Feature Claim Registry¶

This page is the single index of all Feature Claims for the OLAV platform. Each Claim corresponds to a verifiable platform capability.

Methodology: Doc-Driven Testing → Fresh Venv Validation (updated v0.17.0)

Document claim → Build wheel → Fresh venv install → CLI verification → Verified or Narrowed

v0.17.0 audit: 12 of 118 claims (10%) that were ✅ in dev environment failed in fresh venv. All claims are now re-verified from pip install dist/*.whl in a clean environment. See dev_docs/45. CLAIM_AUDIT_v0.17.md for the full audit report.

Status legend:

✅ Verified — Passed CLI end-to-end verification, recorded in dev_docs/23. CLI_VERIFICATION_LOG.md
⬜ Pending — Code implemented, awaiting CLI verification
⚠️ Narrowed — Available under limited conditions; Claim scope has been narrowed
🔶 Env-Blocked — Requires an external environment (running service, CLAB, etc.) to verify

Level 1 — Platform Core Capabilities¶

ID	Claim	Doc Page	Status
C-L1-01	OLAV lets you control infrastructure with natural language	Home	✅

Level 2 — Feature-Level Claims¶

Basic Commands¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-01	CLI correctly reports version and system info	`olav version`	Installation	✅	v0.10.0
C-L2-02	CLI lists all available Agents with descriptions	`olav list`	First Query	✅	v0.10.0
C-L2-03	List and switch the active workspace	`olav workspace list/use`	Manage Workspaces	✅	v0.10.0
C-L2-13	Initialize project directory structure + admin user + embedding model	`olav init`	Installation	✅	v0.17.0
C-L2-27	Reset Agent conversation history	`olav reset --agent <name>`	CLI Commands	✅	v0.10.0

Skill Management¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-04	Install a Skill from a local directory (single or multi-workspace)	`olav skill install <path>`	Build a Skill	✅	v0.17.0
C-L2-25	Install a Skill from a Git URL	`olav skill install <url>`	Build a Skill	✅	v0.10.0
C-L2-06	Append tools to an existing workspace	`olav skill install --merge-into`	Build a Skill	✅ ⚠️	v0.10.0
C-L2-36	List and view installed skill status	`olav skill list/status`	Manage Workspaces	✅	v0.10.0
C-L2-37	Skills declaring Python dependencies auto-create an isolated venv	`requires_packages` in MANIFEST	Build a Skill	✅	v0.10.0

Service Registry¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-19	Register an OpenAPI service with one command	`olav registry register <url>`	Connect a Service	✅	v0.10.0
C-L2-20	DevOps agent generates environment-aware automation scripts	`olav --agent devops "write script..."`	DevOps Agent	✅	v0.13.0

Audit and Logs¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-08	Query audit logs and list recent operations	`olav log list/errors`	Audit and Logs	✅	v0.10.0
C-L2-22	Export audit data in training formats	`olav log export trajectory/sft/atif`	Audit and Logs	✅	v0.10.0
C-L2-12	Multi-user concurrent audit with no write conflicts	5 concurrent `olav log list`	Audit and Logs	✅	v0.10.0

Background Services¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-16	Web service provides browser UI and REST API (HTTP + HTTPS cookie support)	`olav service web start`	Background Services	✅	v0.17.0
C-L2-17	Daemon accelerates CLI responses	`olav service daemon start`	Background Services	✅	v0.10.0
C-L2-18	Syslog receiver accepts network device logs (SO_REUSEADDR)	`olav service logs start`	Background Services	✅	v0.17.0

HTTP API¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-29	SSE streaming query endpoint	`POST /runs/stream`	HTTP API	✅	v0.10.0
C-L2-30	Multi-turn conversation thread management	`POST /threads`	HTTP API	✅	v0.10.0

Interactive Mode¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-14	TUI interactive mode supports multi-turn conversations (via deepagents-cli Textual)	`olav` (no arguments)	First Query	✅	v0.17.0
C-L2-15	Resume a previous session	`olav --session <id>`	First Query	✅	v0.10.0
C-L2-24	/trace-review analyzes failure patterns and writes to memory	`/trace-review`	Self-Improving Loop	⚠️	v0.10.0 — TUI migration pending
C-L2-26	Skip tool call confirmation	`olav --auto-approve`	Using Agents	✅	v0.10.0

