메뉴
Weekly bounded maintenance loop across ingestion, hygiene, compaction, lifecycle management, and concepts
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Run a marketing/positioning pass on an open source repo's README and public-facing docs so a developer landing on it for the first time immediately understands what it is, why it exists, and how to use it. Use this whenever the user wants to improve a repo's README, write launch copy for an OSS project, sharpen positioning ("does this README explain what the project does"), prep a repo for a Show HN / Hacker News / Product Hunt / Twitter launch, or asks for an "OSS marketing pass," "README audit," or similar. This is about clarity and conversion (visitor → user → star), not contribution mechanics — for CONTRIBUTING.md, issue templates, CODEOWNERS, or governance docs, that's a different concern.
Audit and prepare a GitHub repo for open source release, with a focus on developer experience (DevEx) — the things that determine whether a stranger can go from "found this repo" to "opened a PR" without friction. Use this whenever the user wants to open-source a repo, make a private repo public, do a "pre-launch checklist" or "OSS readiness audit," improve their README/CONTRIBUTING/issue templates, or asks something like "is this repo ready for people to use/contribute to." Also trigger for narrower asks that are really pieces of this — "write a CONTRIBUTING.md," "set up issue templates," "add a CODEOWNERS file" — since those are almost always better done as part of the full readiness pass rather than in isolation.
Design, build, or review command-line interfaces using the Command Line Interface Guidelines from clig.dev. Use this skill whenever the user is creating a new CLI, adding commands/subcommands/flags, designing help text, choosing stdout/stderr/exit-code behavior, adding JSON or plain output modes, improving error messages, making destructive operations safer, or asking whether a CLI feels good, scriptable, discoverable, or Unix-friendly.
Collapse semantically overlapping notes, ideas, and project docs into clearer canonical surfaces while preserving distinct claims, decisions, examples, and source provenance.
Maintain canonical concept pages and surface recurring themes
Hygiene pass over active notes, ideas, and projects
| name | vault-maintain |
| description | Weekly bounded maintenance loop across ingestion, hygiene, compaction, lifecycle management, and concepts |
Run a comprehensive weekly maintenance sweep across the active vault surface. This is the unified entry point for periodic vault care, coordinating ingestion, hygiene, semantic compaction, project lifecycle management, and concept promotion into a single bounded workflow.
Portent reference: ../references/portent-knowledge-base-spec.md.
Work the active surface newest-first, ranked by each object's updated
frontmatter date (fall back to file modification time when updated is absent).
Concentrate scrutiny on the freshest material — recently-touched objects are
where active thinking lives and where new mismatches, overlaps, and gaps appear.
Long-untouched objects are assumed settled and rank lower, but are never
excluded: they remain valid link targets, merge destinations, and canonical
homes, and still get a structural check (broken links, orphaning).
Every sub-audit below inherits this newest-first ordering, and the weekly summary leads with findings on recently-updated objects.
vault-maintain [--mode report|apply-safe|apply]
--mode report|apply-safe|apply (default: report)
report: audit and preview all findings (read-only)apply-safe: apply high-confidence, low-risk updates onlyapply: include medium-confidence actions when evidence is clearRun these steps in order to produce a complete weekly maintenance report:
Run vault-ingest --mode report to:
raw/sources/ for unprocessed capturesRun a report-mode pass over active objects to find:
typestatusRun vault-lint --mode report to:
Run vault-tracker --mode report to:
Run an aggressive report-mode review of ideas/fleeting/ and
ideas/incubating/:
keep incubatingpromote to projectmove to notescompact into <canonical>rejectdelete absorbed curated filesDefault recommendation should be opinionated. Do not leave stale ideas as "maybe" when the evidence supports a clear cleanup action.
Run vault-compact --mode report to:
vault-lintRun vault-concepts --mode report to:
Compile findings into a weekly summary including:
raw/processed/YYYY-MM-DD/)external, owned notes not tagged
external)Append the maintenance entry to log.md manually when the vault uses one:
vault-maintainreport (Default)apply-safeApply only high-confidence, low-risk changes:
vault-lintlog.md if presentExcluded from apply-safe:
report)applyAdditionally allow medium-confidence actions:
vault-compact when provenance, unique content, and links
can be preservedvault-compact
merges their unique content into a canonical destination and rewrites inbound
linksraw/processed/YYYY-MM-DD/ when a summary exists without an immutable source
archiveraw/ is immutablevault-compact may delete
curated files after absorption into a canonical destination; list every
deleted file and its canonical replacementarchive/ from active maintenance unless explicitly scopedraw/processed/* files; additive archived source
copies are allowed for synthesis archive repair--mode reportThis skill orchestrates the following specialized skills:
| Step | Skill | Purpose |
|---|---|---|
| 1 | vault-ingest | Source classification and routing |
| 2 | vault-lint | Portent object and hygiene audit |
| 3 | vault-tracker | PORT lifecycle management |
| 4 | vault-compact | Semantic merge and absorption |
| 5 | vault-concepts | Topic promotion and creation |
Return:
ideas/, and recommended
action# Preview weekly maintenance (read-only)
vault-maintain --mode report
# Apply safe updates only
vault-maintain --mode apply-safe
# Full maintenance with source moves and concept creation
vault-maintain --mode apply