f-distill
Skill14 starsChanged 20 days agoMITView on GitHub
Promote the highest-value Logbook cards into halt-shaped NEVER/ALWAYS rules: product lessons into the repo's constitution, method lessons into Topping Out's own RULES.md or playbooks. Preview first; write only on the owner's word. Use for /f-distill, "distill the logbook", "pull the best learnings into the rules". Not for capturing one card (f-card), not for merging or retiring cards (f-dream).
In this repository32 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/llms-full.txtllms.txt
- docs/llms.txtllms.txt
- adjudicateSkill
- agent-councilSkill
- converseSkill
- f-archive-searchSkill
- f-archiveSkill
- f-blueprintSkill
- f-boardSkill
- f-cardSkill
- f-catchupSkill
- f-complySkill
- f-contextSkill
- f-dreamSkill
- f-greploopSkill
- f-inspectSkill
- f-night-shiftSkill
- f-plancheckSkill
- f-scheduleSkill
- f-searchSkill
- f-setupSkill
- f-shiftSkill
- f-signoffSkill
- f-sketchSkill
- f-statusSkill
- f-surveySkill
- model-councilSkill
- orca-pi-connectSkill
- pack-reviewSkill
- speakSkill
- xreviewSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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 23
No auto-synced "distillation" section, no HUD, no "run /f-distill to refresh" inside the constitution or the rules.
- Very specific: line 30
Read foreman/logbook/logbook_index.md and foreman/logbook/logbook_links.md, and the current constitution and Topping Out's RULES.md.
- Distinctive: line 46
Prefer CRITICAL/HIGH on ties.
- Distinctive: line 52
Ask: Proceed with these bullets?
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Topping Out: Distill
Score Logbook cards. Preview halt-shaped bullets. Write nothing until the owner approves the exact bullets.
Two destinations
- Product lessons (about this codebase): the repo's constitution, which the repo profile names (
AGENTS.md, orCLAUDE.mdwhere that is the constitution). IfCLAUDE.mdis only@AGENTS.md, writeAGENTS.mdand leaveCLAUDE.mdalone. - Method lessons (about how Topping Out works): a proposed change to
RULES.mdin the Topping Out package (the foreman rules) orForeman-Playbook.md/Builder-Playbook.md. Show it as a diff; the package is changed only on the owner's word.
Halt
- Preview first. No write until the owner approves the exact bullets.
- No auto-synced "distillation" section, no HUD, no "run /f-distill to refresh" inside the constitution or the rules.
- Preview mode (from
/f-signoffStep 3.2a): steps 1 to 3 only, written to<day folder>/DISTILL-PREVIEW.mdwith its date and the cards it read; it asks nothing and writes nowhere else. The owner applies it later with/f-distill, which re-scores before showing the bullets (cards may have changed). - A card never overrides a rule by itself: a bullet that contradicts an existing rule is shown as a proposed change to that rule, with both texts.
- Not
f-card. Notf-dream.
1. Gather
Read foreman/logbook/logbook_index.md and foreman/logbook/logbook_links.md, and the current constitution and Topping Out's RULES.md.
2. Score (a ranking heuristic)
The multipliers can push a score past 100: it orders candidates, it isn't a probability or a percentage. Inputs a card type doesn't have count as 0 (BUILD has no confidence: use 0 and rely on relationships, recency and impact).
confidence (0–100) × 0.25
relationship_count (cap 10) × 2.5
recency (<14d=100, <30=80, <90=60, else 40) × 0.15
actionability × 0.15
impact (CRITICAL=100, HIGH=75, MEDIUM=50, LOW=25) × 0.20
× 1.5 anti-pattern · × 1.4 CRITICAL · × 1.3 cross-cluster · × 1.2 HIGH · × 0.7 product-specificPrefer CRITICAL/HIGH on ties. One card, one bucket. Caps: 5 anti-patterns, 5 insights, 3 methods, 3 recent (last 14 days), 3 builds (last 30 days).
3. Preview
Halt-shaped bullets only (one or two lines, NEVER/ALWAYS), each tagged with its destination (the repo's constitution, or Topping Out's RULES.md and playbooks) and its source card ids. Also list HIGH/CRITICAL cards that missed the cutoff only on recency or relationship decay; don't auto-promote them.
Ask: Proceed with these bullets?
4. Write (after the owner's go)
Merge the approved bullets into the existing NEVER/ALWAYS (or matching) sections of each destination. No new "Distillation" heading. Record in each source card's front matter distilled: <date> -> <destination>.
5. Report
Cards analysed, bullets written per destination, the aged-out table (or empty).
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.