Knowledge Base and Memory¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-21	Knowledge base indexes documents and supports semantic search	`olav --agent config "index..."`	Knowledge Base	✅	v0.10.0

Users and Authentication¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-23	Admin creates/manages users and tokens	`olav admin "add-user..."`	Users and Roles	✅	v0.10.0
C-L2-28	Supports multiple authentication modes	`auth.mode` config	Security Model	✅	v0.10.0

Core Agent Tools¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-31	Core Agent executes Python code	`olav --agent core "run python:..."`	Core Agent	✅	v0.10.0
C-L2-32	Core Agent executes SQL queries	`olav --agent core "what tables..."`	Core Agent	✅	v0.10.0
C-L2-33	Core Agent web search	`olav --agent core "search:..."`	Core Agent	✅	v0.10.0
C-L2-34	Core Agent executes shell commands	`olav --agent core "run: df -h"`	Core Agent	✅	v0.10.0
C-L2-35	Config Agent workspace health check	`olav --agent config "health check"`	Core Agent	✅	v0.10.0

Export and Integration¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-10	Export Claude-compatible plugin	`olav export claude-plugin`	CLI Commands	✅	v0.10.0
C-L2-11	Auto-truncate large API responses	Python SDK	CLI Commands	✅	v0.10.0

Configuration¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L2-38	Assign different LLM models to different Agents	`agent_overrides` in api.json	Configuration	✅	v0.10.0
C-L2-39	Environment variables override config file	`OLAV_LLM_*`	Configuration	✅	v0.10.0
C-L2-40	`olav log show <run-id>` displays full event sequence	`olav log show <id>`	Audit and Logs	✅	v0.10.0
C-L2-41	Service management: status / start --all / stop --all	`olav service status/start --all/stop --all`	Background Services	✅	v0.10.0
C-L2-42	Web API introspection: /openapi.json, /docs, /threads/search	`GET /openapi.json` `GET /docs`	HTTP API	✅	v0.10.0
C-L2-43	`olav workspace status` shows Agent status	`olav workspace status`	Manage Workspaces	✅	v0.10.0
C-L2-44	Admin rotate-token and add-user --expires	`olav admin "rotate-token <n>"`	Users and Roles	✅	v0.10.0
C-L2-45	Audit database queryable directly via DuckDB	`duckdb .olav/databases/audit.duckdb`	Audit and Logs	✅	v0.10.0
C-L2-46	`/model <name>` switches LLM model within a session	TUI: `/model gpt-4o-mini`	CLI Reference	✅	v0.10.0
C-L2-47	`@file.txt` file content injection	TUI: `@/path/to/file.txt`	CLI Reference	✅	v0.10.0
C-L2-48	`!command` shell passthrough	TUI: `!echo hello`	CLI Reference	✅	v0.10.0
C-L2-49	`--sandbox modal/daytona/runloop` remote sandbox flag	`olav --sandbox modal version`	Agent Harness	✅	v0.10.0
C-L2-50	`olav registry list/refresh/status`	`olav registry list`	Connect a Service	✅	v0.10.0
C-L2-51	`olav config evolve --list/--approve`	`olav config evolve --list`	CLI Reference	✅	v0.10.0

NetOps Extension Claims¶

Quick Start¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-01	`/netops_init --dry-run` passes environment checks + validates Nornir inventory	`/netops_init --dry-run`	Quick Start	✅	v0.17.0
C-NE-02	`olav list` includes all 9 agents (core + 8 netops) after single `skill install`	`olav list`	Quick Start	✅	v0.17.0
C-NE-03	`/netops_init` populates `netops.parsed_outputs` with platform-aware commands	`/netops_init`	Quick Start	✅	v0.17.0
C-NE-04	`/netops_init` generates topology links + populates `netops.devices`	`/netops_init`	Quick Start	✅	v0.17.0

