ar9av/obsidian-wiki
37 skills55.1k installs65.2k starsGitHub
Install
npx skills add https://github.com/ar9av/obsidian-wikiSkills in this repo
1Llm Wikillm-wiki is the foundational theory skill for Andrej Karpathy’s LLM Wiki knowledge-distillation pattern in Obsidian. Solo builders and small teams use it when they want a durable second brain: originals stay untouched in Layer 1, curated notes form Layer 2, and Layer 3 schema keeps linking and retrieval predictable for agents. The skill walks through why distillation beats re-asking the model, how to place raw articles, PDFs, logs, and vision-readable images under configured source paths, and when to delegate ingestion or linting to sibling skills. It fits early research organization, mid-build documentation systems, and long-run operations where product and GTM context must compound without drift. Expect architectural guidance and naming structure—not a one-click ingest pipeline.2.4kinstalls2Wiki IngestWiki Ingest provides mental frameworks and prompt templates for distilling arbitrary sources into an Obsidian-style personal wiki. Indie builders maintaining a second brain use it when a paper, blog post, or transcript should become durable concept, entity, and skills pages rather than a forgotten chat thread. The skill walks through extraction (top ideas, who/what deserves a page, procedures, claims), synthesis when new material overlaps existing notes, and deliberate linking so the graph compounds. It shines in Build when documentation is the product glue for an agent or SaaS, but it also supports Idea-phase research capture and Grow-phase content reuse because the same ingest loop feeds validation notes and public writing outlines.2.3kinstalls3Wiki QueryWiki-query is an agent skill for solo builders who treat Obsidian as a compiled knowledge base. Instead of grepping random markdown in a repo, the agent resolves vault configuration (including ~/.obsidian-wiki/config), loads QMD-related settings, and retrieves answers from wiki pages with citations. Users ask what they already know about a topic, request everything related to a concept, or run multi-hop graph questions that follow typed links across hops. A fast path honors explicit user intent to answer from page summaries and frontmatter only, saving tokens when full body reads are unnecessary. It is journey-wide: the same skill supports Idea research from prior notes, Validate scoping from internal playbooks, Build decisions documented in the vault, and Grow lifecycle recall of support or content strategy. It complements the parent obsidian-wiki tooling referenced in SKILL.md and assumes a built wiki, not a one-off note.2.3kinstalls4Wiki Setupwiki-setup is an agent skill for initializing or reconfiguring an Obsidian wiki vault so solo builders can turn scattered documents and Claude history into a structured, queryable knowledge base. It walks through creating a .env from .env.example, collecting vault location, source document paths, optional Claude export paths, and optional QMD semantic-search settings including MCP versus CLI transport. Use it when you say set up my wiki, initialize obsidian, create a new vault, or need to fix a broken vault layout. The skill emphasizes absolute paths and sensible defaults such as ~/Documents for sources and auto-discovery under ~/.claude. For indie builders shipping with agents, this is the on-ramp before ingestion and search skills—not a one-off file copy but a repeatable vault bootstrap that keeps agent context and personal notes in one Obsidian-shaped system.2.2kinstalls5Wiki LintWiki-lint is an agent skill for Obsidian vault owners who treat their wiki as a product. It performs a health audit: inventory from index.md, recent context from log.md, and targeted greps instead of reading every file on large vaults—following the llm-wiki framework in SKILL.md. Default mode is report-only; add --consolidate to enter a dream cycle that can fix links, cross-reference orphans, correct lifecycle states, demote stale peripheral pages, normalize tags, and add contradiction callouts, always with dry-run preview and explicit confirmation before writes. Solo builders use it when a second brain grows messy after shipping features, research spikes, or multi-project note sprawl. It pairs with the broader obsidian-wiki skill family for config resolution and retrieval discipline. Intermediate complexity: you need a configured OBSIDIAN_VAULT_PATH and comfort approving automated edits to markdown.2.2kinstalls6Wiki UpdateWiki Update is the outbound half of the obsidian-wiki stack for builders who treat Obsidian as a personal or team knowledge base. Unlike wiki skills that only run inside the vault repo, this