f-dream
Skill14 starsChanged 20 days agoMITView on GitHub
Reflect across completed jobs and the Logbook to propose verified deduplication, supersession, and recurring-pattern updates. Use for /f-dream, "reflect across the jobs", "are we repeating mistakes", "clean up contradictions in the logbook", or when entries may contradict or repeat. Reconciles content: not for promoting knowledge upward (f-distill), nor for rotating by count (f-archive).
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-distillSkill
- 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
7 rules. About 0 are common, 6 are distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Default is a read-only dry run over the last ten completed jobs (foreman/completed-jobs/*/JOB_*.md).
- Distinctive: line 26
Read the selected jobs, active cards, Logbook index, relationships, and earlier dream proposals.
- Distinctive: line 28
Verify every candidate against its cited source.
- Distinctive: line 38
update canonical content and relationships
- Distinctive: line 39
verify references and index
- Distinctive: line 40
only then remove the retired active copy
- Distinctive: line 42
Stop on ambiguous identity, missing provenance, copy-verify failure, or relationship inconsistency.
≈ “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: Logbook dream
Default is a read-only dry run over the last ten completed jobs (foreman/completed-jobs/*/JOB_*.md). --since and --jobs bound the corpus. --save-proposal <id> requires explicit write authority. --apply --proposal <id> applies that saved proposal only, no most-recent default.
Automatic mode (from /f-signoff Step 3.2a only)
The close is the mutation authority, for this one proposal. Run the dry run over the jobs since the last dream line in foreman/logbook/upkeep.md, save it (--save-proposal <date>-J<n>), then apply only the items that pass every guard:
- a merge of two cards with the same lesson in the same context, or a supersession where the newer card cites the evidence that overturns the older one, at the proposal's highest confidence;
- never a CRITICAL card, a contradiction that needs judgement, or an item whose provenance can't be verified: those stay in the saved proposal, listed for the owner;
- a new recurring-pattern card only with three or more cited jobs.
The Apply steps below run unchanged (copy, verify, update, verify, then remove the active copy, with a manifest that reverses every change), so nothing is ever lost: a retired card sits in the proposal's retired area. Report the proposal id, every change, and every path written or removed (the cards, the retired copies, the proposal and manifest) to
/f-signoff, and return; it never stops the job. Never stage, commit or push; the close's records commit carries it.
Dry-run
- Read the selected jobs, active cards, Logbook index, relationships, and earlier dream proposals.
- Root does three non-mutating passes with fresh notes: duplication, contradiction/supersession, recurring mistakes. Do not dispatch unnamed roles. Do not apply during these passes.
- Verify every candidate against its cited source. Same lesson in different contexts is not a duplicate. Context-dependent advice is not a contradiction.
- Numbered proposal: evidence, target cards, relationship changes, proposed bodies, confidence, rollback, unresolved. Cap 10 changes. Never auto-retire CRITICAL.
- Return the proposal and suggested
foreman/logbook/archived/dreams/<id>/path. Dry-run writes nothing. Persist only via--save-proposal.
Apply
Existing proposal + explicit mutation authority. Accepted items only. Retire/merge/supersede:
- copy original to the proposal's retired area
- verify the copy
- update canonical content and relationships
- verify references and index
- only then remove the retired active copy
Stop on ambiguous identity, missing provenance, copy-verify failure, or relationship inconsistency. Keep an applied manifest that can reverse every change.
Output
Proposal ID, corpus size, merges, supersessions, new patterns, skipped/flagged, files changed if apply, validation evidence, follow-up. Follow-up distill writes the constitution (the repo profile names it: AGENTS.md, or CLAUDE.md where that is the constitution) only, on the owner's word. Never stage, commit, or push.
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.

