skill-publish-log · Publish Log
A publish-record management skill: it logs, queries, and aggregates every social content publish (platform, title, link, time, initial metrics) as the data base for review and attribution.
Three modes: record (writes a publish event with auto-incrementing ID), query (filters by time / platform / keyword / account, returned as a table), and stats (aggregates counts and engagement by platform / type / month). All storage operations run deterministically via script; it stores publish events only, not follower time series.
Example invocation: "Log what I just published: Xiaohongshu, '5 Underrated AI Tools'."
Full brief
Positioning
skill-publish-log is the publish-event base of the attribution chain: the single authoritative store outputs/_analytics/publish-log.json. It records "what was published" — no content analysis, no follower time series — for upstream review and attribution skills to consume.
Core capabilities
- Record mode: writes a publish event (platform and title required; link, time, type, initial metrics, tags, source skill optional), auto-incrementing ID with atomic write-back, returns a record summary with its number.
- Query mode: filters by time range, platform, keyword, or account Profile, or fetches the latest entry — returned as a table.
- Stats mode: aggregates publish counts, total engagement, and average composite score by platform / type / month; low-sample warnings are passed through verbatim.
- Deterministic execution: reads, writes, ID increments, filtering, and aggregation are all done by
scripts/log.py; the LLM only parses intent, assembles parameters, and interprets the returned JSON. The data file is auto-initialized if missing. - Profile awareness: with a Profile, platform and account fields auto-fill and stats/queries can group by account; without one, the platform must be explicitly specified.
Workflow
- Identify the operation mode: record / query / stats
- In record mode, ask for missing info first (platform and title are the minimum)
- Call
scripts/log.py(record / query / stat) - Interpret the returned JSON into a human-readable summary, table, or stats report
Inputs & outputs
| Input | Notes |
|---|---|
| Record mode | Platform + title required; link, time, type, initial metrics, tags, notes optional |
| Query mode | Time range / platform / keyword / account / latest entry |
| Stats mode | Aggregation dimension (platform / type / month) + time range |
Output: record summary (with number), matching records as a table, structured statistics.
Boundaries with adjacent skills
| Base | Stores | Maintained by |
|---|---|---|
| skill-publish-log (this) | Publish events: platform / title / link / time / initial metrics / tags / source skill | This skill |
| skill-data-tracker | Follower/engagement time-series snapshots (snapshots/) |
skill-data-tracker |
Consumers (read-only): skill-publish-analytics (timing / tag / type / growth attribution), skill-social-performance-review (monthly review).
Fit
- Logging each publish as it happens, building a searchable publish ledger
- Month-end review: how many posts, per-platform split, engagement rankings
- The data base for attribution analysis feeding upstream analytics skills
Before you start
- No API key; the data file auto-initializes when absent.
- Record mode needs at least the platform and title — both are required.
skill-publish-log is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.