skill executes from whichever project you are in when you ask to update the wiki, sync to obsidian, or save recent work to your vault. It follows the Config Resolution Protocol to locate OBSIDIAN_VAULT_PATH, repo metadata, and link format, then compares against .manifest.json and index.md so new material complements existing structure instead of blindly overwriting. Agents scan README, docs, and source layout to understand the project, then distill durable concepts, decisions, and references with correctly formatted internal links. Solo builders juggling multiple repos use it continuously across Idea research dumps, Validate scope notes, Build implementation detail, and Operate runbooks. Complexity is beginner-friendly on the surface but assumes a configured vault and willingness to curate what gets promoted from ephemeral chat into long-lived notes.2.2kinstalls7Wiki StatusWiki Status is an agent skill for solo builders who treat Obsidian as the system of record for Claude/Codex history and project notes. It computes what has been ingested, what is still pending, and how source directories differ from vault pages by reading the vault-root manifest and resolved paths for sources and agent histories. That dashboard moment matters before you burn time on a full rebuild or miss new files that only need an append ingest. Triggers cover plain-language requests for status, delta, dashboard, and structure questions such as central hubs and connectivity. Insights mode analyzes the wiki graph itself—not just file counts—to surface top hubs and bridge pages. Pair it with ingest skills in the same Obsidian Wiki pack after config is valid. It is read-heavy audit workflow guidance; it does not replace the ingest pipeline or modify the vault without your follow-up commands.2.2kinstalls8Wiki History Ingestwiki-history-ingest is a unified router skill for the Obsidian wiki toolkit. Solo builders who journal work across Claude Code, GitHub Copilot, Codex, Pi, Hermes, and OpenClaw often lose context in scattered session stores. This skill gives one verbal entry—wiki-history-ingest with an optional target—so the agent loads the correct specialized history ingester instead of you memorizing six skill names. It deliberately handles history sources only; document ingestion stays on wiki-ingest. Routing tables map familiar paths like Claude memory JSONL, Copilot session-store.db, and Codex rollout indexes to the right downstream skill. That makes it valuable in Operate when you iterate on process, and in Grow when you turn shipped work into searchable notes. Install it when your second brain lives in Obsidian and your primary work happens inside coding agents.2.1kinstalls9Wiki RebuildWiki Rebuild is a documentation operations skill for solo builders who treat an Obsidian vault as an agent-maintained wiki synced from external sources. It governs high-risk maintenance: snapshot the current graph into `_archives` with timestamps, optionally reprocess every source from scratch, or restore an older archive when a bad merge polluted links and manifests. Before any step the agent resolves vault configuration the same way as the parent llm-wiki flow, reads `.manifest.json` for state, and confirms whether the user wants archive-only safety, full rebuild, or rollback—never assuming consent. Use it when someone says rebuild the wiki, start over, nuke and repave, or when incremental ingestion left the knowledge base irreconcilable with repos and APIs. The outcome is a recoverable history plus either a clean regenerated tree or a known-good prior version, which keeps long-running agent wikis trustworthy without manual copy-paste rescue jobs across hundreds of notes.2.1kinstalls10Data Ingestdata-ingest is an agent skill from the Obsidian wiki stack that turns arbitrary unstructured text into curated Obsidian pages. Solo and indie builders use it when exports and dumps outpace normal note-taking—conversation logs, team threads, meeting transcripts, journals, partial CSV rows, bookmark lists, or email archives that are not already handled by narrower ingest skills in the same repo. The workflow starts by resolving vault configuration (walking CWD for .env, falling back to ~/.obsidian-wiki/config), then checking .manifest.json and index.md so repeat runs do not clobber prior work. The agent infers source shape, extracts durable facts and themes, and writes distillations as wiki entries with links that respect OBSIDIAN_LINK_FORMAT. It is intentionally the universal fallback after more specific ingest paths fail, which keeps one command surface for “add this export to the wiki” without forcing builders to guess which specialist skill applies. Pair it with llm-wiki setup