Source profileQuality 91/100

volcengine/OpenViking/examples/openclaw-plugin/skills/openviking-context-database/SKILL.md

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 OpenClaw agent. For first-time plugin installation, use the install-openviking-memory skill instead.

Source repository stars
33,046
Declared platforms
0
Static risk flags
0
Last source update
2026-08-25
Source checked
2026-08-25

Decision brief

What it does: where it fits

Use this skill after @openviking/openclaw-plugin is installed and configured. It describes the current OpenClaw plugin implementation, not the standalone OpenViking Python SDK.

Best for

    Not for

    • Tasks that require unconfirmed production actions or broad system permissions.
    • Environments where the pinned source and install steps cannot be inspected.

    Compatibility matrix

    Platform support, with evidence labels

    PlatformStatusEvidenceWhat to check
    CodexNot declaredNo explicit evidencePortability before use
    Claude CodeNot declaredNo explicit evidencePortability before use
    CursorNot declaredNo explicit evidencePortability before use
    Gemini CLINot declaredNo explicit evidencePortability before use
    Open the compatibility checker

    Installation

    Inspect first. Install second.

    The source command is displayed only when detected. A safe inspection prompt is always available so your agent can explain every action before execution.

    Source-detected install commandSource
    npx skills add https://github.com/volcengine/OpenViking --skill "examples/openclaw-plugin/skills/openviking-context-database"
    Safe inspection promptEditorial

    Inspect the Agent Skill "openviking-context-database" from https://github.com/volcengine/OpenViking/blob/41044af779015b12fa40dee49e8433ea14515dbe/examples/openclaw-plugin/skills/openviking-context-database/SKILL.md at commit 41044af779015b12fa40dee49e8433ea14515dbe. List every install step, command, network request, credential, file read/write, external action, and rollback step. Explain whether it fits my task. Do not install or execute anything until I approve.

    Workflow

    What the source asks the agent to do

    1. 01

      Scope and Safety Rules

      The plugin is remote-only. It talks to an existing OpenViking server through HTTP and does not start or manage openviking-server.

      The plugin is remote-only. It talks to an existing OpenViking server through HTTP and does not start or manage openviking-server.Do not invent OpenViking REST endpoints. Use the registered OpenClaw tools and commands described below.The agent-visible addresource tool is disabled by default (enableAddResourceTool=false). Do not use addresource during search, retrieval, URI reading, or search-result optimization. Use ovsearch and ovread in those flow…
    2. 02

      Current Architecture

      OpenClaw owns agent execution, prompts, and tool invocation. OpenViking owns long-lived context:

      OpenClaw owns agent execution, prompts, and tool invocation. OpenViking owns long-lived context:Long-term memories are usually extracted on /compact or on threshold-triggered commit. A fact mentioned in a fresh conversation may still be present as recent session context before it becomes a long-term memory.
    3. 03

      Configuration Quick Reference

      Core config lives under plugins.entries.openviking.config:

      Core config lives under plugins.entries.openviking.config:
    4. 04

      Tool Selection Guide

      Review the “Tool Selection Guide” section in the pinned source before continuing.

      Review and apply the “Tool Selection Guide” source section.
    5. 05

      Tool Interface Reference

      Semantic search over memories/resources. Use archive tools for session history.

      Semantic search over memories/resources. Use archive tools for session history.Notes: when targetUri is omitted, the plugin resolves a search plan from resourceTypes or configured recallTargetTypes, fetches more candidates than requested, deduplicates, filters leaf memories, reranks, and respects…Persist text immediately by writing a session and committing with wait=true.

    Permission review

    Static risk signals and limitations

    No configured static risk pattern was detected

    This is not proof of safety. Runtime behavior, indirect dependencies, and hidden external systems are outside the static scan.

