trkbt10/indexion-skills
9 skills44.3k installs9 starsGitHub
Install
npx skills add https://github.com/trkbt10/indexion-skillsSkills in this repo
1Indexion ReadmeThe indexion-readme skill covers README construction: scaffolding templates, generating per-package READMEs from doc comments, planning writing tasks, assembling a root README from docs/ and package READMEs via doc.json, and verifying edits with plan drift. Start by checking git diff for README.md symlinks, doc.json at the repo root or under .indexion/readme/, and .indexion.toml config. Greenfield projects run indexion doc init to create template.md and doc.json. Configure packages, root sections (static, toc, packages), and static prose in docs/intro.md. indexion doc readme --per-package generates non-overwriting API skeletons from KGF doc comments; hand-write Overview, Usage, Options, and Examples afterward. Assemble with indexion doc readme --config=doc.json or template placeholders. Optional indexion plan readme emits per-section writing tasks. After regeneration or hand edits, indexion plan drift confirms changes are purely additive with cosine similarity checks suitable for CI guards.5kinstalls2Indexion SddThis skill bridges indexion quantitative spec alignment with SDD agent workflows using cc-sdd, Codex, and Claude. It drafts requirements from RFCs or ISO specs with indexion spec draft, checks fidelity with spec align diff, then runs cc-sdd design, tasks, and phased implementation. Phase A commits types and API surfaces; Phase B adds parsing and algorithm logic with per-task drift gates. indexion spec align status flags DRIFTED, SPEC_ONLY, and SHALLOW when requirements match type stubs without same-file function bodies. The validation loop runs spec verify, align diff, trace, and status with threshold 0.3, feeding reports into agent GO or NO-GO review and optional auto-fix via spec-impl. Operational guidance covers stall detection, external drift gate proxies, E2E tests after SHALLOW zero, vocabulary fixes on public declarations, and plan reconcile for documentation drift. Set cc-sdd lang to match implementation language, not the source spec language.4.9kinstalls3Indexion DocumentationThe indexion-documentation skill covers the evaluation side of the documentation lifecycle: what exists, what is missing, and what is stale. Run indexion plan documentation --style=coverage for overall coverage percentage, per-package breakdown, and functions versus types split. Use indexion plan documentation with github-issue or json formats for prioritized action items. Find undocumented pub declarations with indexion grep --undocumented. Detection uses KGF tokenization pairing visibility keywords with declarations and associating doc comments, language-agnostic across KGF-supported languages. Marker-only comments count as documented even without descriptive text; check doc_preview for quality. Detect drift with indexion plan reconcile reporting vocabulary divergence, stale docs, and missing docs. Scope checks to package-docs, tree-docs, or specific doc paths. Choose git commit timestamps or file mtimes. Generate dependency diagrams via indexion doc graph in Mermaid, DOT, D2, text, or JSON. Clear .indexion/cache/reconcile after schema changes. Coverage measures presence not accuracy; reconcile checks implementation to docs direction only.4.9kinstalls4Indexion RefactorThe indexion-refactor skill detects and eliminates duplication at textual, structural, and conceptual levels using indexion analysis commands, then verifies single source of truth enforcement. Phase 1 uses plan refactor at 90 percent threshold for duplicate code blocks and function-level duplicates within directories. Phase 2 uses plan solid --from=dirA,dirB for cross-package extraction candidates with apted or tfidf strategies. Phase 3 runs plan unwrap with dry-run then fix to remove trivial delegation wrappers. Phase 4 uses explore for concept-level duplication where files share vocabulary without copy-paste matches. Phase 5 consolidates into SoT: one concept, one module, one function. Phase 6 verifies with re-run refactor, explore, unwrap, and moon test. Phase 7 proves non-recurrence with structural invariant tests. Invoke after new abstractions, file format changes, fixes touching three or more files, guard workarounds, or periodic SoT health checks. Stop signals include platform stubs, type method similarity, CLI boilerplate, and semantic-but-different functions.4.9kinstalls5Indexion SegmentThe indexion-segment skill teaches agents to split text into contextual chunks for RAG and embedding pipelines with the indexion segment CLI. Default window strategy uses sliding window divergence detection; alternatives include TF-IDF topic change detection, punctuation and sentence boundaries, and hybrid NCD plus TF-IDF weighting for mixed-content documents. Operators tune granularity via min-size, max-size, target-size, threshold, window-size, and