so paths and linking rules stay consistent across ingest runs.2.1kinstalls11Wiki ExportWiki Export is an agent skill for solo builders who keep research and specs in Obsidian and need the wikilink knowledge graph outside the vault. After resolving OBSIDIAN_VAULT_PATH through the shared obsidian-wiki config flow, it materializes the link graph as JSON, GraphML, Neo4j-ready Cypher, and a self-contained HTML explorer in a wiki-export folder. Default behavior includes every page; you can request a public export to omit internal and PII-tagged notes. Use it when integrations, analytics, or stakeholder-ready visuals need structured graph data instead of raw markdown. It fits builders shipping documentation-heavy products, second brains, or agent context pipelines where links matter as much as page text.2.1kinstalls12Tag TaxonomyTag Taxonomy is an agent skill for solo builders who run an Obsidian wiki as their second brain and need tags that actually filter and link—not hundreds of near-duplicates. Before creating or updating any page, the agent resolves vault config, loads _meta/taxonomy.md, and applies rules like broad-over-narrow tags and alias normalization. Use it when audits show messy frontmatter, when you add a new canonical tag to the taxonomy, or when you ask what tags a page should use. It pairs with the parent obsidian-wiki workflow but stands alone whenever tagging discipline matters. Works with Claude Code and Cursor agents that can read files in your vault path.2.1kinstalls13Claude History IngestClaude History Ingest is an Obsidian wiki skill for solo builders who accumulate months of Claude Code transcripts and want that material in a vault instead of lost in ~/.claude. The agent follows a defined ingest flow: resolve vault and CLAUDE_HISTORY_PATH, honor .manifest.json for incremental append runs, and align new notes with what index.md already covers. It triggers when you say you want to process Claude history, import the .claude folder, or pull insights from prior sessions. The skill is procedural knowledge management—it does not run builds or deploys—but it pays off across the journey whenever you need to remember decisions, patterns, or failed approaches from earlier agent work. Pair it with the broader llm-wiki setup so paths and manifest semantics stay consistent.1.9kinstalls14Skill CreatorThis skill documents a Post-hoc Analyzer Agent for builders who iterate on agent skills using blind comparisons. Once a comparator chooses side A or B without revealing which skill was which, this workflow re-opens the identities, reads the blind comparator JSON, both SKILL.md files and referenced assets, and both execution transcripts. The agent explains what the comparator valued in the winning output and why the loser fell short—covering instruction specificity, tool usage, examples, and edge cases. The deliverable is saved analysis at output_path that feeds the next edit cycle on the weaker skill. It suits solo maintainers of skill libraries or Obsidian wiki skill packs who treat skills like products that need evidence-based revision, not one-shot prompts.1.9kinstalls15Ingest Urlingest-url is an Obsidian wiki skill that fetches a single web page and turns it into a structured note in your vault. Solo builders use it when research tabs pile up and they want durable, searchable knowledge next to their code instead of brittle browser history. If the agent runs from a recognized project path, the note lands in that project folder and can create the project stub when missing; otherwise it goes to misc/ until links and tags pull it toward a project. The skill emphasizes a strict trust boundary: fetched content is data to summarize and attribute, not instructions to follow, and the agent must not run embedded commands or expand network access beyond the requested URL. It fits early discovery and ongoing documentation workflows for agent-assisted builders who treat the vault as the source of truth for what they read and decided.1.7kinstalls16Graph ColorizeGraph Colorize is an agent skill for Obsidian users who want the graph view to communicate structure instead of looking like one gray blob. When you ask to color the graph by tag, category, folder, or visibility, the agent resolves your vault path through the obsidian-wiki config flow, inspects how your notes are actually organized, builds a `colorGroups` array of `{query, color}` pairs in Obsidian’s packed RGB format, and merges that into `.obsidian/graph.json` while preserving other settings. It is aimed at solo builders and researchers who live in a vault for specs, research, and shipping checklists and need quick visual distinction between note types. The skill assumes the