    Evidence record

    Why each signal appears

    EvidenceSourceComputedTestedEditorial
    SignalValueEvidence typeMeaning
    Quality score91/100ComputedDocumentation, specificity, maintenance, and trust rules
    Repository stars33,046SourceRepository attention, not individual Skill quality
    Compatibility0 platformsSourceDeclared in the catalog source record
    Usage guideautomated source guideEditorialGenerated or reviewed according to the visible evidence level

    Pinned source

    Provenance and original SKILL.md

    Repository
    volcengine/OpenViking
    Skill path
    examples/openclaw-plugin/skills/openviking-context-database/SKILL.md
    Commit
    41044af779015b12fa40dee49e8433ea14515dbe
    License
    AGPL-3.0
    Collected
    2026-08-25
    Default branch
    main
    View the original SKILL.md

    OpenViking Context Database — OpenClaw Plugin Operator Skill

    Use this skill after @openviking/openclaw-plugin is installed and configured. It describes the current OpenClaw plugin implementation, not the standalone OpenViking Python SDK.

    Scope and Safety Rules

    • The plugin is remote-only. It talks to an existing OpenViking server through HTTP and does not start or manage openviking-server.
    • Do not invent OpenViking REST endpoints. Use the registered OpenClaw tools and commands described below.
    • The agent-visible add_resource tool is disabled by default (enableAddResourceTool=false). Do not use add_resource during search, retrieval, URI reading, or search-result optimization. Use ov_search and ov_read in those flows.
    • Use manual /add-resource, or add_resource only when it is explicitly enabled and the user explicitly asks to import, add, upload, save, or index a resource.
    • Use add_skill only when the user explicitly asks to import, add, install, or register an Agent Skill into OpenViking.
    • For local files and directories, pass the local path to the plugin tool. The plugin uploads them through /api/v1/resources/temp_upload; do not send raw local filesystem paths to a remote server yourself.
    • Never log or echo API keys. The plugin sends API keys as X-API-Key / setup probe headers and masks them in setup output.

    Current Architecture

    OpenClaw owns agent execution, prompts, and tool invocation. OpenViking owns long-lived context:

    LayerCurrent behavior
    assembleRebuilds compressed session history from OpenViking and injects relevant recall into the latest user message.
    afterTurnAppends only the new turn to the OpenViking session; may trigger async commit when pending_tokens >= tokenBudget * commitTokenThresholdRatio.
    compactRuns commit(wait=true), waits for archive/extraction completion, and reads back latest archive overview.
    ToolsMemory recall/store/forget, archive search/expand, resource/skill import/search, recall trace query, tool-result list/search/read.

    Long-term memories are usually extracted on /compact or on threshold-triggered commit. A fact mentioned in a fresh conversation may still be present as recent session context before it becomes a long-term memory.

    Configuration Quick Reference

    Read status first:

    openclaw openviking status --json
    openclaw config get plugins.entries.openviking.config
    openclaw config get plugins.slots.contextEngine
    

    Core config lives under plugins.entries.openviking.config:

    FieldDefaultPurpose
    baseUrlhttp://127.0.0.1:1933OpenViking HTTP endpoint. Can also come from OPENVIKING_BASE_URL / OPENVIKING_URL.
    apiKeyemptyOptional API key. Can also come from OPENVIKING_API_KEY.
    peer_roleassistantPeer identity mode: none, assistant, or person. Session messages use body peer_id; data-plane recall/search uses X-OpenViking-Actor-Peer.
    peer_prefixemptyOptional prefix for assistant peer_id / actor peer values when peer_role=assistant.
    accountId / userIdemptyAdvanced tenant identity headers for root-key or trusted deployments.
    targetUriviking://~/memoriesDefault search scope for legacy targeted memory search.
    autoCapturetrueAppend sanitized turn text to OpenViking sessions.
    captureModesemanticsemantic or keyword; affects server-side extraction filtering.
    captureMaxLength24000Max sanitized text length per captured turn.
    autoRecalltrueRun recall before replies and inject relevant context.
    recallTargetTypesuser,agentDefault target types when targetUri is omitted. Allowed: resource, user, agent.
    recallResourcesfalseCompatibility shortcut that appends resource to default recall targets when recallTargetTypes is unset.
    recallLimit6Max selected recall items.
    recallScoreThreshold0.15Min score after post-processing.
    recallMaxInjectedChars4000Total injected character cap; complete memories that do not fit are skipped.
    commitTokenThresholdRatio0.5Async-commit threshold as a fraction (0-1) of the model context window (e.g. 0.5 = 50%); 0 commits every turn.
    commitKeepRecentCount10Recent messages kept live after afterTurn commit. Compact always uses 0.
    bypassSessionPatternsemptyGlob-like session keys that completely bypass OpenViking (* segment, ** multi-segment).
    emitStandardDiagnosticsfalseEmit structured openviking: diag {...} lines.
    logFindRequestsfalseLog routing for find/session writes. Also enabled by OPENVIKING_LOG_ROUTING=1 or OPENVIKING_DEBUG=1.
    traceRecallfalseRecord recall traces in memory.
    traceRecallPersistfalsePersist recall traces as local JSONL files.
    traceRecallDir~/.openclaw/openviking/recall-tracesRecall trace directory when persistence is enabled.

