graqle

GraQle — query your architecture, prove your AI's decisions

Index any codebase as a knowledge graph so AI agents reason about architecture instead of grepping files. Every decision they make — at build-time or in production — gets a cryptographic receipt anchored to a public transparency log. One Python package, two surfaces: dev intelligence for engineers, runtime governance for regulators.

pip install graqle

Website · Quickstart · Runtime governance · EU AI Act docs · Changelog · VS Code Extension

Two surfaces, one substrate

	Build-time (dev intelligence)	Run-time (production governance)
Governs	how your AI writes code	what your deployed AI decides
Trigger	a code change	a production decision (loan, hiring, triage, …)
Emits	reviewed, impact-analysed, audit-logged changes	a tamper-evident, third-party-verifiable record per decision
Built on	typed code knowledge graph + multi-agent reasoning	Layer 5 cryptographic substrate (RFC 8785 JCS → RFC 6962 Merkle → ed25519 → Sigstore Rekor)
Status	GA	GA — attest() capture (v0.60.0) + FastAPI middleware / @governed (v0.61.0) + continuous anchoring worker graqle govern serve (v0.62.0)

Build-time governance proves we hold ourselves to this standard — GraQle is developed through its own governance. Run-time governance lets you hold your deployed AI to the same cryptographically-verifiable standard. Same substrate, both surfaces.

90-second quickstart

Build-time — query your codebase as a graph

# 1. Scan any codebase into a knowledge graph
graq scan repo .
# → typed graph: functions, classes, modules, imports, calls — full architecture mapped in seconds

# 2. Ask GraQle to audit it
graq run "find every authentication bypass risk"
# → Graph-of-agents activates across relevant nodes
# → Traces cross-file attack chains the LLM alone cannot see
# → Returns: confidence score + evidence trail + active nodes + tool hints

# 3. Fix it — GraQle shows exact before/after for each file (governed)

# 4. Teach it back — the graph never forgets
graq learn "cancel endpoint must require admin auth"
# → Lesson persists. Every future audit activates this rule.

Run-time — attach governance to a deployed AI in one line

from graqle.governance.runtime import GovernedRuntime

gov = GovernedRuntime(salt="your-deploy-salt")

def score_application(app):
    decision = model.predict(app)                # your deployed AI, untouched
    gov.attest(                                  # <-- the one added line
        domain="loan", model_id="credit-risk-v4",
        inputs={"applicant_ref": gov.pseudonymize_ref(app.id)},   # PII-safe
        output={"decision": decision.label, "reason_code": decision.reason},
    )
    return decision

Each call produces a durable, PII-safe governed record. Its leaf hash is computed with the same shipped primitive the build-time batcher uses, so a runtime record is byte-compatible with the cryptographic substrate (RFC 8785 JCS → RFC 6962 Merkle → ed25519 → Sigstore Rekor). Capture is out-of-band — it adds 0 ms to your write path.

See examples/runtime_attest_production_decisions.py and examples/runtime_govern_serve_anchoring.py.

Run it as a continuous service (v0.62.0)

# Long-lived anchoring worker — flushes batches + drains the replay queue every tick
graqle govern serve --config graqle.yaml

# Cron-style one-shot tick (single flush + single replay-drain)
graqle govern serve --once

# Article-72-style monitoring snapshot — JSON suitable for any external monitor
graqle govern health
# → { "running": true, "ticks": 47, "records_anchored": 3120, "replay_queue_depth": 0, ... }

The serve loop writes .graqle/govern.health.json atomically after every tick — pipe it into your existing monitoring (Prometheus, Datadog, an oncall dashboard, a simple curl).

Independently verifiable, by anyone. Committed batches anchor to the public Sigstore Rekor transparency log. Any third party can verify a record — auditor, regulator, counter-party — without access to your infrastructure, or ours. Verification doesn't depend on Quantamix staying online.

💰 Token economics — a worked case study

A 4-developer team on a 50,000-node enterprise codebase burns ~$40 per developer per day on flat-file AI-coding tokens in 2026. The same team using GraQle's substrate:

Scenario	Annual (4 devs)	Saving
Flat-file baseline (Cursor / Claude Code default)	$42,240	—
GraQle + frontier API (Sonnet 4.6)	$19,874	−53%
GraQle + local SLM (Year 2, 90% migrated)	$5,174	−88%

Every number is auditable. Every assumption is sourced (Anthropic pricing, Cursor power-user data, Microsoft's killed Claude Code pilot, NCBI biomedical-KG research showing >50% token reduction, Qwen3-Coder SWE-Bench benchmarks). Scale linearly to a 40-developer enterprise: ~$224k/year saved in Year 1, ~$371k/year in Year 2.

