Chimera v0.2.0 — security, adoption, intelligence, interop

Chimera v0.2.0 — the security → adoption → intelligence → interop cycle.

An open-source, self-evolving AI agent whose reasoning core is an LLM-Fusion engine (panel → judge → synthesizer). This release makes the security story measured, the setup out-of-the-box, memory & skills measurable, and adds the two winning agent protocols.

🔐 Security — measured, not asserted

• Quarantined reader (dual-LLM / CaMeL): a tool-less model extracts only schema-validated fields from untrusted content, so injected instructions never reach the privileged agent.
• Taint provenance on memories & learned skills — a skill distilled during a tainted run is held for review (the "Zombie Agents" anti-poisoning defense). A taint-adaptive allowlist narrows dangerous tools once a run is tainted; fetched content is data-fenced.
• Red-team suite (chimera redteam) reports the injection attack-success rate with vs without defenses: 100% → ~14% on the built-in corpus — and it names the remaining leak (exfiltration via an allowed tool) instead of claiming 100%.

🚀 Adoption — works out of the box

• chimera init — one command: create .env, set a key, verify, point at a real example.
• Consumer recipes: email_triage, research_brief, repo_watchdog + an MCP guide.
• Task-oriented docs site (mkdocs-material).

🧠 Intelligence — measurable

• chimera memory-bench — recall@k vs scale, lexical vs paraphrase. Surfaces the honest keyword ceiling (lexical 1.00 holds at 1000 facts, paraphrase 0.00).
• Opt-in semantic recall — embeddings bridge paraphrases; FTS fallback on any embedder error. Measured lift: paraphrase 0.00 → ~1.00.
• Skill retirement — under-performing skills proposed-with-review (reversible, never a silent delete).

🔌 Interop — speaks both winning protocols

• Chimera as an MCP server (chimera serve --mcp): solve / fuse / memory-search as MCP tools for any client (Claude Desktop, IDEs, other agents).
• A2A adapter (chimera serve --a2a): Agent Card + JSON-RPC task lifecycle, so a LangGraph/CrewAI/AutoGen orchestrator can delegate a task to Chimera. (Synchronous core; streaming/push not yet implemented.)

691 tests, mypy --strict clean, ruff clean. Apache-2.0. Full detail in CHANGELOG.md.

Full Changelog: v0.1.0...v0.2.0

Chimera v0.1.0

First public release of Chimera — an open-source AI agent that reasons by blending many models at once, does real work on its own, keeps learning, and runs anywhere (your laptop or a $5 server). Apache-2.0, self-hostable, fully yours.

What it does

• Many minds, one answer — LLM-Fusion runs a panel of models → a judge → a synthesizer, behind a cost-aware router (cheap turns use one model; hard problems get fused).
• It does the work, not just chat — plans, uses tools, checks its own output, and self-corrects with verify-or-revert (executable evidence is the ground truth, so a strict reviewer can't discard verified-correct work).
• Gets better the more you use it — layered memory (SQLite + FTS5 recall, cross-session user profile, LLM consolidation of fact clusters, preference nudges) plus skills it writes and refines for itself.
• Safe by design — a governance kernel (allow / warn / block / review), a static validator for self-modification, worktree isolation, and human-in-the-loop for anything destructive.

Highlights in this release

• Subagent layer — an isolated read-only Context Explorer (FastContext-style grep/glob that returns a compact file:line map), a generic spawn_subagent tool, tool-using role agents, and an IsolatedCrew that runs workers in parallel git worktrees with per-worker verify gates and an optional --synthesize supervisor.
• Runs on a server — serve --cron fires scheduled jobs on a real clock (proactive, not just reactive); Dockerfile + docker-compose + install.sh + a full deploy guide.
• Talks to people where they are — native Discord, Telegram, Slack, and Signal adapters; two-way WhatsApp via inbound webhook; webhook-triggered SOPs.
• Connects to anything — MCP client + OpenAPI→tool import + a reference tool library (code interpreter, email/IMAP, calendar, image/TTS, arXiv, and more).
• Onboarding from other agents — migrate config, skills, and long-term memory from Hermes & OpenClaw (memory is merged, not overwritten).

Quality

463 tests · mypy --strict clean · ruff clean · CI green on Python 3.11 / 3.12.

Get started

git clone https://github.com/brcampidelli/chimera-agent.git
cd chimera-agent
pip install -e .          # or: uv sync
cp .env.example .env      # add one provider key (an OpenRouter key unlocks the fusion panel)
chimera doctor

Full guide: docs/usage.md · Deploy: docs/deploy.md

Honest status

Chimera is alpha — it builds, runs, and is heavily tested, but has no production mileage yet. Start with low-stakes tasks, keep the guardrails on (--guard, CHIMERA_SANDBOX=docker), and tell us what breaks. Contributions and issues very welcome. 💜

🤖 Generated with Claude Code