cellm/skills/hefesto/SKILL.md
CELLM feature construction engine — builds new capabilities from CONSTRUCT findings or direct feature requests. Designs architecture, decomposes into CellmOS specs, implements end-to-end, certifies via Argus. Use when: 'build', 'construct', 'create feature', 'implement new', 'hefesto on X', 'implement the CONSTRUCT items'. Does not fix existing code (that is Asclepius).
npx skillsauth add murillodutt/cellm cellm/skills/hefestoInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Build new capabilities from specs using SCE as the context and governance backbone.
spec_search / spec_create_node).context_preflight before phase execution (flow='hefesto').context_certify for critical/strict flows before execution-ready transitions.context_record_outcome.spec_* tools), not in markdown planning only.spec_search + spec_get_tree).context_certify) when risk/strictness requires.context_record_outcome).Every check MUST have 3 levels: check -> phase -> task.
| Level | nodeType | Body fields | Purpose |
|-------|----------|-------------|---------|
| Check | check | context, problem, principle | Top-level spec |
| Phase | phase | description, briefing, specialist | Work group |
| Task | task | action, fileRef, diffExpected | Atomic unit of work |
NEVER execute a phase directly as if it were a task.
ALWAYS create 1+ tasks under each phase before starting execution.
Tasks are what spec_claim_next returns — phases without tasks break swarm execution.
Decomposition sequence:
spec_create_node(nodeType: 'check') — top-levelspec_create_node(nodeType: 'phase', parentId: checkId) — per work groupspec_create_node(nodeType: 'task', parentId: phaseId) — per atomic actionspec_add_edge — dependency edges between phases/tasksWhen prompt contains [OLYMPUS CONTEXT]:
| Aspect | Standard | Olympus |
|--------|----------|---------|
| Feature source | Direct request / CONSTRUCT findings | OLYMPUS CONTEXT FINDINGS TO BUILD list |
| Resolution | Journal only | Also triad_resolve_finding after each build |
| Session | None | Extract session_id from context |
| Spec linkage | Normal | Pass specId to triad_resolve_finding |
| Report path | docs/cellm/reports/ | ~/.cellm/reports/ |
Resolutions: built, blocked, false_positive.
data-ai
Prose override — temporarily disable quantization and respond in readable prose. Use when relational density matters, for safety-critical explanations, onboarding handoffs, or when token economy is not the priority.
development
Govern explicit weekly Super PRs or maintainer-requested PR merges. Evaluates a 10-criterion readiness checklist and performs governed merge only when a user-requested PR is READY. Never creates or keeps permanent PRs. Use when: 'pr-check', 'pr-merge', 'merge this PR safely', 'is PR ready', 'guard merge', or /sk-git delegates pr-merge.
data-ai
Operational surface for the compress-llm Layer-1 token I/O compressor. Enable, disable, switch mode, and inspect status without editing config files. Use when tuning compression pressure for the current session or project.
tools
Generate structured upstream feedback for the CELLM engineering team. Produces evidence-first Markdown at docs/evidence/<date>-cellm-feedback-*.md for bugs, anti-patterns, deprecation gaps, and harness surprises, with optional atom registration via knowledge_ops. Use when: 'feedback for CELLM', 'send to CELLM team', 'register this as atom', 'document this anti-pattern', 'report this bug upstream'. Trigger proactively on MCP schema/runtime mismatches, mechanical edit loops (>=3 sequential edits), short deprecation windows (<6 weeks), or reusable harness surprises. Do NOT trigger for routine feature work or project-local bugs.