Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.

Hash-anchored read/edit/undo_last_edit tools for DeepSeek Harness (dsh). Every line gets a unique 3-char hash (A-Za-z0-9) that stays stable across edits; stale or ambiguous anchors are rejected, never fuzzy-matched. Undo persists across restarts.

Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.

Enables capturing a tamper-evident, hash-chained local record of every AI agent tool call — both the decision before a tool runs and its outcome — so you can review what the agent actually did in order, detect edits, deletions, or reordering at the exact index, and export a signed evidence pack any third party can verify with no installs.

Read-only checks for money that agents move. verify_receipt: schema, Ed25519 signature, mandate expiry, amount = quantity x unit_price, refund not above payment. audit_spend: duplicate charges, price escalation, failed-but-billed, runaway loops, missing refunds, model overpay, budget breaches - three separate amounts, never one. reserve_report: NAV with an evidence hash per line; refuses a backing

Discover Agent Skills in configured local directories, plugin marketplaces and Git sources. Load them through DeepSeek Harness or a local stdio MCP server, with catalog budgets, content hashes and delivery receipts.

Discover Agent Skills in configured local directories, plugin marketplaces and Git sources. Load them through DeepSeek Harness or a local stdio MCP server, with catalog budgets, content hashes and delivery receipts.

Deterministic pre-execution audit for trading agents. PASS/WAIT/FAIL, reproducible verdict_hash.

Open-source antivirus for AI agents: block risky tools, secret access, prompt injection, malicious packages, MCP servers, plugins, and skills at runtime.

Enables agents to ingest local registry snapshots, detect removed or newly deprecated tools and schema-breaking changes, verify deprecation claims against hash-chained evidence, and generate migration diffs for agent configurations.

Enables agents to download single model weight files from Hugging Face repositories or Civitai model versions directly into allowlisted ComfyUI model folders. Source identifiers, destination categories, and file extensions are validated, and existing files are never overwritten.

The open-source memory and observability layer for AI agents — persistent memory, loop detection, hash-chained audit trails, and a live dashboard, automatic on pip install.