القائمة
Use when initializing soe in a project or when asked about the .soe/ layout, config, or the durable-vs-ephemeral commit policy. Explains what /setup scaffolds and how state is committed vs ignored.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
The Evaluate-Loop coordinator for soe. Runs the BRAINSTORM → PLAN → EVALUATE_PLAN → EXECUTE → EVALUATE_EXEC → (FIX↺ | COMPLETE) state machine for a track: dispatches leaf agents, dispatches workers into isolated worktrees and applies their validated returns serially, and is the SOLE writer of .soe/tracks/{id}/state.json. Resumes crash-safely from committed state and bounds its fix/plan loops. Use when: 'run the loop', 'orchestrate', 'run the track', 'evaluate-loop', 'drive the track to completion'.
Session-model-led multi-model tiering. Use in ANY conversation (ambient) or in the pipeline to decide which model tier does which slice of work, and to delegate to the right model-pinned agent (strategist/deep-reasoner/fast-worker). The session model you selected IS the orchestrator; it self-selects its topology from its own model identity.
Use when the orchestrator needs to delegate an implementation task to an isolated worker — dispatches each worker as a Task-tool subagent in its own git worktree, awaits its return as the completion signal, and applies results serially as the sole state.json writer behind a validated context firewall
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
Use when creating new skills, editing existing skills, or verifying skills work before deployment
Use at run start to discover what reviewers/analyzers/generators any installed plugin provides and route work by ROLE. Builds a role→provider map via lib/capability-scan.js so the loop prefers the best-matching installed specialist (a Go reviewer, a Flutter reviewer, AgentShield, ...) and falls back to soe-core's generic (soe:code-reviewer, soe:security-reviewer, soe:architect, ...) when none is installed. soe-core is self-sufficient and NEVER hard-depends on packs.
| name | soe-setup |
| description | Use when initializing soe in a project or when asked about the .soe/ layout, config, or the durable-vs-ephemeral commit policy. Explains what /setup scaffolds and how state is committed vs ignored. |
soe keeps its orchestration bookkeeping in a per-project .soe/ directory,
scaffolded by the /setup command (backed by the real, tested lib/setup.js
helper). This skill describes that layout, when/how setup runs, and the
durable-vs-ephemeral commit policy that keeps cross-session memory shareable
without git noise.
Run /setup once per project, before the first /go. It is idempotent and
resumable: re-running never clobbers an existing .soe/config.json (your edits
are preserved) and never duplicates the managed .gitignore block.
.soe/setup_state.json ({ "last_step": ... }) records progress so an
interrupted setup can resume.
Internally /setup calls runSetup(projectDir, { model }), which reuses the
shared applyGitignore helper for .gitignore — the gitignore logic is NOT
reinvented per command.
.soe/ layoutdocs/plans/ human design docs + plans (committed, single source of truth)
.soe/ machine orchestration bookkeeping (hidden)
├── config.json mode + thresholds + model defaults
├── setup_state.json { last_step } resume marker
└── tracks/{id}/ per-track state (created later by /go)
├── state.json AUTHORITATIVE execution state (atomic writes, orchestrator-only)
├── progress.md rendered FROM state.json (human-readable, not hand-edited)
├── decision-log.md durable decision record (committed)
├── retrospective.md durable retrospective (committed)
└── run/ ephemeral worker/run scratch (gitignored)
Human specs and plans live in docs/plans/, NOT under .soe/. .soe/ only
holds machine state that references those plans.
{
"mode": "autonomous-guardrailed",
"max_fix_cycles": 5,
"max_plan_revisions": 3,
"models": { "orchestrator": "<session model>", "reasoner": "opus", "worker": "sonnet" }
}
The models block is documentation defaults only. mode selects the interaction
model (autonomous-guardrailed / interactive / fully-agentic); the caps bound the
fix loop (5) and plan-revision loop (3).
This is the core rule the managed .gitignore block enforces:
docs/plans/* and per-track .soe/tracks/{id}/*.md
(decision-log.md, retrospective.md, progress.md) + state.json. This is
the engine's memory: it survives across sessions, machines, and teammates..soe/ run/ directories and
transient worker-status files. These are recreated every run and must never be
committed. The managed block in .gitignore (see gitignore-manager) targets
ONLY this ephemeral state — it never hides durable memory.Commit the durable files so soe's orchestration memory is shareable; let git ignore the ephemeral run-state so history stays clean.