update-playbook
Skill0 starsChanged 2 months agoMITView on GitHub
Use when the user wants to analyze content performance, update a Content Playbook (LinkedIn, X, or Substack), run the Growth Agent, or explicitly invokes /update-playbook. Derives evidence-backed rules from Analytics/ data, scoped to one platform at a time, and updates that platform's playbook file. Refuses to invent a pattern when the sample size is too small.
In this repository38 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- audit-draftSkill
- check-plagiarismSkill
- critique-draftSkill
- critique-draft-substack-articleSkill
- critique-draft-substack-noteSkill
- critique-draft-xSkill
- draft-commentSkill
- draft-replySkill
- extract-hookSkill
- generate-ideasSkill
- generate-visualSkill
- generate-visual-substack-articleSkill
- generate-weekSkill
- generate-week-substackSkill
- generate-week-xSkill
- humanize-draftSkill
- interviewerSkill
- monitor-engagementSkill
- optimize-profileSkill
- plan-advocacySkill
- plan-weekSkill
- plan-week-substack-articleSkill
- plan-week-substack-noteSkill
- plan-week-xSkill
- publish-substackSkill
- pull-analyticsSkill
- pull-analytics-xSkill
- repurpose-postSkill
- research-topicSkill
- review-draftsSkill
- schedule-approvedSkill
- schedule-approved-xSkill
- write-draftSkill
- write-draft-substack-articleSkill
- write-draft-substack-noteSkill
- write-draft-xSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 22
Never mix platforms' evidence into one rule — a pattern proven on LinkedIn is not assumed to transfer to X or Substack, and vice versa.
- Distinctive: line 63
If a previously-recorded rule is contradicted by newer data, update or remove it rather than leaving stale, wrong guidance in place — note …
- Distinctive: line 82
Never add a rule backed by fewer than 3 posts.
- Very specific: line 83
Never treat a missing/null metric (see /pull-analytics) as a zero when computing averages — exclude it, don't count it against the post.
- Distinctive: line 85
Never silently overwrite a rule's evidence trail — corrections are logged, not erased.
≈ “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
Update Playbook (Growth Agent)
Turns real Analytics data into structured, evidence-cited rules in a platform-scoped playbook file, closing the loop described in REQUIREMENTS.md §17-18. This is the only skill that writes to any playbook file. The algorithm below is identical regardless of platform — only which files it reads/writes changes.
Arguments
- platform (optional) —
linkedin(default),x, orsubstack. Determines which Analytics/Published-Posts notes to gather (filtered by theirplatformfield) and which playbook file to update:Content-Learnings/playbook.md(linkedin),playbook-x.md(x), orplaybook-substack.md(substack — update the Articles and Notes sections independently within that one file, per its own header note). Never mix platforms' evidence into one rule — a pattern proven on LinkedIn is not assumed to transfer to X or Substack, and vice versa.
Minimum evidence bar
Do not add or change a rule unless at least 3 published posts support it, and always cite their ids and the sample size next to the rule (e.g. "n=4"). With fewer than 3 published posts total, run the report step only and say plainly there isn't enough data yet — do not lower the bar to produce something to write.
Process
1. Gather data
For every note in Published-Posts/ whose platform field matches this
run's argument (default linkedin), read its metadata (category, format,
length, hook_style, hashtags, posting day/time) and its linked
Analytics/ snapshot history (latest values, and trend across snapshots
if there are several). For substack, treat Articles and Notes as separate
populations throughout steps 1-4 — don't average a 2,000-word article's
metrics against a one-line Note's.
2. Look for patterns, only where the evidence bar is met
Compare average engagementRate (and other metrics where meaningful)
across groupings: by category, by content_type, by hook_style, by length
bucket (e.g. <1000 / 1000-2000 / >2000 chars), by posting weekday, by
hashtag set. A pattern is reportable only if:
- at least 3 posts sit in each side of the comparison, and
- the difference is large enough to plausibly not be noise (use judgment, but a marginal 5% difference on n=3 is not a rule — say so).
3. Update the Playbook
For each pattern meeting the bar, add or update a row in this run's target
playbook file (playbook.md / playbook-x.md / playbook-substack.md,
per the platform argument) — its Best-Performing Patterns or
Anti-Patterns table (for substack, the Articles or Notes variant of
those tables specifically): the rule in plain language, the evidence (post
ids), a confidence level (low for n=3-4, medium for n=5-9, high for
n=10+), and today's date. Bump version and last_updated in that file's
frontmatter — never another platform's file.
If a previously-recorded rule is contradicted by newer data, update or remove it rather than leaving stale, wrong guidance in place — note the reversal explicitly rather than silently deleting the old row.
4. Topic Fatigue Watch
Flag categories/topics posted 3+ times in the last ~4 weeks with flat or declining engagement across those posts, in the Topic Fatigue Watch section.
Report back
State plainly: how many published posts exist, how many met the evidence bar for any comparison, which rules were added/updated/removed, and — just as important — which comparisons were attempted but skipped for insufficient data. A playbook that honestly says "not enough data yet" is more useful than one padded with weak inferences.
Hard rules
- Never add a rule backed by fewer than 3 posts.
- Never treat a missing/null metric (see
/pull-analytics) as a zero when computing averages — exclude it, don't count it against the post. - Never silently overwrite a rule's evidence trail — corrections are logged, not erased.
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.