vault has been opened in Obsidian at least once so `.obsidian/` exists. It is narrowly scoped to graph styling—not general note editing or sync—and is ideal when your tag taxonomy is stable enough to encode as persistent color rules.1.7kinstalls17Wiki Researchwiki-research turns your coding agent into a librarian that hunts the open web in multiple passes, merges findings, and commits them into an Obsidian vault using the obsidian-wiki conventions. Before searching, it resolves vault path and link format, inspects index and hot notes for prior art, and honors optional research-config rules about sources and confidence. Triggers include slash commands like /wiki-research, plain-language requests to research or deep-dive a topic, and autonomy-oriented phrasing when the user wants comprehensive coverage filed directly into the wiki. Solo builders use it to compound research from idea exploration through content and product decisions without losing citations in chat history. It pairs naturally with the parent llm-wiki skill for linking standards and setup. Confirm ambiguous topics with the user, then let the loop run until structured pages land under the vault tree you already maintain.1.7kinstalls18Wiki CaptureWiki Capture is an agent skill for solo builders who use an Obsidian-backed llm-wiki vault and want conversations to compound into searchable knowledge. When you say save this, invoke /wiki-capture, or ask to preserve or file what you just discussed, the skill resolves OBSIDIAN_VAULT_PATH and OBSIDIAN_LINK_FORMAT from .env or ~/.obsidian-wiki/config, scans index.md (and hot.md when present) to avoid redundant notes, then identifies what would still matter in three months without this chat. It rewrites that substance as declarative knowledge—especially decisions and why they were made—classifies the topic, and writes a structured note into the right category with wikilinks per the shared llm-wiki link rules. That makes agent sessions part of a durable second brain instead of disposable threads, whether you were debugging, planning a launch, or researching competitors.1.6kinstalls19Wiki Dashboardwiki-dashboard helps solo builders turn a sprawling Obsidian vault into queryable, self-maintaining views without exporting to a separate BI tool. When someone asks for a dashboard, content index, or “show all X as a table,” the skill chooses between Obsidian Bases—native since 1.8, GUI-friendly YAML—and Dataview when they need SQL-like power or aggregations. It follows the obsidian-wiki config protocol to find the vault, reads index.md to see what already exists, and clarifies folder, tag, or category scope before generating files. That makes it valuable during Build while writing docs and PM notes, and again in Grow when you want lifecycle or content pipelines visible in one pane. Prefer Bases unless the user explicitly wants Dataview or lacks 1.8+. The output is durable vault artifacts agents and humans can reopen as the wiki grows.1.6kinstalls20Wiki SynthesizeWiki-synthesize is an agent skill for solo builders who treat an Obsidian vault as a structured wiki. It systematically scans the full page inventory (starting from index.md), maps which concepts appear together via links, tags, and headings, and finds pairs or clusters that repeat across many pages but still lack a dedicated synthesis page. It then ranks opportunities using co-occurrence tiers and hub rules, drafts synthesis pages that state explicit cross-cutting conclusions rather than summaries, and wires the graph with reverse links, index entries, and optional synthesis-map metadata. Triggers include natural language requests to synthesize the wiki, find connections, or a /wiki-synthesize command, especially after a large ingest when the vault has grown enough that implicit overlap outpaces explicit synthesis coverage. It assumes you have resolved vault path and link format via the shared llm-wiki config protocol and that index.md reflects current pages.1.6kinstalls21Codex History IngestCodex History Ingest is an agent skill for solo builders who run Codex locally and want reliable ingestion of session history instead of hand-parsing JSONL. It explains the on-disk layout under ~/.codex, how session_index.jsonl lists thread ids and titles, and how rollout files stream typed events from session metadata through tool calls and model response items. Use it when building importers, Obsidian vault sync, or analytics over past agent runs, including delta modes driven by updated_at on the index. It turns opaque Codex logs into a predictable schema so you can search decisions, token usage, and command outcomes long after a thread ends.1.6kinstalls22Hermes History IngestHermes History