How do I install the Wiki Reconcile skill?

Run npx skillmds@latest add yarlson/wiki-reconcile in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Wiki Reconcile safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Wiki Reconcile?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Wiki Reconcile free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Wiki Reconcile?

yarlson (@yarlson) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Wiki Reconcile by yarlson — Agent Skill

Name: Wiki Reconcile
Author: yarlson

Wiki Reconcile

Search, compare, and reconcile pending Yarbrain memory candidates with canonical semantic notes through reviewable patches. Use when processing inbox/memory candidates, resolving conflicting evidence, or applying an explicitly approved knowledge patch; do not capture sessions or promote procedures.

yarlson 78c2a85 3.4 KB Updated Sep 17, 2026

File contents

Reconcile Wiki Knowledge

Maintain current understanding without erasing evidence. A candidate is a claim to evaluate, not a fact to append.

Search Before Write

Do not create a file under notes/ until all of these steps succeed:

Resolve ../../scripts/yarbrain.py relative to this SKILL.md.
Search by proposed title, entities, aliases, project, and distinctive phrases. Repeat searches and raise the result limit when five results cannot cover every plausible overlap.
Read every plausible overlap with the helper's read command.
Compare scope, dates, confidence, volatility, status, and provenance.

If search is unavailable or ambiguous, stop without changing canonical notes.

Reconcile

Choose exactly one operation:

NOOP when the current note already represents the candidate accurately.
MERGE when the candidate adds compatible evidence or detail.
SUPERSEDE when newer evidence replaces a time-sensitive claim.
SPLIT when different projects, environments, users, or versions explain an apparent contradiction.
CONFLICT when credible evidence disagrees and cannot yet be resolved.
CREATE only when no existing note represents the concept.

Draft a focused patch before editing canonical knowledge. Hash the UTF-8 string with SHA-256 and use proposal- as its ID and Markdown filename. Digest the UTF-8 proposal body, excluding its digest field, with SHA-256, then store it beside its candidate under inbox/memory/. Show the operation, every target, search evidence, source episodes, changed current understanding, history effect, and rollback steps. Store unresolved conflicts under inbox/conflicts/. If the current request does not authorize a vault write, show the draft without storing it.

Apply a patch only after explicit approval, including a request that clearly names the reviewed proposal ID and digest. Any proposal change invalidates that approval. For a multi-file change, require a working rollback mechanism and restore the pre-edit state after partial failure. Preserve old provenance and record corrected or superseded claims in History. A canonical note must contain id, kind, scope, status, created, updated, verified_at, volatility, aliases, and sources frontmatter plus Current understanding, Why, Edge cases, Evidence, and History sections when those sections have content.

After an approved edit, read every target back and run lint. Restore the pre-edit state when the patch introduced a lint failure; report unrelated pre-existing issues separately. Mark the candidate reconciled with the applied proposal ID and then rebuild INDEX.md. An index failure does not invalidate a correct canonical edit because the index is derived; report it as stale and leave the approved note intact.

If the canonical edit succeeds but the candidate status write fails, keep the canonical edit, report incomplete finalization, and retry only the status write after verifying the applied proposal. Do not apply the canonical patch again.

Never silently resolve a conflict, remove the only evidence copy, rewrite an episode, or edit installed plugin skills.

---name: wiki-reconciledescription: Search, compare, and reconcile pending Yarbrain memory candidates with canonical semantic notes through reviewable patches. Use when processing inbox/memory candidates, resolving conflicting evidence, or applying an explicitly approved knowledge patch; do not capture sessions or promote procedures.---# Reconcile Wiki KnowledgeMaintain current understanding without erasing evidence. A candidate is a claimto evaluate, not a fact to append.## Search Before WriteDo not create a file under notes/ until all of these steps succeed:1. Resolve ../../scripts/yarbrain.py relative to this SKILL.md.2. Search by proposed title, entities, aliases, project, and distinctive phrases. Repeat searches and raise the result limit when five results cannot cover every plausible overlap.3. Read every plausible overlap with the helper's read command.4. Compare scope, dates, confidence, volatility, status, and provenance.If search is unavailable or ambiguous, stop without changing canonical notes.## ReconcileChoose exactly one operation:- NOOP when the current note already represents the candidate accurately.- MERGE when the candidate adds compatible evidence or detail.- SUPERSEDE when newer evidence replaces a time-sensitive claim.- SPLIT when different projects, environments, users, or versions explain an apparent contradiction.- CONFLICT when credible evidence disagrees and cannot yet be resolved.- CREATE only when no existing note represents the concept.Draft a focused patch before editing canonical knowledge. Hash the UTF-8 string<candidate-id><newline><operation> with SHA-256 and useproposal-<first-16-lowercase-hex-characters> as its ID and Markdown filename.Digest the UTF-8 proposal body, excluding its digest field, with SHA-256, thenstore it beside its candidate under inbox/memory/. Show the operation, everytarget, search evidence, source episodes, changed current understanding,history effect, and rollback steps. Store unresolved conflicts underinbox/conflicts/. If the current request does not authorize a vault write, showthe draft without storing it.Apply a patch only after explicit approval, including a request that clearlynames the reviewed proposal ID and digest. Any proposal change invalidates thatapproval. For a multi-file change, require a working rollback mechanism andrestore the pre-edit state after partial failure. Preserve old provenance andrecord corrected or superseded claims in History. A canonical note must containid, kind, scope, status, created, updated, verified_at, volatility, aliases, andsources frontmatter plus Current understanding, Why, Edge cases, Evidence, andHistory sections when those sections have content.After an approved edit, read every target back and run lint. Restore thepre-edit state when the patch introduced a lint failure; report unrelatedpre-existing issues separately. Mark the candidate reconciled with the appliedproposal ID and then rebuild INDEX.md. An index failure does not invalidate acorrect canonical edit because the index is derived; report it as stale andleave the approved note intact.If the canonical edit succeeds but the candidate status write fails, keep thecanonical edit, report incomplete finalization, and retry only the status writeafter verifying the applied proposal. Do not apply the canonical patch again.Never silently resolve a conflict, remove the only evidence copy, rewrite anepisode, or edit installed plugin skills.

Frequently asked questions