    Normal setup command:

    openclaw openviking setup --base-url <OPENVIKING_URL> --api-key <API_KEY> --json
    

    Useful variants:

    openclaw openviking setup --base-url <URL> --api-key <KEY> --peer-prefix openclaw-prod --json
    openclaw openviking setup --base-url <URL> --api-key <ROOT_KEY> --account-id <ACCOUNT_ID> --user-id <USER_ID> --json
    openclaw openviking setup --base-url <URL> --api-key <KEY> --recall-target-types resource --json
    openclaw openviking setup --base-url <URL> --api-key <KEY> --allow-offline --json
    openclaw openviking setup --base-url <URL> --api-key <KEY> --force-slot --json
    

    Tool Selection Guide

    User intentUse
    “What did I say before?”, preferences, decisions, known factsmemory_recall
    “Remember this now”memory_store
    “Forget X”memory_forget
    Summary lacks an exact command/path/snippet from old chatov_archive_search, then ov_archive_expand if needed
    Import docs, PDFs, local dirs, URLs, Git repos, media attachmentsmanual /add-resource; add_resource only if enableAddResourceTool=true
    Import/register an Agent Skilladd_skill
    Search imported resources or skillsov_search
    Read an exact viking://... hit from ov_search or recall traceov_read
    Explain why recall/search returned somethingov_recall_trace
    A previous tool result shows only a preview/refopenviking_tool_result_list, openviking_tool_result_search, openviking_tool_result_read

    Tool Interface Reference

    memory_recall

    Semantic search over memories/resources. Use archive tools for session history.

    ParameterRequiredDescription
    queryYesSearch query.
    limitNoMax selected results. Defaults to recallLimit.
    scoreThresholdNoScore threshold 0..1. Defaults to recallScoreThreshold.
    targetUriNoExact search URI. If set, only this URI is searched.
    resourceTypesNoArray of resource, user, agent; used only when targetUri is omitted.

    Notes: when targetUri is omitted, the plugin resolves a search plan from resourceTypes or configured recallTargetTypes, fetches more candidates than requested, deduplicates, filters leaf memories, reranks, and respects recallMaxInjectedChars.

    memory_store

    Persist text immediately by writing a session and committing with wait=true.

    ParameterRequiredDescription
    textYesInformation source text.
    roleNoSession role, default user.
    sessionIdNoExisting OpenViking/OpenClaw session reference. If omitted, a temporary memory-store-* session is created.

    memory_forget

    Delete a memory.

    ParameterRequiredDescription
    uriNoExact viking://user/.../memories/... memory URI. Non-memory URIs are refused.
    queryNoSearch query when uri is unknown.
    targetUriNoSearch scope URI, default targetUri.
    limitNoSearch limit, default 5.
    scoreThresholdNoSearch threshold, default recallScoreThreshold.

    If query mode finds multiple candidates, report candidates and ask the user to choose the exact URI; do not delete ambiguous memories.

    ov_archive_search

    Keyword grep across archived original conversation messages of the current session.

    ParameterRequiredDescription
    queryYesSingle keyword or short phrase; prefer names, dates, file paths, commands, or distinctive nouns.
    archiveIdNoRestrict to one archive, e.g. archive_002.

    Try at least two concrete keyword variants before concluding archived detail is unavailable.

    ov_archive_expand

    ParameterRequiredDescription
    archiveIdYesArchive ID from [Archive Index], e.g. archive_002.