Ingest is a reference skill for builders who run Hermes agents and want those runs preserved in an Obsidian wiki via the obsidian-wiki ingest path. It specifies how to read ~/.hermes/ (or HERMES_HOME): markdown and JSON memories with tags, timestamps, project, and source linkage, plus optional session JSONL under dated directories. Solo operators use it when automating knowledge capture from agent sessions—turning episodic chat into tagged notes without hand-copying. It does not execute ingest by itself in the excerpt; it equips the agent with correct field semantics, normalization rules, and directory layout so downstream wiki generation stays consistent. Pair it with broader obsidian-wiki skills when you are operating a personal knowledge base alongside Claude Code or similar agents.1.4kinstalls23Openclaw History Ingestopenclaw-history-ingest is an agent skill for solo and indie builders who run OpenClaw and want their agent work preserved in an Obsidian wiki instead of buried in session logs. It targets the structured long-term MEMORY.md file and JSONL session transcripts, emphasizing durable insights over ephemeral tool chatter. Before writing, it resolves OBSIDIAN_VAULT_PATH and OPENCLAW_HISTORY_PATH, then reads the vault’s .manifest.json and index.md so ingestion extends what you already documented rather than duplicating pages. Append mode is the default: each source file is checked against the manifest and only new or unprocessed material is ingested. Triggers include phrases like processing OpenClaw history, ingesting ~/.openclaw, or adding prior OpenClaw sessions to the wiki. It pairs naturally with the broader Obsidian wiki stack in the same repo and the wiki-history-ingest router when you want a consistent entry point across history sources.1.3kinstalls24Wiki AgentWiki Agent is a journey-wide skill for solo builders who split work across Claude Code, Codex, Copilot, Hermes, and OpenClaw and need targeted recall instead of re-explaining solved problems. When you invoke a /wiki-* command with an optional search topic, the agent finds sessions in that tool’s history about the topic, ingests only those blobs into your Obsidian wiki, and answers with synthesis you can use immediately in the active session. That makes it cross-referencing infrastructure: you stay in one agent while mining another’s trail. It deliberately differs from bulk wiki-history-ingest, which archives all new history. Use it during Build when debugging a pattern you solved in Codex last month, during Operate when tracing how you fixed an incident in Claude, or during Validate when you want prior pricing or scope conversations without manually scrolling exports.1.3kinstalls25Daily Updatedaily-update is an Obsidian-wiki agent skill that runs a lightweight maintenance pass over your vault: verify source freshness, refresh the index, regenerate hot.md, and write runtime state that powers terminal notifications. Solo builders who treat an Obsidian wiki as living ops documentation—not a static export—can invoke it when they want a repeatable morning sync instead of manually grep-ing staleness across plugins and repos. The skill begins with the shared Config Resolution Protocol from llm-wiki/SKILL.md, walking CWD for .env, then ~/.obsidian-wiki/config, so OBSIDIAN_VAULT_PATH and OBSIDIAN_WIKI_REPO resolve before any bash work. It derives a per-vault STATE_DIR hash so multiple vaults on one machine do not clobber each other, then reads .manifest.json at the vault root as the source of truth for what to check. Default run mode executes the full cycle; alternate phrasing in the description also covers installing or verifying launchd cron and the terminal reminder UX. This pairs naturally with other obsidian-wiki skills for capture and publishing, but daily-update itself is the hygiene loop that keeps downstream agents and humans oriented on what changed overnight.1.3kinstalls26Memory BridgeMemory Bridge is an agent skill for solo builders who maintain an Obsidian wiki with llm-wiki-style provenance tracking. It turns `.manifest.json` and per-page `sources:` metadata into navigable views: browse everything Codex (or another tool) contributed, search within a tool’s corpus, or run diff mode to see what one agent documented that another did not. That gap analysis is the differentiator—many builders rotate between Claude Code, Codex, and other assistants without realizing their wiki is unevenly sourced. The skill assumes you already sync or generate wiki pages with source tags; it does not replace wiki authoring. Invoke when switching tools mid-project, auditing institutional memory, or answering “what did Codex know about X?” before re-asking the same question