Concepts¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-05	Snapshots produce unique `snapshot_id` shared across tables	`execute_sql`	Concepts	✅	v0.10.0
C-NE-06	Custom TextFSM templates take priority over NTC built-ins	manual test	Concepts	✅	v0.10.0
C-NE-07	`CommandRegistry.reload()` picks up new templates without restart	Python API	Concepts	✅	v0.10.0
C-NE-08	`netops.devices` populated by Device ETL from parsed_outputs (hostname, ip, platform, vendor, model, os_version)	`SELECT * FROM netops.devices`	Database Schema	✅	v0.17.0
C-NE-09	BGP data queried directly from `netops.parsed_outputs` (no auto-generated views)	`SELECT parsed_data FROM netops.parsed_outputs WHERE command='show ip bgp summary'`	Database Schema	✅	v0.17.0
C-NE-10	`netops.topology_links` includes `discovery_protocol`	`SELECT discovery_protocol FROM netops.topology_links`	Database Schema	✅	v0.10.0
C-NE-11	`netops.raw_output_store` preserves latest raw CLI output	`SELECT * FROM netops.raw_output_store LIMIT 5`	Database Schema	✅	v0.10.0
C-NE-12	`nornir/hosts.yaml` defines device inventory	manual test	Configuration	✅	v0.10.0
C-NE-13	`blacklisted_commands.yaml` supports regex; invalid patterns skipped	manual test	Configuration	✅	v0.10.0
C-NE-14	`cron_schedules.yaml` customizes trace_learner schedule	`crontab -l`	Configuration	✅	v0.10.0

Core Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-15	Natural language query returns results from snapshot data	`olav "R1 BGP neighbors?"`	Core Agent	✅	v0.10.0
C-NE-16	Blacklisted commands rejected by `execute_cli`	manual test	Core Agent	✅	v0.10.0
C-NE-17	`/learn_cmd` executes command and generates TextFSM template	`/learn_cmd "show version" -d R1`	Core Agent	✅	v0.10.0
C-NE-18	`take_snapshot` triggers collection + auto-ingest into DB	`take_snapshot`	Core Agent	✅	v0.17.0
C-NE-19	Core Agent suggests Ops Agent for complex analysis	complex query	Core Agent	✅	v0.10.0

Ops Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-20	Ops routes analysis queries to Analysis subagent	`olav --agent ops "analyze..."`	Ops Agent	✅	v0.10.0
C-NE-21	Ops routes device check queries to Probe subagent	`olav --agent ops "check..."`	Ops Agent	✅	v0.10.0
C-NE-22	What-If simulation recalculates connectivity after link failure	`olav --agent ops "simulate..."`	Analysis	✅	v0.10.0
C-NE-23	Analysis uses only `run_python_simulation`, no live device access	SKILL.md inspection	Analysis	✅	v0.10.0
C-NE-24	`execute_cli_parallel` validates whitelist before execution	manual test	Probe	✅	v0.10.0
C-NE-25	Device names restricted to `[a-zA-Z0-9_\-.]`	manual test	Probe	✅	v0.10.0
C-NE-26	`diff_topology_drift` detects link state changes	`olav --agent ops "what changed..."`	Diff	✅	v0.10.0
C-NE-27	`diff_sql_state` compares any table between snapshots	`olav --agent ops "compare..."`	Diff	✅	v0.10.0

Lab Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-28	`deploy_lab` builds ContainerLab topology from snapshot data	`olav --agent ops-lab "deploy lab"`	Lab Agent	🔶	v0.10.0
C-NE-29	Lab agent pushes SRL config and verifies BGP/OSPF convergence	`olav --agent ops-lab "push config and verify"`	Lab Agent	🔶	v0.10.0
C-NE-30	`deploy_lab` auto-handles SRL topology.yml quirks and veth injection	internal tool	Lab Agent	🔶	v0.10.0
C-NE-31	Lab agent has no `execute_cli` — no live device access	code inspection	Lab Agent	✅	v0.10.0

Audit Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-32	Designer guides profile creation interactively	`olav --agent audit-designer "create..."`	Audit Agent	✅	v0.10.0
C-NE-33	Auditor executes profile and outputs Markdown report	`olav --agent audit "run..."`	Audit Agent	✅	v0.10.0
C-NE-34	Designer validates table/column existence before SQL	internal validation	Audit Agent	✅	v0.10.0
C-NE-35	`analyze_thresholds` provides P50/P90/P95 recommendations	internal tool	Audit Agent	✅	v0.10.0

