agentleFS
koloskuskoloskus/jotbook/skills/jotbook-review/SKILL.md

jotbook-review

Skill0 starsChanged 5 months agoMITView on GitHub

Review the jot backlog and dispatch decisions in place — drop, merge, tweak, pencil, or ink. Surfaces overlap between related jots and flags any with a pencil already in progress.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 6 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read .claude/jotbook.local.md if present.

    line 12
  2. Very specific:

    If jots_dir doesn't exist or contains no jots, report that in one line and stop.

    line 19
  3. Very specific:

    Pencil status: check whether <pencils_dir>/<slug>.md or <pencils_dir>/<slug>.html exists.

    line 32
  4. Distinctive:

    When a jot has a pencil, prefer recommending the user review the pencil (which has more material to judge from) rather than re-deciding …

    line 77
  5. Distinctive:

    If the user marked more than one jot for ink and/or pencil in a single decision turn, ask once before dispatching:

    line 98
  6. Distinctive:

    Default to checkpoint if the user doesn't pick.

    line 102
  7. Distinctive:

    After all actions complete, give a one-line summary:

    line 104
  8. Distinctive:

    Don't expand jot files into paragraphs during review — they stay lightweight until inked or penciled.

    line 114
  9. Distinctive:

    Don't auto-drop "old" jots just because they're old.

    line 115
  10. Very specific:

    Don't propose creating a new jot that wasn't already staged.

    line 116
  11. Very specific:

    Don't read every referenced source file during review — the jot text is enough to make keep/drop decisions.

    line 117
  12. Very specific:

    Don't invoke jotbook-ink or jotbook-pencil without an explicit user mark.

    line 118
  13. Very specific:

    Don't quietly re-decide for a jot that already has a pencil.

    line 119

≈ “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.

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

Review the jot backlog

The user is reviewing the jot list to decide which subjects still feel worth inking into a long-form entry — or worth penciling first as a provisional draft before committing. Your job is to surface and recommend, not to delete or merge unilaterally. Every destructive action needs explicit confirmation.

Resolve configuration

Read .claude/jotbook.local.md if present. From its frontmatter, extract:

fielddefaultmeaning
jots_dirdocs/jotbook/_jots/where staged jots live
pencils_dirdocs/jotbook/_pencils/where pencils live — checked to flag jots that already have one

If jots_dir doesn't exist or contains no jots, report that in one line and stop.

Procedure

1. Inventory

List every *.md file under jots_dir (excluding any README.md). For each, read the file and extract:

  • Slug (filename without YYYY-MM-DD- prefix and .md suffix)
  • staged: date
  • Age in days (today − staged date)
  • Description line
  • Files line (if present)
  • Pencil status: check whether <pencils_dir>/<slug>.md or <pencils_dir>/<slug>.html exists. If so, note "has pencil from <drafted-date>" (and "revised <date>" if the pencil's frontmatter has a revised: field).

Sort by age, oldest first.

2. Detect overlap

Group jots that look topically related. Signals:

  • Overlapping file references on the Files: line
  • Shared keywords or domain terms in descriptions
  • Same subsystem or topic cluster

Note these as suggested merges, not forced ones.

3. Present

Output a compact, scannable summary. Don't over-format — the user will be making decisions from this view. Suggested shape:

Jots (N total):

  ▸ 2026-05-12  cache-eviction-policy      (10d)  [has pencil from 2026-05-18]
    How the LRU cache evicts on hot keys.
    Files: src/cache/lru.ts, src/cache/metrics.ts

  ▸ 2026-05-15  feature-flag-rollout       (7d)
    How the gradual rollout schedule advances cohorts.

  ▸ 2026-05-15  feature-flag-telemetry     (7d)
    How flag-evaluation events flow to analytics.

Suggested consolidations:
  • feature-flag-rollout + feature-flag-telemetry → both touch the flag
    subsystem and could share an entry; consider merging into
    "feature-flag-system".

Recommendations:
  • cache-eviction-policy: pencil already exists — consider running
    `/jot review pencils` to evaluate it instead of re-deciding from
    the jot alone.
  • feature-flag-system (merged): plausible candidate for pencil if
    you want to see the long-form before committing.
  • drop none unless the user has lost interest in any.

When a jot has a pencil, prefer recommending the user review the pencil (which has more material to judge from) rather than re-deciding from the jot alone.

Lead with the inventory; put your recommendations at the end. Keep the whole thing scannable — short lines, no walls of prose.

4. Solicit decisions

Ask the user, in plain language, what to do. Don't present a numbered menu — let them respond in free form. Every action surfaced here is acted on directly in step 5 — ink and pencil dispatch into their respective skills inside this same flow; no follow-up slash command is required. Mention all six available actions so the user knows the surface. Example:

Decisions? You can say things like "drop the feature-flag pair", "merge X and Y into Z", "tweak cache-eviction to focus on hot keys", "ink cache-eviction-policy", "pencil feature-flag-system" (draft first, decide later), or "keep everything for now".

5. Apply

Apply structural actions (drop / merge / tweak) first, then dispatch ink and pencil actions. For each decided action:

  • Keep — no-op.
  • Drop <slug> — delete the jot file. If more than one is being dropped, list what you're about to delete first and confirm.
  • Merge <slug-a> + <slug-b> → <new-slug> — create a new jot file combining the descriptions (still one or two lines total); union the Files: lines; delete the merged-in originals. Date the merged jot today.
  • Tweak <slug> — <new framing> — the inferred subject was close but the framing needs a nudge. Edit the jot's description line in place to reflect the user's directive. Optionally update the Files: line if the new framing implicates different sources. This stays lightweight — apply the change in one pass, do not enter a back-and-forth discussion. If the directive is genuinely ambiguous, ask exactly one clarifying question, then apply and move on.
  • Ink <slug> — leave the jot file in place and hand off to jotbook-ink on that slug. The ink skill takes it from there (including its own pencil-promotion question if a pencil already exists for the slug).
  • Pencil <slug> (optionally with --html) — leave the jot file in place and hand off to jotbook-pencil on that slug, forwarding --html if the user specified it.

If the user marked more than one jot for ink and/or pencil in a single decision turn, ask once before dispatching:

"You marked <N> jots for ink/pencil. Checkpoint between each (so you can redirect partway), or batch through them in order?"

Default to checkpoint if the user doesn't pick. Either way, dispatch them in the order the user named them.

After all actions complete, give a one-line summary:

Dropped N, merged M into K, tweaked T, kept P, inked: <list>, penciled: <list>.

Omit the trailing fields when their lists are empty.

What NOT to do

  • Don't expand jot files into paragraphs during review — they stay lightweight until inked or penciled.
  • Don't auto-drop "old" jots just because they're old. Age is information, not a verdict.
  • Don't propose creating a new jot that wasn't already staged. Use jotbook-stage for that.
  • Don't read every referenced source file during review — the jot text is enough to make keep/drop decisions. Save deep reading for jotbook-ink or jotbook-pencil.
  • Don't invoke jotbook-ink or jotbook-pencil without an explicit user mark. The user's free-form reply IS the mark — once they've said "ink X" or "pencil Y", dispatch directly; do not ask them to re-issue the request as a slash command.
  • Don't quietly re-decide for a jot that already has a pencil. Point the user at /jot review pencils so they can judge from the long form.

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.