    Use after an archive search or when the [Archive Index] already identifies the archive likely to contain exact detail.

    add_resource

    Import resources into viking://resources/....

    This agent tool is disabled by default. Prefer manual /add-resource for resource ingestion. If enableAddResourceTool=true exposes the tool, use it only for explicit import/index requests and never as part of search/retrieval optimization.

    ParameterRequiredDescription
    sourceYesLocal path, OpenClaw media attachment path, directory path, public URL, or Git URL.
    toNoExact target URI, e.g. viking://resources/project-docs. Mutually exclusive with parent.
    parentNoParent URI under viking://resources. Mutually exclusive with to.
    reasonNoReason/note for import.
    instructionNoProcessing instruction for semantic extraction.
    waitNoWait for processing completion.
    timeoutNoTimeout in seconds when wait=true.

    The current OpenClaw tool exposes the parameters above. The underlying client also supports server-facing resource options such as strict, ignore_dirs, include, exclude, and preserve_structure for command/internal paths; do not pass them to the tool unless the registered schema exposes them.

    add_skill

    Import Agent Skills into viking://~/skills/....

    ParameterRequiredDescription
    sourceNoLocal SKILL.md path or skill directory. Exactly one of source or data is required.
    dataNoRaw SKILL.md content or an MCP tool dict. Exactly one of source or data is required.
    waitNoWait for processing completion.
    timeoutNoTimeout in seconds when wait=true.

    Agent Skill best practice: a skill should have precise frontmatter (name, trigger-oriented description, useful tags), clear scope boundaries, explicit “when not to use” guidance if needed, and executable steps with concrete parameters. Keep secrets out of skill content.

    ov_search

    ParameterRequiredDescription
    queryYesSearch query.
    uriNoSearch URI. Defaults to resources plus agent skills.
    limitNoMax results per scope, default 10.

    Use after importing resources/skills, or when the user asks to search OpenViking-managed knowledge.

    Important: ov_search returns OpenViking virtual URIs such as viking://resources/project-docs/api.md#chunk-3. These are not local file paths. Do not use filesystem read tools for them; call ov_read with the exact URI when full content is needed.

    ov_read

    Read full content for one exact OpenViking virtual URI through /api/v1/content/read.

    ParameterRequiredDescription
    uriYesExact viking://... URI returned by ov_search or recall trace results. openviking://... aliases and local file paths are refused.

    ov_recall_trace

    ParameterRequiredDescription
    turnNolatest or all, default latest.
    traceIdNoExact trace ID.
    sessionId / sessionKey / ovSessionIdNoFilter by OpenClaw/OpenViking session.
    sourceNoauto_recall, memory_recall, ov_search, or ov_archive_search.
    resourceTypesNoArray/string containing resource, user, agent.
    since / untilNoUnix timestamp bounds in milliseconds.
    includeContentNoRead selected/displayed URI content previews on demand.
    limitNoMaximum traces to return, default 20.

    Trace records exist only when traceRecall=true; persisted lookup requires traceRecallPersist=true and accessible trace files.

    Externalized tool-result tools

    Use when a preview contains viking://session/<session_id>/tool-results/<tool_result_id>.

    ToolParameters
    openviking_tool_result_listtool_name?, limit? (default 50)
    openviking_tool_result_searchtool_output_ref, query, limit? (default 20), context_chars? (default 300)
    openviking_tool_result_readtool_output_ref, offset? (default 0), limit? (default 20000, -1 accepted by server path for all remaining content)

    The plugin refuses to read/search a tool-result ref from another session.

    Slash Commands

    /add-resource ./README.md --to viking://resources/openviking-readme --wait
    /add-skill ./skills/install-openviking-memory --wait --timeout=30
    /ov-search "OpenViking install" --uri viking://resources/openviking-readme --limit=5
    /ov-recall-trace --turn latest --source auto_recall --include-content
    

    Command parsers support quoted args and flags. Resource-only flags are rejected for skill imports.