Infra Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-36	`olav registry register` registers service and generates API reference	`olav registry register <url>`	Infra Agent	🔶	v0.12.0
C-NE-37	`olav --agent infra` queries registered services via `api_request`	`olav --agent infra "query..."`	Infra Agent	🔶	v0.12.0
C-NE-38	`api_request` returns clear error for unregistered services	`api_request("nonexistent", ...)`	Infra Agent	✅	v0.12.0
C-NE-39	Write operations require `--enable-api-write` + dry-run + mandatory approval	`--enable-api-write`	Infra Agent	🔶	v0.12.0
C-NE-40	`network_isolation=True` sandbox hard-blocks HTTP write operations	sandbox test	Infra Agent	✅	v0.12.0

DevOps Agent¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-NE-41	DevOps agent queries `netops.devices` before generating scripts	`olav --agent devops "write..."`	DevOps Agent	✅	v0.13.0
C-NE-42	Generated scripts include `--dry-run` + error handling + summary	script inspection	DevOps Agent	✅	v0.13.0
C-NE-43	Scripts use real device names/IPs from database, not placeholders	script inspection	DevOps Agent	✅	v0.13.0
C-NE-44	Scripts exported to `exports/scripts/` as executable files	file check	DevOps Agent	✅	v0.13.0

Level 3 — Knowledge Store (UKS)¶

Top-Level Capability¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L3-UKS	Unified Knowledge Store: single LanceDB table for documents, memories, and graph nodes	`olav kb status`	Knowledge Base	✅	v0.13.0

Schema & Migration¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-01	Memory table has `origin`, `confidence`, `tags` columns	unit test	Knowledge Base	✅	v0.13.0
C-KB-02	`add_memory()` uses correct defaults when new params omitted	unit test	Knowledge Base	✅	v0.13.0
C-KB-03	Old-schema tables auto-migrate on first access	unit test	Knowledge Base	✅	v0.13.0
C-KB-09	`migrate_memory_table()` infers `origin`/`confidence` from `metadata.source`	unit test	Knowledge Base	✅	v0.13.0
C-KB-25	`migrate_kb_chunks()` moves `kb_chunks` rows to the unified memory table	unit test	Knowledge Base	✅	v0.13.0
C-KB-26	`from olav.core.knowledge` raises `ImportError` after deletion	unit test	Knowledge Base	✅	v0.13.0
C-KB-27	`search_knowledge_lancedb` tool file is deleted	unit test	Knowledge Base	✅	v0.13.0

Capture & Decay¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-04	Extracted items contain a `tags` array	unit test	Knowledge Base	✅	v0.13.0
C-KB-05	`capture` writes `origin="agent"`, `confidence=importance` from LLM response	unit test	Knowledge Base	✅	v0.13.0
C-KB-06	`document` origin does NOT decay	unit test	Knowledge Base	✅	v0.13.0
C-KB-07	`agent` origin decays normally	unit test	Knowledge Base	✅	v0.13.0
C-KB-08	High-frequency recall (`access_count >= 5`) halves the decay rate	unit test	Knowledge Base	✅	v0.13.0

Graph¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-10	`materialize_graph()` returns a dict with `nodes` and `edges` keys	unit test	Knowledge Base	✅	v0.13.0
C-KB-11	Vector similarity creates implicit `similar` edges with a `weight` field	unit test	Knowledge Base	✅	v0.13.0
C-KB-12	Shared tags create explicit edges through entity nodes	unit test	Knowledge Base	✅	v0.13.0
C-KB-15	`graspologic` unavailable → `cluster_knowledge()` returns `None`	unit test	Knowledge Base	✅	v0.13.0

Export & Sync¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-13	`export_obsidian()` generates `.md` files with YAML frontmatter	unit test	Knowledge Base	✅	v0.13.0
C-KB-14	`export_visjs()` generates an HTML file with vis.js content	unit test	Knowledge Base	✅	v0.13.0
C-KB-17	`sync_from_files()` UPDATEs entries that have changed on disk	unit test	Knowledge Base	✅	v0.13.0
C-KB-18	`sync_from_files()` INSERTs entries for new files	unit test	Knowledge Base	✅	v0.13.0
C-KB-19	`sync_from_files()` DELETEs entries when the file is removed	unit test	Knowledge Base	✅	v0.13.0
C-KB-20	`sync_from_files(dry_run=True)` does NOT modify data	unit test	Knowledge Base	✅	v0.13.0

