-time 30 -X POST https://api.notion.com/v1/search \
    -H "Authorization: Bearer ${NOTION_API_TOKEN}" \
    -H "Notion-Version: 2025-09-03" \
    -H "Content-Type: application/json" \
    -d "$BODY")
  echo "$RESP" | jq -c '.results[] | {
    id, url, last_edited_time,
    title: (.title[0].plain_text // "Untitled"),
    properties: (.properties | to_entries | map({
      name: .key,
      type: .value.type,
      options: ((.value.select.options // .value.status.options // .value.multi_select.options // []) | map(.name)),
      relation_data_source_id: (.value.relation.data_source_id // null)
    }))
  }' >> "$SCHEMA_FILE"
  HAS_MORE=$(echo "$RESP" | jq -r '.has_more')
  NEXT=$(echo "$RESP" | jq -r '.next_cursor // empty')
  [ "$HAS_MORE" = "true" ] || break
done

Step 2: Pick Role Candidates from the Probe Report

For each canonical role, pick the top-confidence candidate at ≥ 0.80 (the probe's "Best guess"). Skip any role with no high-confidence match — those need either template install or manual mapping during /setup.

Parse the probe report (markdown) OR re-score in-process by piping $SCHEMA_FILE through scripts/notion_probe_score.py. The in-process route is more robust:

SCORED="${TMP_DIR}/${RUN_ID}-scored.json"
python scripts/notion_probe_score.py < "$SCHEMA_FILE" > "$SCORED"

# Extract best-guess assignments (highest score per role, gated at 0.80)
ASSIGNMENTS=$(python - <<'PY'
import json, sys
with open("'"$SCORED"'") as f:
    data = json.load(f)
out = []
for role in ["master_tasks", "provider_crm", "activity_log", "meeting_notes"]:
    ranked = sorted(data, key=lambda d: -d["scores"][role]["score"])
    if ranked and ranked[0]["scores"][role]["score"] >= 0.80:
        out.append(f'{role}={ranked[0]["id"]}')
print(" ".join(out))
PY
)

If ASSIGNMENTS is empty, surface a warning and exit with status no_candidates — there's nothing to map.

Step 3: Derive Field Mappings

MAPPING_JSON="${TMP_DIR}/${RUN_ID}-mapping.json"
python scripts/notion_field_map.py --schemas "$SCHEMA_FILE" $ASSIGNMENTS > "$MAPPING_JSON"

notion_field_map.py applies deterministic rules per role:

  • Master Tasks: status/date/assignee/workspace fields + status option buckets (done/in-progress/waiting/not-started) derived from the status or select property's option names against canonical word sets.
  • Provider CRM: stage/last_contact/next_step/workspace fields.
  • Activity Log: type/date/outcome/next_action/workspace fields.
  • Meeting Notes: title/date/summary/attendees/workspace + related-tasks relation (matched against the Master Tasks data source id from Step 2).

For ambiguous select properties (a DB might have both Workspace and Category selects), the scanner prefers more specific keyword matches in priority order: workspace > department > project > category > area. See scripts/notion_field_map.py :: find_prop for the rule.

Step 4: Render YAML Proposal

Write $OUT in the same shape as config/sources.yaml :: notion_databases, with one block per role that landed a high-confidence assignment:

# Notion mapping proposal — {{TODAY}}
# Generated by /notion-map from probe run {{RUN_ID}}.
# Review against your actual workspace before adopting into config/sources.yaml.

notion_databases:
  - name: "Master Tasks"
    data_source: "collection://{master_tasks_ds_id}"
    status_field: "Status"
    status_done_values: ["Done"]
    status_in_progress_values: ["In progress"]
    status_waiting_values: ["Waiting"]
    status_not_started_values: ["Not started"]
    date_field: "Due"
    assignee_field: "Assignee"
    workspace_field: "Workspace"
    # Detected workspace option values — map these to your streams during /setup:
    # ["Stream A", "Stream B", "Stream C", "Other"]
    last_activity_field: "Last Activity"
    updated_field: "Updated"

  - name: "Provider CRM"
    ...

Convert notion_field_map.py's JSON output to YAML inline (use Python yaml.safe_dump, fall back to manual rendering if pyyaml is missing — the project's state/coo_mode.yaml does this manually if needed).

For each role's workspace_values_raw / stage_options_raw / type_values_raw arrays from the mapper, render them as a YAML comment above the field — they are seed data for the upcoming /setup stream-mapping step, not authoritative config values yet.

Step 5: Trust Gate

Print a summary to stdout:

Notion field mapping proposal — {{TODAY}}
Roles mapped: {{N_ROLES}} / 4

  Master Tasks  → "Master Tasks"   (10 fields)
  Provider CRM  → "Provider CRM"   ( 4 fields)
  Activity Log  → "Activity Log"   ( 5 fields)
  Meeting Notes → "Meeting Notes"  ( 6 fields)

Proposal written to: state/notion-mapping-{{TODAY}}.yaml

Next steps:
  - Open the file, review each field name.
  - If anything looks wrong, edit and re-run.
  - When /setup ships, it will read this file as the field-mapping seed.

No further interaction. The file IS the trust-gate artifact. The user reviews on disk and either lives with it or hand-edits.

Step 6: Telemetry

END_MS=$(date +%s%3N 2>/dev/null || python -c "import time; print(int(time.time()*1000))")
DURATION=$((END_MS - RUN_START_MS))

EXTRA=$(python -c "import json; print(json.dumps({
  'n_roles_mapped': $N_ROLES,
  'n_dbs_probed': $N_DBS,
  'mode': '$MODE',
  'out_file': '$OUT'
}))")

scripts/telemetry.sh notion-map "$RUN_ID" "$DURATION" "$STATUS" "$EXTRA"

Status values:

  • ok — ≥ 1 role mapped
  • no_candidates — probe ran but no role cleared 0.80
  • auth_failed / error — as in /notion-probe

Step 7: Cleanup

rm -f "${TMP_DIR}/${RUN_ID}-"*

Keep $OUT. Don't delete prior mappings.

Success Criterion

Run against a workspace containing the four canonical DBs. The proposal at state/notion-mapping-{TODAY}.yaml must match config/sources.yaml :: notion_databases field-for-field for the 4 canonical DBs:

  • Master Tasks: status_field, status_done_values, status_in_progress_values, status_waiting_values, status_not_started_values, date_field, assignee_field, workspace_field, last_activity_field, updated_field
  • Provider CRM: stage_field, last_contact_field, next_step_field, workspace_field
  • Activity Log: type_field, date_field, outcome_field, next_action_field, workspace_field
  • Meeting Notes: title_field, date_field, summary_field, attendees_field, workspace_field, related_tasks_field

If any field deviates, fix the rule in notion_field_map.py — never hand-edit the proposal to make it match (that defeats the calibration).

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.