{"name":"Awareness","tagline":"The Cognitive Memory Cloud for AI Agents","description":"Persistent, searchable memory for AI agents with Auto-Cognition — workflow rules that teach agents when to remember, recall, and record automatically.","website":"https://awareness.market","github":"https://github.com/everest-an/Awareness-SDK","updated_at":"2026-08-10","capabilities":{"mcp_tools":[{"name":"awareness_init","description":"Load cross-session project memory and context","detail":"Load cross-session project memory. Returns session_id, user_preferences (personal preferences, identity, career — highest priority, present FIRST), context (narratives, tasks, knowledge, active_skills, attention_summary), and usage rules. Call ONCE at session start.\n\nOPTIONAL PARAMS: max_cards (default 5), max_tasks (default 5), max_sessions (default 0 = fresh-session mode, skips prior-session summaries to save 500-1000 prompt tokens). Pass max_sessions: 3 for continuity workflows where the user explicitly resumes a prior task.\n\nATTENTION PROTOCOL: When context.attention_summary.needs_attention is true, review all attention items (stale tasks, high risks) and automatically: (1) remind the user about stale tasks, (2) warn about high risks before starting work, (3) update resolved items via awareness_record."},{"name":"awareness_get_agent_prompt","description":"Fetch full activation prompt for a specific agent role (sub-agent spawning)","detail":"Fetch the full activation prompt for a specific agent role. Call before spawning a sub-agent to get its dedicated system prompt."},{"name":"awareness_recall","description":"Search memory — pass ONE query string, daemon picks the rest","detail":"Search persistent memory by passing a single natural-language query. Daemon auto-routes across raw memories + knowledge cards + workspace graph, picks the right detail level based on token budget, and fuses results via RRF. Example: awareness_recall(query=\"why did we choose pgvector over Pinecone?\"). Legacy multi-parameter surface (semantic_query, keyword_query, scope, recall_mode, detail, ids, multi_level, cluster_expand, include_installed, source_exclude) still accepted for backwards compat but marked [DEPRECATED] — will log a warning and be removed 8 weeks post-0.8.0."},{"name":"awareness_lookup","description":"Structured data: tasks, knowledge, risks, timeline","detail":"Fast structured data lookup (<50ms). Types: context, tasks, knowledge, risks, session_history, timeline, handoff."},{"name":"awareness_record","description":"Save memory — pass ONE content string, daemon handles the rest","detail":"Save what you did/decided/learned to persistent memory. Pass a single content string and the daemon defaults action=remember and triggers client-side salience-aware extraction asynchronously. Example: awareness_record(content=\"Today I decided to switch from Pinecone to pgvector because...\"). Advanced actions (remember_batch, update_task, submit_insights) still require explicit action= for backwards compat."},{"name":"awareness_mark_skill_used","description":"Report skill usage outcome — success, partial, or failed","detail":"Mark a skill as used after applying it. Pass outcome='success' (default), 'partial', or 'failed'. Success resets decay timer fully. Partial gives a reduced boost. Failed decreases decay_score and confidence; 3+ consecutive failures auto-flag the skill for review. Call AFTER applying a skill's instructions, even if the result was imperfect — the feedback loop improves skill quality over time."},{"name":"awareness_facts","description":"Read bi-temporal facts from the knowledge graph (active or as_of)","detail":"Query the F-074 bi-temporal fact store. Default returns only active facts (valid_to IS NULL). Pass as_of=<ISO ts> to rewind to a point in time. Optional filters: subject_id, predicate. Use this when you need ground-truth structured claims, not narrative recall."},{"name":"awareness_related","description":"n-hop graph traversal — find entities related to a starting entity","detail":"Walk the knowledge graph up to max_hops (1-4) from a given entity_id. Returns linked entities with the predicate path. Use when you need connection discovery: 'what is X related to', 'who knows X', 'what depends on X'."},{"name":"awareness_timeline","description":"Bi-temporal fact stream for one entity (active + superseded)","detail":"Returns the full time-machine view for a single entity, ordered by valid_from desc — both active and superseded facts. Use this for 'how did X evolve over time' / 'what did we believe about X last month' queries."},{"name":"awareness_find_deals","description":"Search the public Open Deal Board (business supply and demand)","detail":"Search a PUBLIC board of business supply and demand broadcasts. Use when the user wants to find a counterparty — a supplier, buyer, spare capacity, inventory or partner (e.g. \"who has H100s in Singapore\", \"找液冷机柜托管\").\n\nNOT a memory tool: it neither reads nor writes the user memory, and it needs NO API key, so it works even when Awareness memory is unconfigured.\n\nPARAMS: query (free text, works in Chinese and English), direction (\"supply\" | \"demand\"), category, region, limit (1-50, default 20).\n\nREPORTING RULES — Awareness vouches for no one and issues no badges. Report what publisher.evidence shows (days active, broadcasts published, deals confirmed by both parties, upheld reports) and let the user judge. provenance.anchored means the record is anchored on-chain (ERC-8350): it proves the record existed then and is unaltered — it does NOT mean the contents are true, so never call an anchored listing \"verified\" or \"guaranteed\". Contact details are never returned; link the user to the listing URL.