mcpbeat Sign in

Wiki Stage Commit Agent Skill

> Review and promote staged wiki pages to their final locations. Use when WIKI_STAGED_WRITES=true and the user says "/wiki-stage-commit", "review staged pages", "commit staged writes", "promote staged pages", "approve staged changes", or "what's waiting in staging". Shows each staged file, lets the user accept or reject it, and moves accepted files to their final wiki locations. Rejected files are moved back to _raw/ for manual editing.

2k tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
3103
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/Ar9av/obsidian-wiki --skill wiki-stage-commit

The instruction itself

15 sections, as written by the author

Wiki Stage Commit — Staged Write Promotion

You are reviewing LLM-written pages that are waiting in _staging/ for human approval before they land in the live wiki. This skill is only useful when WIKI_STAGED_WRITES=true in the vault config.

Before You Start

  • Resolve config — follow the Config Resolution Protocol in llm-wiki/SKILL.md. This gives OBSIDIAN_VAULT_PATH and WIKI_STAGED_WRITES.
  • If WIKI_STAGED_WRITES is not set or is false, tell the user: "Staged writes mode is not enabled. Set WIKI_STAGED_WRITES=true in your .env to use this feature." Then stop.
  • Read the _staging/ directory inventory.

Invocation Forms

/wiki-stage-commit               # interactive review: show each file and ask accept/reject
/wiki-stage-commit --all         # accept all staged files without per-file review
/wiki-stage-commit --reject-all  # reject all staged files (move to _raw/ for manual editing)
/wiki-stage-commit --list        # list staged files with summary, no changes

Step 1: Inventory Staged Files

Glob $OBSIDIAN_VAULT_PATH/_staging/**/*.md — these are the pending pages.

Also glob $OBSIDIAN_VAULT_PATH/_staging/**/*.patch.md — these are pending *updates* to existing pages (diff-style files showing proposed additions and deletions).

Report the inventory:

Staged files: 4 new pages, 2 updates

New pages:
  _staging/concepts/attention-mechanism.md        (ingested 2 days ago)
  _staging/entities/andrej-karpathy.md            (ingested 2 days ago)
  _staging/skills/fine-tuning-llms.md             (ingested yesterday)
  _staging/references/attention-is-all-you-need.md (ingested 3 hours ago)

Updates (patch files):
  _staging/concepts/transformer-architecture.patch.md  (target: concepts/transformer-architecture.md)
  _staging/skills/prompt-engineering.patch.md          (target: skills/prompt-engineering.md)

If _staging/ is empty, report: "Nothing staged. All writes have been committed or no staged writes have been produced yet."

Step 2: Per-File Review (interactive mode)

For each staged file (new pages first, then updates):

For new pages:

Display a summary:

--- New page: concepts/attention-mechanism.md ---
Title:    Attention Mechanism
Tags:     #ml #architecture
Summary:  Core building block of transformers — computes weighted sum of values based on query-key similarity.
Tier:     supporting
Confidence: 0.72
Sources:  papers/attention.pdf

[Preview first 20 lines of body]
...

Accept [a], Reject [r], Skip [s], Preview full [p]?

For patch files:

Display a structured diff:

--- Update: concepts/transformer-architecture.md ---
Source: _staging/concepts/transformer-architecture.patch.md

Proposed additions (+):
+ Transformers outperform RNNs on tasks requiring long-range dependencies. ^[inferred]
+ New source: papers/survey-2026.pdf

Proposed deletions (-):
- The attention mechanism was first described in [Bahdanau 2015].  (to be replaced by updated claim)

⚠️  Conflict check: target page was modified 3 days after staging. Review carefully.

Accept [a], Reject [r], Skip [s], Preview full diff [p]?

If --all flag is set, skip prompting and accept every file.

If --reject-all flag is set, skip prompting and reject every file.

If --list flag is set, stop after printing the inventory (Step 1).

Step 3: Apply Decisions

Accepting a new page

  • Move _staging/<category>/page.md<category>/page.md (the final location)
  • Update index.md with the new page entry
  • Remove the staged file

Accepting a patch/update

  • Read the current page at the target path
  • Apply the proposed additions and deletions (merge, don't just overwrite)
  • Update the updated frontmatter timestamp
  • Update index.md if the summary changed
  • Remove the staged patch file

Rejecting a file

Move it to $OBSIDIAN_VAULT_PATH/_raw/ for manual editing:

  • _staging/concepts/page.md_raw/rejected-concepts-page.md
  • _staging/concepts/page.patch.md_raw/rejected-patch-concepts-page.md
  • Prefix with rejected- so the user can identify it

Conflict detection on patch accept

Before applying a patch, check whether the target page's updated frontmatter is newer than the patch file's own updated field:

  • If the target was modified AFTER the patch was staged, warn: ⚠️ Conflict: target was updated since this patch was staged. Applying may lose recent changes.
  • Give the user a chance to abort: Apply anyway [y], Skip [s], Reject [r]?

Step 4: Update Tracking Files

After processing all staged files:

  • hot.md — update the Recent Activity section: "Committed N staged pages; rejected M."
  • log.md — append:
   - [TIMESTAMP] STAGE_COMMIT accepted=N rejected=M skipped=K

Step 5: Report

Stage commit complete.

✅  Accepted (N):
  concepts/attention-mechanism.md     → now live
  entities/andrej-karpathy.md         → now live
  concepts/transformer-architecture.md → updated (patch applied)

❌  Rejected (M):
  skills/fine-tuning-llms.md          → moved to _raw/rejected-skills-fine-tuning-llms.md

⏭️  Skipped (K):
  references/attention-is-all-you-need.md → still in _staging/

Staging queue: K files remaining

Notes

  • Staged files use the same page template as live pages — they are ready to land, just awaiting approval
  • Patch files use a human-readable diff format: lines starting with + are additions, lines starting with - are deletions
  • index.md and log.md are always updated immediately on ingest (they are low-risk tracking files) — only category pages go through staging
  • The _staging/ directory is not tracked by Obsidian's graph view — pages only appear in the wiki after promotion

Other skills for the same job

different authors, same section of the catalogue
Doc Coauthoring
by anthropics
vendor ×10

Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.

4k tokens
Changelog Generator
by frostant
×9

Automatically creates user-facing changelogs from git commits by analyzing commit history, categorizing changes, and transforming technical commits into clear, customer-friendly release notes. Turns hours of manual changelog writing into minutes of automated generation.

774 tokens
Test Driven Development
by w95
×7

Use when implementing any feature or bugfix, before writing implementation code

2k tokens
Writing Plans
by ZhanlinCui
×4

Use when you have a spec or requirements for a multi-step task, before touching code

816 tokens
Writing Skills
by ZhanlinCui
×4

Use when creating new skills, editing existing skills, or verifying skills work before deployment

26k tokens scripts
Crafting Effective Readmes
by softaworks
×3

Use when writing or improving README files. Not all READMEs are the same — provides templates and guidance matched to your audience and project type.

15k tokens
Humanizer
by softaworks
×3

| Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.

6k tokens
Opentrons Integration
by christophacham
×3

Official Opentrons Protocol API for OT-2 and Flex robots. Use when writing protocols specifically for Opentrons hardware with full access to Protocol API v2 features. Best for production Opentrons protocols, official API compatibility. For multi-vendor automation or broader equipment control use pylabrobot.

9k tokens scripts

How to use it

Copy the folder

Take ar9av/wiki-stage-commit from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.