cellm/skills/compress-llm/SKILL.md
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.
npx skillsauth add murillodutt/cellm cellm/skills/compress-llmInstall 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.
Operational skill for the Layer-1 LLM-driven compression stage defined
by ADR-004 and ported in Phases 1–3. Wraps SettingsManager writes so
developers never touch raw config files.
SettingsManager.set('CELLM_COMPRESS_LLM_ENABLED', …)
— never writes raw files, never bypasses precedence (ENV > DB > default).off | lite | full | ultra | wenyan-lite | wenyan | wenyan-full | wenyan-ultra. Any other string
is rejected at parseValue.commit | review | compress) are programmatic and
MUST NOT be settable via this skill — they flow through the
dispatcher, not the UI surface.cellm:prose) wins over this skill at runtime: when
the prose flag is active with level minimal/full, the interceptor
returns early regardless of this skill's flag state.| Command | Backend action |
|---------|-----------------|
| /cellm:compress-llm status | GET /api/ui-settings + read runtime flag |
| /cellm:compress-llm enable | SettingsManager.set('CELLM_COMPRESS_LLM_ENABLED', true) |
| /cellm:compress-llm disable | SettingsManager.set('CELLM_COMPRESS_LLM_ENABLED', false) |
| /cellm:compress-llm mode <X> | Validate X ∈ SELECTABLE_MODES, then set('CELLM_COMPRESS_LLM_MODE', X) |
| /cellm:compress-llm off | Alias for disable |
Status output shape:
{
enabled: boolean,
mode: SelectableCompressMode,
source: 'env' | 'db' | 'default',
prose: { active: boolean, level: 'off'|'minimal'|'full', synced: boolean }
}
CELLM_COMPRESS_LLM_ENABLED / CELLM_COMPRESS_LLM_MODE
— session-scoped, highest precedence.full (per ADR-004, until Phase 5 flips the default).commit, review, compress) in any prompt,
command completion, or CLI hint. They exist for the dispatcher only.~/.cellm/settings.json directly — always route through
SettingsManager.set() so DB cache + change-listeners stay consistent.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.
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.
documentation
Docops: scaffold LLM-first documentation structure, templates, and docops.json for a project. Creates directory hierarchy without overwrite. Use when: 'init docs', 'setup documentation', 'create docops structure'. Not for Oracle setup (use cellm:oracle).