\n\nNo-MCP fallback for other assistants: https://awareness.market/deals.md returns the whole board as ~3 KB of Markdown and accepts the same filters as query params."}],"recall_modes":[{"id":"hybrid","description":"Structured + vector search (default, recommended)"},{"id":"precise","description":"Fast targeted search for specific facts"},{"id":"session","description":"Expand results to full session history"},{"id":"structured","description":"Zero-LLM pure database lookup (<50ms)"},{"id":"auto","description":"Intent-detected mode selection"}],"write_actions":[{"id":"remember","description":"Record a single event with inline insights"},{"id":"remember_batch","description":"Batch record at session end"},{"id":"backfill","description":"Import past conversations"},{"id":"ingest","description":"Bulk data import (files, docs)"},{"id":"update_task","description":"Create or update tracked tasks"},{"id":"submit_insights","description":"Structured knowledge extraction"}],"lookup_types":["context","tasks","knowledge","risks","session_history","timeline","handoff","rules","graph","agents"],"knowledge_categories":{"engineering":["problem_solution","decision","workflow","key_point","pitfall","insight","skill"],"personal":["personal_preference","important_detail","plan_intention","activity_preference","health_info","career_info","custom_misc"]}},"marketplace":{"description":"Browse, install, and hire from Awareness's agent & memory marketplace — AI agents built from real memory, ready to deploy.","url":"https://awareness.market/market","agent_hiring":"https://awareness.market/hired","memory_packs":"https://awareness.market/market"},"deals":{"description":"Live supply & demand board of business deals (AI hardware, compute, colocation, logistics). Machine-readable at /deals.md, JSON API at /api/v1/public/deals, per-listing schema.org Offer/Demand.","url":"https://awareness.market/deals","markdown":"https://awareness.market/deals.md","api":"https://awareness.market/api/v1/public/deals","human_page":"https://awareness.market/deals"},"differentiators":[{"feature":"Auto-Cognition","description":"Workflow rules injected into IDE config (CLAUDE.md, .cursor/rules/, etc.) so agents automatically init, recall, and record — no manual orchestration needed."},{"feature":"Multi-Level Retrieval","description":"Searches across chunk, broader context, topic expansion, and structured knowledge card levels for the right depth of context per query."},{"feature":"Structured Insights","description":"Automatically extracts 13 categories of knowledge cards from raw events: decisions, solutions, workflows, pitfalls, skills, preferences, and more."},{"feature":"Memory Market","description":"Community marketplace for knowledge templates. Install expert knowledge into your agent's memory with one click. Cross-memory search included."},{"feature":"Active Skills","description":"Reusable procedures loaded at session start. Agents inherit proven workflows without re-deriving them each time."}],"integrations":{"protocols":["MCP (Model Context Protocol)","REST API","Python SDK","TypeScript SDK"],"ide_agents":["Claude Code","Cursor","Windsurf","Cline","GitHub Copilot","Codex","Kiro","Trae","Zed","JetBrains (Junie)","Augment","Google AntiGravity (Jules)","Manus","ChatGPT","Gemini CLI","Devin","OpenClaw"],"agent_frameworks":["LangChain","CrewAI","PraisonAI","AutoGen / AG2"],"plugins":[{"name":"Claude Code Plugin","url":"https://awareness.market/plugins/claude-code"},{"name":"OpenClaw Plugin","url":"https://awareness.market/plugins/openclaw"}]},"getting_started":{"fastest":{"command":"npx @awareness.market/setup","description":"One-command setup: auto login → select memory → write MCP config + workflow rules"},"python_sdk":{"install":"pip install awareness-memory-cloud","description":"Python client with interceptor pattern for OpenAI/Anthropic clients"},"typescript_sdk":{"install":"npm install @awareness-sdk/memory-cloud","description":"TypeScript/Node.js client"},"mcp_config":{"description":"Add to your IDE's MCP settings","example":{"mcpServers":{"awareness":{"url":"https://awareness.market/mcp/{your-api-key}"}}}}},"pricing":[{"tier":"Free","price":"$0/mo","memories":2,"api_calls":"1K/mo","storage":"500 MB"},{"tier":"Starter","price":"$5/mo","memories":5,"api_calls":"10K/mo","storage":"5 GB"},{"tier":"Pro","price":"$12/mo","memories":20,"api_calls":"100K/mo","storage":"50 GB"},{"tier":"Team","price":"$29/mo","memories":"Unlimited","api_calls":"500K/mo","storage":"200 GB","seats":5},{"tier":"Enterprise","price":"Custom","description":"Private deployment, SSO/SAML, dedicated support"}],"links":{"documentation":"https://awareness.market/docs","developers":"https://awareness.market/developers","mcp_tools_reference":"https://awareness.market/docs?doc=MCP_TOOLS_REFERENCE","python_sdk_docs":"https://awareness.market/docs?doc=python","typescript_sdk_docs":"https://awareness.market/docs?doc=typescript","memory_market":"https://awareness.market/market","deal_board":"https://awareness.market/deals","deal_board_markdown":"https://awareness.market/deals.md","deal_board_api":"https://awareness.market/api/v1/public/deals","agent_hiring":"https://awareness.market/hired","pricing":"https://awareness.market/pricing","faq":"https://awareness.market/faq","blog":"https://awareness.market/blog","enterprise":"https://awareness.market/enterprise","discord":"https://discord.gg/nMDrT538Qa","llms_txt":"https://awareness.market/llms.txt"}}