Search across all indexed project knowledge (copilot-instructions, skills docs, runbooks, session docs). Uses hybrid BM25 keyword + semantic search with temporal decay. Returns the most relevant chunks with source file, section, and line numbers.

List all knowledge sources indexed by ContextEngine, each with a one-line summary (from the file's own head: frontmatter description, title plus first sentence, or module docstring), status (found/missing) and chunk counts. Read the summary to pick the right source before calling read_source.

Read the full content of a specific knowledge source by name.

Force a full re-index of all knowledge sources. Use after adding new files or changing contextengine.json.

Discover and analyze all projects in the workspace. Shows tech stack (framework, runtime, key dependencies), infrastructure (git, docker, pm2), and git remote status for each project. Requires Pro license.

Scan all projects for port declarations (ecosystem.config.js, docker-compose.yml, .env, package.json) and detect port conflicts. Returns a port allocation map with conflict warnings. Requires Pro license.

Run the Compliance Agent audit across all projects. Checks: port conflicts, git remotes (origin + gdrive), git hooks (post-commit auto-push), .env files (existence + gitignore), Docker config (restart policy, workdir), PM2 config (treekill, kill_timeout, no bash wrappers), version issues (EOL runtimes, outdated deps, MUI v4/v5 coexistence). Returns a structured plan with findings and remediation steps.

Score one or all projects on AI-readiness (0-100%). Checks documentation (copilot-instructions, README, CLAUDE.md, .cursorrules, SKILLS.md, .env.example), infrastructure (git, hooks, Docker, CI, deploy scripts, PM2), code quality (tests, TypeScript, linting, npm scripts), and security (.env gitignored, secrets exposure, lockfiles). Returns letter grade (A+ to F) with detailed breakdown.

Save a key-value entry to a named session. Use to persist decisions, context, plans, and findings between coding sessions. Each session can hold multiple keys (e.g., 'summary', 'active_tasks', 'decisions'). Keys are updated in place if they already exist.

Load a previously saved session by name. Returns all stored key-value entries with timestamps. Use at the start of a session to restore context from a previous conversation.

List all saved sessions. Shows session names, entry counts, and timestamps. Use to discover what context is available from previous conversations.

Delete a saved session by name. Returns success/not-found. Use for cleanup of stale or obsolete session context.

Verify the integrity of the local audit log chain. Returns OK + record count, or BROKEN + break index when a record has been edited or the chain otherwise diverges. Produces evidence aligned with SOC 2 CC7.2 (change monitoring) and ISO 27001 A.12.4.1 (event logging) — evidence artifacts, not a certification (OpsContext is not itself SOC 2– or ISO 27001–certified; see docs/compliance/). The audit log lives at ~/.contextengine/audit.log and records every state-changing operation (learning save/delete/import, session save/delete, activation activate/deactivate) as a hash-chained JSONL line.

Multi-agent cost report from Claude Code's own transcripts on this machine: tokens moved (cache read/write, fresh input, output), valued cost at API list prices (marked NOTIONAL on a subscription, UNPRICED when no rate matches), capacity intensity (subagents, failed, died at window, tool calls per agent, cache reuse), top runs, and context_burn / fanout_without_canary signals. Call it after a fan-out to read what it consumed, or before one to compare with the last. Thresholds come from .contextengine/policy.json agent_cost, else built-in defaults.

Returns active drift / loop / stuck-tool / fabrication / silent-failure signals detected over the recent audit-log window. Use to self-check before starting a major task phase. If any 'critical' signal is active (fabrication_suspect or silent_failure), pause and surface to the human.

MUST be called before ending any coding session. Checks all project repos for uncommitted changes, verifies documentation freshness (copilot-instructions.md, SKILLS.md, session docs), and returns a checklist of required actions. Will report PASS/FAIL for each check. The AI agent should resolve all FAIL items before ending.

Save a permanent operational rule learned during a coding session. Unlike sessions (ephemeral), learnings persist forever and auto-surface in search_context results so AI agents don't repeat mistakes. Duplicate rules (same category + rule text) are updated in place.

List all permanent learnings, optionally filtered by category. Shows operational rules that have been discovered across sessions. Use search_context to find learnings by keyword — they're automatically included in search results.

Delete a learning by its ID. Use list_learnings first to find the ID of the learning you want to remove.

Bulk-import learnings from a Markdown or JSON file. By default only MARKED learnings are imported: inline bullets with a [category] prefix, anything inside a *LEARNINGS.md file, anything under a heading that says learnings / lessons / gotchas / rules, and JSON arrays of {category, rule, context}. Set permissive=true to also import every H3 heading, bold bullet and table row (H2=category, H3=rule, bullets=context). Deduplicates against existing learnings.

Activate a ContextEngine Pro license to unlock premium tools (score_project, run_audit, check_ports, list_projects, HTML reports). Get a license at https://api.compr.ch/contextengine/pricing

Check current ContextEngine license status, plan, and available premium tools.