in chat. Complexity is intermediate because you need a configured vault and familiarity with the parent wiki workflow.1.3kinstalls27Impl ValidatorImpl Validator is a journey-wide quality subagent skill for solo builders who want agent work checked against intent before it ships to the user. In subagent mode it consumes a structured check block listing the goal, artifacts (files, commands, or text), and verifiable assertions; in user mode it responds to direct review requests. The reviewer role is deliberately critical: match stated goals to what was actually produced and surface concrete failures early. Other Obsidian-wiki skills can spawn it automatically so bridges and daily updates self-check. Pair it after any substantial agent implementation or documentation pass when you need a lightweight audit without a full human code review ceremony.1.2kinstalls28Obsidian Wiki IngestObsidian Wiki Ingest is a workflow agent skill that acts as the automation layer for the obsidian-wiki repository: it takes new or changed sources from a configured sources directory and runs the wiki-ingest pipeline into your Obsidian vault. Solo builders who treat notes as part of the product—specs, research, runbooks—use it to avoid manual copy-paste and duplicate pages. The skill enforces manifest-backed deduplication, consistent frontmatter, and cross-linking so the vault stays navigable as content grows. You choose append versus full versus raw ingest depending on whether you are incrementally adding files or rebuilding from _raw_. It expects the obsidian-wiki workspace layout and environment variables for vault and sources paths. Use it during Build when consolidating docs, and in Operate when continuously feeding operational notes and imports into the same knowledge base.1.1kinstalls29Wiki SwitchWiki Switch is an agent skill for solo builders and writers who maintain more than one Obsidian wiki vault through the obsidian-wiki CLI layout. Each vault is a named file at `~/.obsidian-wiki/config.NAME`, and the active vault is whichever file `config` symlinks to—so switching is re-pointing that symlink, not copying secrets around. The skill routes invocations to switch, list, show, or create profiles, confirms the vault path after activation, and handles empty invocations as a vault list. Use it when you bounce between work and personal knowledge bases, onboard a new machine profile, or ask the agent which wiki is active before sync or publish steps. It reduces symlink and path mistakes that otherwise break automation hooked to `OBSIDIAN_VAULT_PATH`.1.1kinstalls30Wiki Digestwiki-digest generates a periodic, human-readable summary of what changed in your Obsidian knowledge base: new insights, updated notes, emerging themes, and items worth revisiting. It intentionally reports on knowledge, unlike wiki-status which focuses on ingestion of external sources. Before writing, the agent resolves vault configuration through the shared llm-wiki protocol, then interprets the requested window—default seven days, or daily, monthly, ISO since-dates, and explicit day counts. Solo builders use it as a lightweight review ritual after research sprints, shipping notes, or learning-heavy weeks so loose graph updates become an actionable narrative. Pair it with your broader Obsidian wiki toolchain when you want reflection and prioritization rather than pipeline status alone.1kinstalls31Copilot History Ingestcopilot-history-ingest is the detailed reference arm of the obsidian-wiki project for GitHub Copilot CLI artifacts on a developer machine. Solo builders accumulate dozens of UUID session folders under ~/.copilot/session-state, each mixing workspace.yaml timing metadata, optional VS Code workspace and branch context, and line-delimited events.jsonl transcripts. This skill tells an agent exactly which fields matter—like customTitle for labeling wiki pages and summary_count for checkpoint expectations—so ingestion does not drop short sessions. It fits when you want coding-agent history searchable beside product notes instead of lost in local JSONL. Most work happens in Build docs while consolidating knowledge, and again in Operate iterate when reviewing what the agent changed on a branch. It is reference-heavy rather than a single shell recipe; pair it with broader obsidian-wiki flows that write markdown into the resolved vault path.899installs32Wiki DedupWiki-dedup is an agent skill for solo builders who run an Obsidian-backed llm-wiki and keep hitting the same idea under two filenames. It walks identity-resolution: inventory from index.md, compare titles and frontmatter, then optionally merge duplicates after you confirm. The workflow is write-heavy and intentionally