Argus captures only durable conversation and project-state changes, routing them to the correct Git-backed Archivum record.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install argus-skill
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install argus-skill using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Argus?
Argus is an Openclaw Skills context-preservation workflow for turning meaningful conversation changes into durable records without archiving entire conversations. It identifies decisions, corrections, preferences, commitments, evidence, project state, blockers, methods, and next actions, while excluding filler, duplicate summaries, unsupported claims, and unauthorized secrets.
The skill routes each durable delta to its canonical owner across configured Archivums, commitment systems, budget authorities, and specialist archives. It respects workspace contracts and version-2 configuration, maintains bidirectional home anchors, validates archive links, and reports exactly what changed. Argus is designed for repository-aware agents that need reliable long-term memory without creating duplicate ledgers or silently expanding scope.
Argus Use Cases
- Archive a meaningful conversation milestone or completed work.
- Capture explicit decisions, corrections, stable preferences, or changed priorities.
- Record promises, deadlines, waiting-on states, current workstream custody, and next actions.
- Preserve project evidence, research results, reusable methods, sources, and generated outputs.
- Update an Archivum-style workspace while preserving its configured directory and file layout.
- Close the loop before ending work that changed durable project state.
- Validate backlinks and archive boundaries across multiple Git-backed records.
- Provide durable memory for repository-aware agents using Openclaw Skills.
How Argus Works
- Identify only the durable delta from the conversation or completed work, separating actionable state from filler, repetition, speculation, and unauthorized secrets.
- Discover the registry and candidate Archivums using the installed skill's discovery script, preferring an explicit registry, then
ARCHIVUM_REGISTRY, then$ARCHIVUM_HOME/00_meta/archivum_registry.toml. - Ask for clarification when multiple archives are plausible and no registry resolves ownership; never silently guess or create a new Archivum.
- Route each fact by ownership using the routing contract, including commitment systems, budget authorities, home archives, research archives, admin archives, career archives, and existing specialist stores.
- Read the selected Archivum's
AGENTS.md,config.yaml, profile, state, and schema only as needed, resolving all paths from the configured map rather than assuming directory names. - Search the existing worktree and records before creating new content, then apply the workspace's commitment and resource contracts where applicable.
- Write each delta to one canonical owner, updating multiple surfaces only when routing requires it.
- Maintain the bidirectional home-anchor contract so private satellite records link to their home Archivum and remain reachable from it.
- Verify changed files, links, schemas, ledgers, and relevant tests; commit only when the archive contract or user requires it.
- Report changed durable records, intentionally untouched surfaces, and one next action when work remains, without publishing or sending anything merely because capture completed.
Argus Setup
- Install or enable the Argus skill in the Openclaw Skills environment and ensure its
scripts/andreferences/directories remain available relative to the installed skill. - Configure an Archivum registry through an explicit
--registryargument, theARCHIVUM_REGISTRYenvironment variable, or$ARCHIVUM_HOME/00_meta/archivum_registry.toml. - Ensure each target workspace exposes its agent contract and, for version-2 workspaces, an authoritative
config.yamlcontaining the relevantworkspace.*,directories.*, andcanonical_files.*mappings. - Discover available archives and their configuration details:
python3 scripts/discover_archivums.py --json
- For a specific registry, run discovery with an explicit path:
python3 scripts/discover_archivums.py --registry /path/to/00_meta/archivum_registry.toml --json
- Validate the archive graph and backlink boundaries with the read-only checker:
python3 scripts/check_backlinks.py --registry /path/to/00_meta/archivum_registry.toml
- Use
references/checkpoint-prompt.mdwhen making Argus available automatically in repository-aware agents. Preserve legacy workspace layouts as-is unless migration is explicitly requested.
Argus Data Schema & Taxonomy
Argus organizes durable state by canonical ownership rather than by keyword matching or conversation chronology.
| Data surface | Typical durable content | Ownership guidance |
|---|---|---|
| Commitment system | Promises, deadlines, waiting-on states, current workstream custody | Use the workspace-configured commitment contract |
| Budget authority | Token and GPU budgets, account windows, autonomous-run gates, session histories, artefact custody | Record resource usage only where the configured system requires it |
| Home Archivum | Cross-domain projects, identity, relationships, personal decisions, stable preferences, cross-archive index | Acts as the home anchor for connected records |
| Research Archivum | Hypotheses, literature, experiments, results, methods, scientific strategy | Use the configured research layout |
| Admin Archivum | Operations, governance, finance, legal, people operations, internal administration | Preserve the existing administrative contract |
| Career archive | Role research, applications, evidence packets, pipeline state | Use when career ownership is canonical |
| Specialist Archivum | Domain-specific project records | Prefer an existing specialist store |
Version-2 workspaces use config.yaml as the authoritative schema and path map. Argus resolves workspace.profile_file, workspace.state_file, workspace.schema_file, directories.*, and canonical_files.* before reading or writing. It does not assume numbered directories such as 01_projects; for example, a research workspace may map projects to 01_active_research.
Each private satellite record must link back to its home anchor, while the home Archivum must retain a route to the satellite. Backlink validation rejects logical paths that escape a registered archive root. The skill does not create a second commitment ledger, budget ledger, or recursive record stating that Argus performed an update.
Argus Advanced Features
- Ownership-based routing for deltas spanning multiple domains or having several plausible destinations.
- Bidirectional home-anchor links between home Archivums and private satellite records.
- Registry discovery that reports Archivum configuration versions and available agent contracts.
- Read-only backlink validation with archive-root boundary protection.
- Version-2 configuration support through authoritative
config.yamlpath mappings. - Archive-local
archivumCLI integration for deterministic creation, status, and doctor checks when available. - Commitment-system integration that avoids parallel promise or task ledgers.
- Budget-authority integration for token, GPU, autonomous-run, session, and artefact governance.
- Bounded context loading that reads only the selected archive's needed profile, state, schema, and records.
- Milestone-based batching that keeps the watcher quiet and prevents checkpointing after every edit or tool call.
- Legacy workspace preservation without automatic migration.
- Automatic-agent support through the supplied checkpoint prompt for repository-aware Openclaw Skills workflows.
SKILL.md
Loading
Related Openclaw Skills
ML Run Provenance
ML Run Provenance creates and validates self-describing machine learning run metadata at initialization so every execution has traceable intent, inputs, code identity, and lineage.
Butler
Butler provides local, project-scoped control for Claude usage estimates, GPU budgets, reservations, and post-run reconciliation.
Walking Calories Calculator
An evidence-based walking calorie calculator that converts body weight and step count into a transparent net calorie estimate using peer-reviewed biomechanics research.
Security Watch
Security Watch monitors CVEs and security advisories through the Chinng AI-Agent Portal and produces evidence-based, actionable vulnerability summaries.
Smart Medical Device IoMT Compliance
A compliance writing assistant that maps connected medical devices across hardware, software, sensors, apps, and cloud services to NMPA, FDA, and MDR requirements.
Everything Is a Battlefield | AI-HIVE Original Content Workflow
An AI-HIVE MCP workflow for planning, generating, reviewing, and delivering original animation and game content without reproducing protected IP.
Featured*
