Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add knowz-io/deepseek-knowz-plugin --skill knowz-flushgit clone --depth 1 https://github.com/knowz-io/deepseek-knowz-pluginWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/knowz-io/deepseek-knowz-plugin/knowz-flush)<a href="https://agentmods.dev/skills/knowz-io/deepseek-knowz-plugin/knowz-flush"><img src="https://agentmods.dev/badge/skills/knowz-io/deepseek-knowz-plugin/knowz-flush.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00052 | $0.01165 |
| Opus 5 | $0.00026 | $0.00583 |
| Sonnet 5 | $0.00010 | $0.00233 |
| Haiku 4.5 | $0.00005 | $0.00117 |
Grade A, and why
knowz-flush scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/knowz-flush — Process pending captures
Drain the canonical project-root knowz-pending.md queue into Knowz vaults. Each queued block declares an Operation (create, amend, or update) and an Idempotency Key. Never replay an amend/update as create.
If enterprise.json exists in the project root, use its brand value instead of "Knowz" in user-facing text.
If Knowz MCP tools are unavailable, stop after reporting the queue count and: "{brand} MCP not connected. In Grok Bot Plugins or Cursor Marketplace, search Knowz → Add → Authorize. Do not paste API keys in chat."
Instructions
- Read both queue locations before doing anything:
- Canonical:
knowz-pending.mdin the project root. - Legacy migration source:
knowzcode/pending_captures.md. If neither contains a----delimited capture block, report that there is nothing to flush and stop.
- Canonical:
- Normalize and migrate legacy blocks without loss:
- Treat any keyless block in project-root
knowz-pending.mdas a legacy producer block. Normalize it in place before replay; do not reject queues produced by older/knowz save,/knowz amend, or/knowz autoversions. - Map
Target Vault Typeto canonicalTarget Vault; the value remains a routing token resolved throughknowz-vaults.md. - If
Payloadis missing but legacyContentexists, mapContenttoPayload. A block with neither is malformed and remains untouched. - Preserve
Operation,KnowledgeId,Semantic Key,Intent,Category,Source, title, and payload exactly. - Preserve an existing
Idempotency Key. If absent, create a stable migration key from the exact operation,KnowledgeIdor semantic key, target-vault token, intent, source, and title. Never include migration time or attempt number. - If that key already exists canonically with identical mutation content, treat the legacy block as an already-migrated duplicate. If the same key has different operation, target, identity, or payload, leave both blocks untouched and report
IDEMPOTENCY_KEY_COLLISION. - For a keyless canonical-root block, insert and confirm the derived key before any MCP call. For a nested legacy block, append a converted block to
knowz-pending.mdbefore removing its legacy source block. Remove the legacy block only after the canonical append is confirmed. Keep the legacy file itself as a deprecation notice when empty.
- Treat any keyless block in project-root
- Parse canonical blocks. Required fields are
Operation,Idempotency Key,Target Vault, andPayload;KnowledgeIdis additionally required foramend/update.Queue Statusdefaults topending;supersededblocks are preserved but MUST NOT be written.Semantic Key,Intent,Category, andSourceare optional metadata. A missingOperationmay migrate ascreateonly when the legacy block has noKnowledgeId, no amend/update action, and contains a complete create payload; otherwise leave it queued as malformed. - Group blocks by
Idempotency Key:- Byte-equivalent logical mutations are one retry group and are executed at most once.
- A key reused for different operation, identity, target, or payload is a collision. Execute none of that group.
- Read
knowz-vaults.mdand resolveTarget Vaultas an explicit ID/name or configured routing token. Empty or ambiguous resolution leaves the block queued. - Verify the MCP tools required by the queued operations before processing them. Create replay also requires a Knowz search capability; amend/update replay requires
get_knowledge_itemplus the matching mutation tool. If a required tool is unavailable, leave that operation queued rather than weakening replay safety. - Preflight every retry group before mutation:
- create: search the resolved vault by the exact title and compare content/semantic identity. One materially identical item means the operation already succeeded; count it as idempotently reconciled. No match permits create. A similar-title item with conflicting content or multiple plausible matches is ambiguous and remains queued.
- amend: fetch
KnowledgeId. If missing, leave queued asMISSING_AMEND_TARGET; never create. If the requested delta is already reflected exactly, count it as reconciled. If applicability is ambiguous, leave queued for review; otherwise call amend once. - update: fetch
KnowledgeId. If missing, leave queued asMISSING_UPDATE_TARGET; do not silently create during replay. If the full replacement is already reflected, count it as reconciled; otherwise call update once.
- After a confirmed success or idempotent reconciliation, remove every byte-equivalent block in that retry group from both queue files. If mutation succeeds but queue cleanup cannot be confirmed, stop and report the exact key so the next run performs the preflight again.
- Report newly created/amended/updated counts, idempotently reconciled counts, migrated legacy count, superseded/quarantined count, and every remaining key grouped by collision, malformed block, ambiguous target, missing target, unavailable tool, or MCP failure.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 40 lines · 52 tokens per session scan A 5193f63bc769
knowz-flush is a skill published in the GitHub repository knowz-io/deepseek-knowz-plugin (1 stars, last pushed 20d ago), licensed MIT. It adds 52 tokens to every session and 1,165 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
krouter-obsidian
KRouter Obsidian — route knowledge-dependent tasks to the smallest current evidence set in an Obsidian vault using deterministic local search. Use when a request depends on durable user preferences, prior corrections or superseded claims, long-term project history or results, reusable methods or evidence, or the…
knowledge-wiki
Manage the personal knowledge wiki. Use when the user shares articles, documents, or asks to organize knowledge; when a conversation produces insights worth preserving as structured knowledge; or when the user asks about the knowledge base.
modlens
Plug-in vision for text-only models. Hard rule: when a file path or URL with an image extension (.png, .jpg, .jpeg, .webp, .gif, .heic, .heif) appears anywhere in the conversation (typed by the user, injected as a [Image: source: ] line, or inside a tag) and you cannot see that image's content, run this skill on it…
j-space
Use this skill to establish and operate the model's inner workspace — the J-space — for any task that needs more than fluent output: multi-step or chained reasoning, planning, long-horizon and agentic work, competition-level problems, complex debugging, keeping many parts of a deliverable globally consistent, holding…
install-openviking-memory
Install and configure the OpenViking long-term memory plugin for OpenClaw via natural conversation. Once installed, the plugin automatically captures facts from chats and recalls relevant context before each reply (auto-capture + auto-recall, cross-session). Covers prerequisites, install through OpenClaw's plugin…
openviking-context-database
Use OpenViking from OpenClaw through @openviking/openclaw-plugin: long-term memory, session archives, resource and Agent Skill import, semantic recall, recall trace debugging, and externalized tool-result recovery. Prefer this skill when the user wants to use, query, debug, or operate OpenViking context from an…