cautious because merges are destructive. It plugs into the ar9av Obsidian wiki stack alongside wiki-lint and cross-linker, using the shared config resolution protocol so agents know vault path and link format before scanning. Use when aliases accumulate after research sprints or when imports created parallel pages—you get one canonical note per concept instead of fragmented stubs that confuse RAG and human browsing.666installs33Wiki Context PackWiki Context Pack is an agent skill for solo builders who keep an Obsidian wiki as system-of-record and need to feed another agent a focused, size-limited knowledge slice. It follows the Obsidian wiki config protocol, reads hot.md and index.md when present, and assembles the most relevant pages into a single markdown block sized to your budget. Triggers include /wiki-context-pack, requests for a context slice, or packing the wiki for a downstream task. Unlike wiki-query, which optimizes for an immediate answer, this skill optimizes for reusable input—architecture decisions, research threads, or recent edits—so coding, planning, or review agents start with grounded context. It fits builders running Claude Code or Cursor against a personal or team vault who repeatedly hit context limits when naively dumping notes.538installs34Wiki Stage Commitwiki-stage-commit is an agent skill for solo builders running an Obsidian LLM wiki with staged writes enabled. When the agent drafts pages into _staging/, you need a deliberate promotion step so bad or hallucinated structure never hits live wiki paths. Invoke it after batch wiki generation or when you ask what is waiting in staging. The skill lists staged files, summarizes them for review, and on acceptance moves content to final wiki locations; rejections route back to _raw/ for hand editing. It depends on correct .env configuration, OBSIDIAN_VAULT_PATH resolution from the parent llm-wiki skill, and WIKI_STAGED_WRITES=true. Use it as the approval gate in a docs-as-code workflow alongside other obsidian-wiki commands, not as a general git commit helper.537installs35Pi History IngestPi-history-ingest is an agent skill that distills Pi coding agent session history into an Obsidian wiki vault. Solo builders who run Pi locally accumulate structured JSONL under ~/.pi/agent/sessions; this skill walks that tree layout, follows the active conversation branch, and compiles knowledge into wiki pages while respecting what is already in .manifest.json and index.md. Before processing, it resolves OBSIDIAN_VAULT_PATH and PI_HISTORY_PATH using the same config protocol as llm-wiki, defaulting the history path when unset. Default append mode only ingests new or changed source files, which keeps long-running vaults maintainable. Use it when you want to answer what have I worked on in Pi, import prior sessions for documentation, or feed an personal knowledge base from agent transcripts. It complements direct wiki editing rather than replacing structured PM tools.1installs36Wiki Importwiki-import is the companion ingest path for obsidian-wiki exports: it takes a validated graph.json (nodes, links, graph envelope) and materializes page stubs in your current vault with consistent frontmatter, wikilinks, and typed edges. Indie builders maintaining a personal knowledge base or a team handbook in Obsidian use it when migrating between machines, merging exports, or bootstrapping AGENTS.md-aware conventions from a prior wiki-export run. The skill enforces a preview step—source path, export timestamp, and counts—before writes, and stops with actionable errors when JSON shape is wrong or empty. It expects you to resolve vault path through the shared llm-wiki config protocol and to honor owner rules in AGENTS.md when present. This is procedural vault surgery, not open-ended note-taking chat: pair it after wiki-export when moving graphs, not when you only need a single manual note.1installs37Wiki Quick Chat Capturewiki-quick-chat-capture is an agent skill that turns messy coding-session insights into Obsidian-ready `_raw/` drafts. Solo builders lose decisions in chat logs; this skill encodes problems, root causes, fixes, and validation notes in a repeatable markdown shape with YAML frontmatter so a separate ingest skill can promote drafts into your vault taxonomy. It targets developers who maintain a personal or team wiki and want agent sessions to feed that system without hand-reformatting every time something surprising breaks. Use it right after you confirm a fix or document a quirk, before the context fades. It pairs naturally with wiki-ingest workflows and emphasizes draft lifecycle, base confidence, and source attribution rather than publishing polished docs in one shot.1installs