Import & Search¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-21	Importing a `.md` file creates a memory entry with `origin='document'`, `confidence=1.0`	unit test	Knowledge Base	✅	v0.13.0
C-KB-22	`hybrid_search()` returns results from both `agent` and `document` origins	unit test	Knowledge Base	✅	v0.13.0
C-KB-28	After migrating `kb_chunks`, `search_by_vector()` returns entries with `origin='document'`	unit test	Knowledge Base	✅	v0.13.0

CLI Commands¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-KB-16	`olav kb export` command is available and exits successfully	`olav kb export`	Knowledge Base	✅	v0.13.0
C-KB-23	`olav kb status` outputs knowledge store statistics	`olav kb status`	Knowledge Base	✅	v0.13.0
C-KB-24	`olav kb graph` generates an HTML file	`olav kb graph`	Knowledge Base	✅	v0.13.0

Level 4 — Multi-User & Security¶

ID	Claim	CLI Command	Doc Page	Status	Version
C-L4-01	`olav init` auto-creates admin user + writes token to `~/.olav/token` + sets `auth.mode=token`	`olav init`	Users and Roles	✅	v0.14.0
C-L4-02	`olav admin add-user` validates that the Linux user exists before creating a token	`olav admin add-user <name>`	Users and Roles	✅	v0.14.0
C-L4-03	`olav sessions` lists current-user sessions from the audit database	`olav sessions`	Users and Roles	✅	v0.14.0
C-L4-04	`olav --resume <thread_id>` resumes a session across interfaces	`olav --resume <id>`	Users and Roles	⬜	v0.14.0
C-L4-05	Thread ownership: non-owner requests return HTTP 403; admin bypasses	`POST /runs/stream`	Security Model	✅	v0.14.0
C-L4-06	`search_logs` tool queries syslog Parquet files via DuckDB	agent tool	Background Services	✅	v0.14.0
C-L4-07	API server emits startup warning when `auth.mode=none`	`olav service web start`	Security Model	✅	v0.14.0

v0.17.0 New Capabilities¶

ID	Claim	CLI Command	Status	Version
C-L2-52	`olav init` auto-downloads embedding model silently	`olav init`	✅	v0.17.0
C-L2-53	`olav skill install` supports multi-workspace via `workspaces:` list	`olav skill install <path>`	✅	v0.17.0
C-L2-54	`olav skill install` auto-runs `pip install -e` when `pyproject.toml` present	`olav skill install <path>`	✅	v0.17.0
C-L2-55	`POST /reload` hot-reloads agent singleton after skill install	`POST /reload`	✅	v0.17.0
C-L2-56	`GET /agents` returns dynamic agent list from workspace registry	`GET /agents`	✅	v0.17.0
C-L2-57	`GET /memory/graph` serves knowledge graph vis.js page	`GET /memory/graph`	✅	v0.17.0
C-L2-58	CIDR allowlist middleware for web API	api.json `security.allowed_cidrs`	✅	v0.17.0
C-L2-59	Semantic cache with auto-invalidation on data ingest	`olav "query"` (2nd time)	✅	v0.17.0
C-L2-60	Embedding dimension auto-detected from model probe	`detect_embedding_dim()`	✅	v0.17.0
C-L2-61	Single-query mode uses native langgraph streaming (no deepagents-cli dependency)	`olav "query"`	✅	v0.17.0
C-NE-45	Platform-aware SSH collection (Junos + IOS command lists)	`/netops_init`	✅	v0.17.0
C-NE-46	Device ETL extracts IP from `show ip interface brief` / `show interfaces terse`	`/netops_init`	✅	v0.17.0
C-NE-47	Dry-run validates full Nornir inventory (catches group reference mismatches)	`/netops_init --dry-run`	✅	v0.17.0

Statistics¶

Status	Count
✅ Verified (wheel)	125
⬜ Pending	1
⚠️ Narrowed	1
🔶 Env-Blocked	6
Total	133

Verification Process¶

# For each ⬜ Claim:
# 1. Run the CLI command
olav <command for the claim>

# 2. Record the result in dev_docs/23. CLI_VERIFICATION_LOG.md
# 3. Pass → change status to ✅, add version number
# 4. Fail → fix code, narrow the claim, or remove the Claim

# For 🔶 Claims: set up the environment first (LLM API Key / running external service)