knowledge-claim-consolidation skill
Review and consolidate Knowledge claims over a chosen scope - the unlinked residue after relinking, the claims on one hub or over-budget page, or the claims about one subject since a date. File claims on their most specific subject page, create pages for substantial subjects, retire duplicates, supersede status snapshots and replaced claims by an existing claim, and review pages so each reads as a current explanation of its subject, leaving current pages alone. Use for claim cleanup, consolidation, page review, or the Knowledge Consolidation scribe.
Is the knowledge-claim-consolidation skill safe?
Clean: nothing in its files matched our rules. We read 22 files in the folder on 2026-11-14.
No findings.
Install the knowledge-claim-consolidation skill
A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description.
git clone --depth 1 https://github.com/openclaw/openclaw.git /tmp/openclaw mkdir -p ~/.claude/skills cp -r /tmp/openclaw/extensions/knowledge/skills/knowledge-claim-consolidation ~/.claude/skills/knowledge-claim-consolidation
In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub
The instructions your agent would load
SKILL.md as published, without the frontmatter. Read it on GitHub
Claim consolidation review
Claims pile up faster than pages are curated: knowledge_remember saves claims that name no page, and hubs collect snapshots, duplicates and claims about more specific subjects. This review files claims on their pages, retires duplicated and replaced claims, and keeps pages current.
Rules card
These hold for the whole run; keep them after compaction.
text, retiring a cited claim or dropping a page from a claim's entities, make sure the page still reaches every record it cites, and that each citation points at a current record.
- Never lose a citation or reachable evidence. Before deleting or moving
from a complete, fresh read, and reread before each edit.
- Read before writing. Read a record before changing it; edit a body only
record's revision and hash, and the latest expectedOntologyFingerprint from your last receipt. A row's 16-character hash prefix fences an annotate, withdraw or supersede; a body edit needs the full read.
- Fences: every write carries a fresh requestId, expectedVaultId, the
fit its type gets fixes only: broken citations, a stale fact a newer claim contradicts. Restructuring or splitting needs a named reason in the report.
- Leave alone by default. A page with a current State and sections that
untouched and report them. Unattended, never ask.
- Never guess. Leave uncertain identities, duplicates and outcomes
retry under a new request ID.
- Uncertain write: read recovery first; never
Tasks
Read setup, then call knowledgestatus and knowledgeontology. Read only the reference for the task at hand, and follow its links only when you reach that step. You don't need to reread a reference after compaction unless you are about to use it again.
linking.
- Enumerate a scope, link claims, create missing pages:
retiring.
- Retire duplicates, replaced claims and snapshot series:
to subject pages and the one layout for the page's type.
- Review a page: page review, which sends you
Scopes
Name the scope in the report and run its tasks in this order:
its subject): enumerate, move claims about a more specific subject, retire, then review the page. Move before retiring: a superseded or withdrawn claim can no longer be relinked.
- Unlinked residue (default; scribe runs): link, then retire.
- One page (a hub, an over-budget page, or a body that no longer explains
review its page if the new claims change its picture.
- One subject, optionally since a date: enumerate, link, retire, then
Verify and report
Run knowledge_verify with expectedEntities for linked claims and expectedLinks for new or edited pages (verify), then enumerate the scope again and handle or report new arrivals.
Report: scope; claims examined, linked, superseded (with the superseding claim) and withdrawn (with the survivor); pages created and why; pages reviewed with their result (no change, fixed, or restructured with its reason); dated sections and Timeline entries retired and kept; and each claim or line you left and why.
Scribe runs
The Knowledge Consolidation scribe runs unattended; its prompt handles access and the automation. Do not ask questions. Each step names its task:
empty, report zero claims examined and go on.
- Link the unlinked residue (linking). If it is
(retiring), then check the residue again.
- Retire duplicates and replaced claims in it
(retire a written Timeline). Find them with knowledge_grep for Timeline and confirm the heading.
- Retire written Timelines: up to 10, largest first
(scribe page choice), whatever the residue's size.
- Review pages: up to 5, one at a time
withdrawn and left; Timelines retired and left; pages reviewed, changed and left unchanged.
- Report. The status line counts claims examined, linked, superseded,
More skills from openclaw/openclaw
- A1passwordSet up and use 1Password CLI for sign-in, desktop integration, and reading or injecting secrets.
- Aacp-routerRoute plain-language requests for Claude Code, Cursor, Copilot, OpenClaw ACP, OpenCode, Gemini CLI, Grok Build, Qwen, Kiro, Kimi, iFlow, Factory Droid, Kilocode, or explicit ACP harness work into either OpenClaw ACP runtime sessions or direct acpx-driven sessions ("telephone game" flow). For coding-agent thread requests, read this skill first, then use only `sessions_spawn` for thread creation. Codex chat binding defaults to the native Codex app-server plugin unless ACP is explicit or background spawn needs ACP.
- Aadd-model-providerAdd and live-prove a model provider with non-interactive config one-liners, without exposing credentials.
- Aagent-transcriptRequested GitHub PR/issue agent transcripts: redact, trim, preview, and insert safely.
- Aapple-notesCreate, view, edit, delete, search, move, or export Apple Notes via the memo CLI on macOS.
- Aapple-remindersList, add, edit, complete, or delete Apple Reminders and reminder lists via remindctl.
- Aauto-qaRun OpenClaw-wide autonomous QA and live/stress campaigns across independent subsystem lanes, with verified fixes and a resumable evidence report.
- AautoreviewStructured code review through the shared OpenClaw agent-skills installation.
- Abear-notesCreate, search, and manage Bear notes via grizzly CLI.
- Ablock-kitUse proactively for structured or interactive Slack replies, and when asked to author or validate native Slack Block Kit JSON.
- AblogwatcherMonitor blogs and RSS/Atom feeds for updates using the blogwatcher CLI.
- AblucliBluOS CLI (blu) for discovery, playback, grouping, and volume.