    Troubleshooting

    SymptomLikely causeFirst action
    configured=falseSetup did not persist configRe-run openclaw openviking setup ... --json; branch on JSON error.
    slotActive=falseAnother context engine owns the slot or gateway has stale stateInspect plugins.slots.contextEngine; use --force-slot only after user confirms.
    health.ok=falseServer unreachable or wrong baseUrl / keyCheck baseUrl, network, /health, and auth.
    No long-term memory after a fresh fact/compact or commit/extraction has not run, or server extraction failedUse memory_store for explicit remember/save/store intents; otherwise run /compact or wait for threshold commit, then check OpenViking server logs.
    Recall misses shared documentsresource target is not enabledUse memory_recall with resourceTypes:["resource"] or configure recallTargetTypes: ["resource"].
    Summary lacks exact detailArchive summary is too coarseUse ov_archive_search with concrete keywords, then ov_archive_expand.
    Large tool output preview is insufficientTool result was externalizedUse openviking_tool_result_search/read with the ref.
    Need to explain recall behaviorTrace disabled or no trace for that turnEnable traceRecall; optionally traceRecallPersist; query ov_recall_trace.

    Reference Docs in This Repo

    • README.md / README_CN.md: feature overview and quick start.
    • INSTALL.md / INSTALL-ZH.md: install, upgrade, uninstall, and JSON setup handling.
    • INSTALL-AGENT.md: agent-oriented installation workflow.
    • docs/openviking-openclaw-plugin-guide.md: comprehensive Chinese operator/developer guide.

    Frequently asked questions

    What to verify before installation and use

    What does the openviking-context-database source document cover?

    Use this skill after @openviking/openclaw-plugin is installed and configured. It describes the current OpenClaw plugin implementation, not the standalone OpenViking Python SDK.

    How do I install openviking-context-database?

    The source record exposes this install command: npx skills add https://github.com/volcengine/OpenViking --skill "examples/openclaw-plugin/skills/openviking-context-database". Inspect the command and pinned source before running it.

    Alternatives

    Compare before choosing

    Computed 10045,511

    coreyhaines31/marketingskills

    ab-testing

    When the user wants to plan, design, or implement an A/B test or experiment, or build a growth experimentation program. Also use when the user mentions "A/B test," "split test," "experiment," "test this change," "variant copy," "multivariate test," "hypothesis," "should I test this," "which version is better," "test two versions," "statistical significance," "how long should I run this test," "growth experiments," "experiment velocity," "experiment backlog," "ICE score," "experimentation program

    Computed 10029,034

    garrytan/gbrain

    bulk-ingestion

    End-to-end discipline for turning any large data source (audio libraries, email takeouts, document corpora, chat exports, API dumps) into brain pages at scale. The lifecycle spine: SCHEMA → ACCESS → TRIAL → EVALUATE → IMPROVE → CODIFY → TEST → SKILLIFY → BULK → MONITOR. State is tracked in a durable JSON manifest (see MANIFEST-PATTERN.md) so any crash, session boundary, or subagent fan-out resumes from ground truth instead of memory.

    Computed 10024,921

    alirezarezvani/claude-skills

    app-store-optimization

    App Store Optimization (ASO) toolkit for researching keywords, analyzing competitor rankings, generating metadata suggestions, and improving app visibility on Apple App Store and Google Play Store. Use when the user asks about ASO, app store rankings, app metadata, app titles and descriptions, app store listings, app visibility, or mobile app marketing on iOS or Android. Supports keyword research and scoring, competitor keyword analysis, metadata optimization, A/B test planning, launch checklist

    Computed 1005,241

    dotnet/skills

    migrate-vstest-to-mtp

    Migrates .NET test projects from VSTest to Microsoft.Testing.Platform (MTP). Use when user asks to "migrate to MTP", "switch from VSTest", "enable Microsoft.Testing.Platform", "use MTP runner", set OutputType=Exe only for test projects in Directory.Build.props, or mentions EnableMSTestRunner, EnableNUnitRunner, or UseMicrosoftTestingPlatformRunner. USE FOR: MTP behavioral differences vs VSTest (exit code 8, zero tests discovered, --ignore-exit-code, TESTINGPLATFORM_EXITCODE_IGNORE); centralizing