progress-digest - SKILL.md Agent Skill

name: progress-digest description: Generate weekly learning progress digests from analytics snapshots and recent trends. version: 1.0.0

Skill: Progress Digest

Purpose

Weekly learning progress digest. Read the last 7 days of analytics snapshots across all active mind maps, compute trends, highlight achievements, flag struggling areas, and deliver a structured summary to the user via their preferred notification channel (email by default for digests).

When to Use

Use this skill when:

The weekly-progress-digest scheduled task fires (cron: 0 9 * * 0, Sundays at 09:00)
The user requests "show me my learning progress" or similar

Digest Composition Protocol

Step 1: Gather Data

Call these tools to collect the analytics data:

analytics_get_cross_topic() — comparative stats across all active mind maps. Returns: topics (list with mastery_pct, retention_rate_7d, velocity per map), portfolio_mastery.
For each mind_map_id in the cross-topic result, call: analytics_get_trend(mind_map_id, days=7) — last 7 daily snapshots, ordered ascending. The metrics dict per snapshot includes: mastered_nodes, total_nodes, mastery_pct, retention_rate_7d, velocity_nodes_per_week, struggling_nodes, estimated_completion_days.
analytics_get_snapshot(mind_map_id) — latest snapshot for each map (if no 7-day trend). Call this at most once per map. It always returns a status field: status="ok" carries the snapshot (with snapshot_date as an ISO-8601 string); status="not_found" means none exists. On status="not_found", do not retry the same call — fall back to mastery_get_map_summary() and note that weekly analytics are not available yet.
memory_search(query="learning preferences") — check if the user prefers verbose vs. concise summaries, or has expressed delivery preferences.

Step 2: Identify Trends

For each topic with at least 2 snapshots in the 7-day window, compute trends:

Positive signals to highlight:

Newly mastered concepts: mastered_nodes increased since last week's snapshot
Retention rate improving: retention_rate_7d trended upward over the snapshots
Velocity accelerating: velocity_nodes_per_week increased vs. the prior week

Concern signals to flag:

retention_rate_7d < 0.60 — retention is low; flag for review focus
len(struggling_nodes) >= 3 — multiple concepts struggling; consider curriculum re-planning
sessions_this_period == 0 over the 7-day window — topic is idle (no study sessions)
estimated_completion_days is very high or None — progress has stalled

Portfolio-level insight: Compare portfolio_mastery to available historical data to show overall momentum.

Step 3: Structure the Digest

Format the digest as follows (keep it concise — readable on mobile):

Weekly Learning Progress — [Date]

Portfolio: [portfolio_mastery]% overall mastery across [N] active topics

[Topic 1 — title]
  Mastered: [mastered_nodes]/[total_nodes] ([mastery_pct]%)  [+N this week if improved]
  Retention: [retention_rate_7d]%  [↑ or ↓ vs last week if trend available]
  Velocity: [velocity] concepts/week
  [Flag if struggling: "⚠ [N] concepts need review attention"]
  [Flag if idle: "💤 No sessions this week"]
  Estimated completion: ~[estimated_completion_days] days

[Topic 2 — title]
  ...

Highlights:
  - [Achievement 1, e.g. "Mastered 5 Python concepts this week"]
  - [Achievement 2, e.g. "Retention rate improved from 68% to 84%"]

Watch areas:
  - [Struggle flag, e.g. "Python closures: low retention (42%) — review sessions recommended"]
  - [Idle flag, e.g. "Calculus: no sessions in 7 days"]

Estimated completions:
  - [Topic]: ~[N] days at current pace

Tone: Keep it encouraging. Acknowledge effort, not just outcomes. Frame struggle areas as opportunities, not failures.

Step 4: Deliver via Email

For the weekly digest, use email as the primary delivery channel:

notify(
    channel="email",
    intent="send",
    subject="Your weekly learning progress — [Date]",
    message=<formatted_digest>,
    request_context=<session_request_context>
)

If the user has a preference stored in memory (from memory_search()), respect it:

If preference is Telegram or another channel, use that instead.
If no preference stored, default to email for digests (it's the appropriate format for longer-form weekly summaries).

Step 5: Trigger Curriculum Re-planning (if needed)

For any topic where retention_rate_7d < 0.60 or len(struggling_nodes) >= 3:

Call:

curriculum_replan(
    mind_map_id=<map_id>,
    reason="analytics feedback: low retention / multiple struggling concepts"
)

Note in the digest (before delivery): "I've adjusted the learning path for [topic] to prioritize your struggling concepts."

Step 6: Store Digest Summary in Memory (Optional)

If the digest contains a notable milestone or significant pattern shift, record it:

memory_store_fact(
    subject=<topic_label>,
    predicate="study_pattern",
    content=<brief note about trend, e.g. "mastery accelerating — 5 concepts/week pace">,
    permanence="standard",
    importance=5.0,
    tags=[<topic_tag>, "progress", "weekly-digest"],
    entity_id=<map_entity_id>
)

Note: map_entity_id is the entity_id of the mind map's root concept node. Neither analytics_get_snapshot() nor mind_map_get() return a map-level entity_id directly — call mind_map_node_list(mind_map_id=<id>) and use the entity_id of the root node (lowest depth, or the node with no incoming prerequisite edges). If no clear root can be determined, omit entity_id — it is not required for user/topic-level study pattern facts.

Exit Criteria

analytics_get_cross_topic() called to get portfolio-level data
analytics_get_trend() called for each active mind map
Trends computed: velocity, retention, newly mastered concepts, struggling nodes
Digest formatted and delivered via notify(channel="email", intent="send", subject=..., ...)
Curriculum re-planning triggered (via curriculum_replan()) for any topic with low retention or 3+ struggling nodes
Session exits after delivery — no teaching or review in this session