adaptive threshold flags, then emit prefixed segment files to an output directory. Invoke when users need document chunking for retrieval, meaningful section splits, or sub-document similarity prep. Workflow runs default segmentation first, adjusts threshold and target-size for chunk length, then enables hybrid mode with ncd-weight and tfidf-weight when single strategies underperform. CLI flags cover --strategy=tfidf, --strategy=punctuation, --hybrid, --prefix, and custom divergence thresholds so downstream vector indexes receive consistently sized, context-aware segments.4.9kinstalls6Indexion Kgfindexion kgf inspects and debugs KGF language specifications that power indexion dependency analysis across source files. Subcommands list installed specs, update from GitHub, add a single spec, and run inspect, tokens, events, or edges on any file with optional --spec and --kgf-dir overrides. Invoke when developing or fixing KGF specs, when indexion analysis output looks wrong, or when grep patterns miss because aliases such as pub map to KW_pub from lex sections and actual token kinds must be read from kgf tokens output first. The workflow starts with inspect for the full pipeline, then drills into tokens, events, or edges compared against kgfs lang.kgf spec files. Documented pitfalls cover PEG first-match ordering when DocComment consumes declaration docs, silent failures from missing NL between doc comments and keywords, bottom-up event order requiring bind and $scope from child to parent rules, and token priority where generic Operator patterns swallow EQ. Verify doc strings on declares edges after every spec edit.4.9kinstalls7Indexion WikiThe indexion-wiki skill is a runbook for navigating and maintaining indexion wikis stored at .indexion/wiki/ as Markdown governed by wiki.json. Each page requires four synchronized artifacts: page content, manifest entry, search indexes such as vectors.db and tfidf-vocabulary.json, and log.json audit entries. Agents must call wiki pages add or wiki pages update instead of editing .md directly, because direct edits leave search, manifest provenance, and logs stale. Workflows start with wiki index build to read index.md hub pages, detect stale coverage via wiki pages ingest source hashing, fix structure with wiki lint, and catch semantic drift using plan reconcile at the project root until New logical reviews reaches zero. The full maintenance cycle ingests stale tasks, updates pages through the CLI, lints cross-references, reconciles drift, rebuilds navigation, and confirms ingest dry-run reports zero pages needing attention. Commands cover pages plan for greenfield structure, export and import for GitHub wikis, wiki index build --full for vector search, and wiki log audit review.4.9kinstalls8Indexion Agent OrientThe indexion-agent-orient skill runs before implementation in unfamiliar or ambiguous codebase areas to stop agents from anchoring on prompt nouns and editing the wrong packages. The indexion agent orient command builds an incremental orientation map across the selected repository, then renders a short evidence-backed brief pasteable into AGENTS.md, CLAUDE.md, slash commands, Codex skills, or subagent prompts. Owner inference draws from the full prebuilt map including code structure, package READMEs, and wiki pages maintained by wiki pages update, wiki pages ingest, and plan reconcile, not only the display limit from --limit. Pipeline steps verify the CLI workflow exists, generate orient.md and orient-map.json with optional --no-update for cached maps, read Likely Implementation Owners, Knowledge Sources, Consumer Surfaces, Do Not Implement Here, Required Preflight, and Orientation Map sections, confirm owners via doc graph, grep, and search, gate implementation against unsafe paths, and delegate to zero-knowledge subagents with a quiz on owners and preflight evidence.4.9kinstalls9Indexion IdentityThe indexion-identity skill guides agents through detecting and reviewing name and content drift across codebases using the indexion identity audit CLI. It scans files, folders, and graph-level declaration symbols, treating each audit row as a review candidate rather than proof of drift. Agents compare scoped names against expected and actual summaries, assessments, and recommendations before follow-up verification with indexion doc graph, semantic grep, and ripgrep. The workflow emphasizes choosing the smallest confirmed operation: treating insufficient-content as an evidence problem, renaming stale symbols or files, adjusting folders, moving misplaced files, or splitting files with multiple dominant responsibilities. File names are evaluated with parent scope, and declaration-heavy files may read evidence-thin until inspection confirms a rename or split. After edits, teams rerun moon info, moon fmt, moon test, and a fresh identity audit without maximizing scores mechanically.4.9kinstalls