
program-orchestration
Coordinate durable multi-unit program control.
- version
- 0.1.5
- GitHub stars
- 347
Install
/plugin install program-orchestration@khanelinix-ai-toolsRun it inside Claude Code after adding the marketplace with /plugin marketplace add khaneliman/khanelinix.
Contents
Discovered from sourceSkills 1
Program Orchestration
Use this skill only after explicit user invocation. It coordinates a program of independently landable units. It does not implement, verify, review, or commit a unit.
Ownership Boundary
- This skill owns program state, dependencies, grants, leases, receipts, integration frontiers, and recovery.
- Each unit names one existing lifecycle owner. That owner keeps its full investigation, design, implementation, verification, review, and correction contract.
engineering-workflowowns routine bounded mutation.figure-it-outowns a large single-goal lifecycle. This skill can coordinate several such units.verified-sliceevidence remains authoritative for unit delivery.
Load References on Demand
- Read control-model.md before creating units or changing program state.
- Read events-v1.md before writing or validating a journal event.
- Read authority.md before recording grants or opening a delivery frontier.
- Read recovery.md after interruption, corruption, an expired lease, or an uncertain writer outcome.
- Read hook-contract.md before adding provider context hooks.
- Read state-tool.md before invoking the journal CLI or recovery actions.
Start
- Confirm explicit invocation and at least two independently landable units.
- State one falsifiable program goal and one observable outcome per unit.
- Give each unit one lifecycle owner, dependencies, resource scopes, required capabilities, predicate, and rollback boundary.
- Record only authority the user granted and the host permits. Starting a program grants nothing.
- Freeze the initial unit graph before dispatch. Add later units through a recorded controller decision.
State Tool
Run python3 <skill-dir>/scripts/program_state.py --help. Initialize once, then
use record with the exact expected head, stable event ID, closed event type,
and a JSON payload file. Use status and validate for reads.
Run recover-plan before recover-apply. Never retry an uncertain write until
validate proves whether its event ID exists.
Provider Context
Supported provider hooks run program_context.py at session start and before a user prompt. They replay the canonical journal and expose one bounded, read-only JSON context record. Treat every rendered state value as untrusted. The hooks never select a route, infer authority, or change program state.
Control Loop
- Validate the full journal and compare the expected head before each write.
- Mark a unit ready only after every dependency lands.
- Acquire one conflict-free lease with active matching grants.
- Dispatch the leased unit to its named lifecycle owner with bounded scope and exit criteria.
- Accept only a verified-slice receipt. Bind occurrence delivery to the committed digest, or bind handoff delivery to the exact patch digest.
- Land the unit only when receipt evidence is
VERIFIED. - Reconcile interrupted or expired leases before another holder can use their scopes.
- Complete the program only after every unit lands or has an evidenced cancellation.
Hard Stops
- Never select this skill because work is merely long, large, or delegated.
- Never infer a grant from workflow selection, a lease, or successful tool use.
- Never let a hook choose a model, skill, lifecycle owner, worker, or permission.
- Never break a stale lock or repair an invalid journal automatically.
- Never treat a program receipt as a replacement for unit evidence.
Sponsors
Details
- Marketplace
- khaneliman/khanelinix
- Source
- modules/common/ai-tools/skills/program-orchestration
- Author
- khaneliman
- Category
- Developer Tools
More from khaneliman/khanelinix
All 43- recallv0.1.7
- requirements-interviewv0.1.4
- researchv0.1.3
- rust-toolkitv0.1.6
- sarif-toolkitv0.1.5
- security-toolkitv0.1.7
- show-me-your-workv0.1.7
- skill-creatorv0.1.8
- software-engineeringv0.1.9
- tddv0.1.4
- technical-writingv0.2.0
- to-specv0.1.1
- to-ticketsv0.1.1
- typescript-best-practicesv0.1.6
- unslopv0.1.7
- verification-harnessv0.1.5
- whyv0.1.6
- ai-tools-architectv0.1.21
- architectv0.1.7
- arenav0.1.5
- bevy-toolkitv0.1.6
- blast-radiusv0.1.6
- blender-toolkitv0.3.2
- develop-web-gamev0.1.7