lN-lifecycle-doctrine
Skill3 starsChanged 3 months agoMITView on GitHub
Shared doctrine reference for child-session lifecycle management at every supervision layer. Defines the generic child state machine (spawned→working→complete→retired, context-pressure axis, edge states), per-state entry/act/exit rules, complete-check guards, teardown discipline, idle-fleet handling, inter-actor inbox-on-nudge delivery (ac_message_send/ac_inbox_query/ac_ack; send-keys = wake nudge only), and per-layer executor bindings for l1-supervise (child=L0) and l2-supervise (child=L1). This skill has no operations — it is reference content read by l1-supervise and l2-supervise at tick time. Never inline this doctrine into the supervise skills themselves.
In this repository41 other files it gives its agents.
- cadence-goalsSkill
- ci-feedback-loopSkill
- compoundSkill
- decision-rights-doctrineSkill
- diataxisSkill
- dispatch-gateSkill
- drift-checkSkill
- family-checkSkill
- fix-then-re-review-ladderSkill
- github-projectsSkill
- gitSkill
- goal-treeSkill
- handoffSkill
- improvement-loopSkill
- kb-captureSkill
- kb-compileSkill
- kb-coreSkill
- kb-lessons-bridgeSkill
- kb-lintSkill
- l1-supervisorSkill
- l2-supervisorSkill
- lessons-learnedSkill
- lN-review-doctrineSkill
- medical-recordsSkill
- operator-interview-doctrineSkill
- operator-interviewSkill
- operator-reviewSkill
- planning-workflowSkill
- pr-review-fanoutSkill
- pull-taskSkill
- ralph-wiggumSkill
- reviewSkill
- root-cause-depthSkill
- scan-documentSkill
- self-improvementSkill
- self-verifySkill
- skillifySkill
- task-workflowSkill
- testing-without-mocksSkill
- vault-cloudSkill
- verify-inherited-premiseSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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 48
If a rule in references/ is wrong for the current iteration,
≈ “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
lN-lifecycle-doctrine — shared child-session lifecycle doctrine
This skill is a doctrine-only reference. It defines child-session
lifecycle management generically — applicable at any supervision
layer in a multi-layer agent tree (L2 supervises L1, L1 supervises
L0). The actual supervision skills (l1-supervise, l2-supervise)
are thin executors that load this doctrine at tick time and apply it
at their layer.
Source of truth
| File | Purpose |
references/lifecycle.md | Generic state machine, per-state rules, complete check, teardown guards, idle-fleet handling (incl. confirmed-idle auto-stop + idle-and-blocked), context-pressure axis, inter-actor inbox-on-nudge delivery, interactive-menu answering, platform constraints + scoped agent-team pattern, and per-layer executor bindings. |
Invariants
Recursive ownership. Each supervisor owns the lifecycle of its direct children — L2 owns L1, L1 owns L0. No layer auto-stops its children; the parent always decides. A supervisor MAY auto-stop its own tick loop once its fleet is confirmed idle.
Doctrine-only, no inlining. This skill has no operations.
l1-supervise (child=L0) and l2-supervise (child=L1) Read
references/lifecycle.md at tick time and never inline its content.
No-nesting + durability is the real platform limit — not "a layer is always a session." Substrate (session / subagent / Workflow / agent team) is chosen per layer within that limit; only the durable multi-session tree must be separate sessions.
Termination is the supervisor retiring the child, gated on all
teardown guards (never-teardown-own-session, PR merged/abandoned,
no-unpushed, no-dirty-tracked, a CLEAN l{N-1}-review marker on the
merged PR — complete-check condition 4, and — for objective-scoped
children — an objective-accept marker).
When this doctrine is wrong
If a rule in references/ is wrong for the current iteration,
fix it here via PR against pfeff/claude-skills. Do not patch
the rule inside l1-supervise or l2-supervise — that is the
doctrine-drift failure mode this split exists to prevent.
See also
lN-review-doctrine— the parallel review doctrine; lifecycle doctrine'spr-open → fixing → mergedstates reference the review-cycle states defined there.goal-treereferences/layer-model.md— canonical "what can be a layer" platform constraint and tree-depth → layer mapping.l1-superviseskill — executes this doctrine at N=1 (child=L0).l2-superviseskill — executes this doctrine at N=2 (child=L1).
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.