Plus six things Cursor / Copilot / Codex do not offer at any subscription tier: cryptographic audit trail, EU AI Act Article 26 readiness (€15M fine exposure), patent-defensible substrate, survive-vendor-disappearance, multi-agent governance, public Sigstore Rekor anchoring.

→ Read the full case study — math, sources, and a bash snippet to re-run it on your own team's numbers.

What is GraQle

A governance-led multi-agent reasoning system for code, with a built-in cryptographic audit substrate for the AI you ship to production. Scan any codebase into a persistent knowledge graph. Every module becomes a reasoning agent. Agents decompose, debate, and synthesize answers with clearance-level governance. Every change — and every production decision — is impact-analysed, gate-checked, and cryptographically committed.

AI assistants see files. GraQle sees architecture. That's why it catches the cross-file bugs they can't, and why its audit trail survives every level of tampering.

Built for engineering teams who need:

• Cross-file reasoning — impact analysis, lesson recall, dependency-aware refactor (the kind of thing that requires reading 5 files; we read the graph instead).
• Auditable AI decisions — confidence scores, evidence trails, tamper-evident logs anchored to a public transparency log.
• EU AI Act–aligned behaviour out of the box — for European customers, regulated deployments, and analyst-grade due diligence.
• Model-agnostic operation — 14 LLM backends, offline-capable via Ollama, runs entirely on your machine by default. No telemetry. Code stays on your machine.

How it works

• Scan → AST + dependency analysis builds a typed graph (functions, classes, modules, imports, calls).
• Activate → A pre-reasoning safety layer scores each node for relevance, confidence, and risk before the LLM runs.
• Reason → Multiple agents debate. Outputs carry confidence, graph_health, active_nodes, evidence pointers.
• Gate → Governance gates (CG-01..CG-20) intercept write-class operations. Plans required. Risks surfaced. Trade-secret + path-traversal hardening enforced.
• Audit → Every tool call is logged to .graqle/governance/audit/ with redaction + secret scanning.
• Commit → For runtime decisions, the audit record gets canonicalised (RFC 8785), Merkle-rooted (RFC 6962), ed25519-signed, and anchored to the public Sigstore Rekor log.
• Learn → Lessons become weighted edges. The graph remembers across sessions, teams, and git operations.

The pipeline runs through five named phases — ANCHOR → ACTIVATE → GENERATE → VALIDATE → COMMIT. Each phase is governance-gated, evidence-attached, and audit-logged.

API defaults: confidence_threshold=0.65 (refusal floor), gate_threshold=0.60 (gate-status floor). Both are configurable per-call.

Model agnostic

Anthropic · OpenAI · AWS Bedrock · Ollama · Gemini · Groq · DeepSeek · Together · Mistral · OpenRouter · Fireworks · Cohere · Azure OpenAI · custom HTTP.

# graqle.yaml — smart task routing
backends:
  reasoning:  anthropic/claude-sonnet-4-6   # quality work
  embedding:  bedrock/titan-v2              # cheap + fast
  summaries:  ollama/llama3                 # local + free

Runs fully offline with Ollama. No telemetry. Code stays on your machine. API keys stay in your local graqle.yaml.

Governance gate — drop-in for Claude Code, Cursor, VS Code

graq init              # sets up a governed project (writes the constitution → CLAUDE.md)
graq gate-install      # one-time, project-local — enforce it for Claude Code

graq init writes the GraQle constitution into your project, so your AI tool behaves like a disciplined senior engineer from the very first command: governed tools only (every change is checked), a defined investigate → plan → review → apply → learn workflow, built-in token-cost rules, and the project's known pitfalls baked in. One rulebook — shipped as graqle/data/constitution/ — renders for every client (Claude Code → CLAUDE.md, OpenAI Codex → AGENTS.md, Cursor → .cursorrules, Windsurf → .windsurfrules), so editing it once keeps them all in sync.

gate-install then routes every native write/edit/bash through GraQle's governance gates and adds a permissions backstop to .claude/settings.json. Plans required for risky changes. Trade-secret scanning on git commits. Path-traversal hardening on subprocess capture. CG-01 through CG-20 — all on, all auditable.

→ Governance Gate spec

MCP-first

// .mcp/config.json
{ "graqle": { "command": "graq", "args": ["mcp", "serve"] } }

76+ MCP tools — every operation Claude Code / Cursor / VS Code Copilot needs is exposed as a governed tool with confidence scores, evidence pointers, and audit-trail entries. No prompt engineering, no glue code.

🇪🇺 EU AI Act–aligned

Articles 6, 9, 12, 13, 14, 15, 25, 50 become applicable on 2026-08-02. GraQle gives your high-risk AI system the signals, audit trail, and disclosure primitives it needs — so the parts of your compliance file you can quote from us, you can quote today.

# One switch flips every EU-AI-Act-aware subsystem at once
graq compliance switch on        # shell snippet → eval to enable
graq compliance switch status    # what's actually armed, in one envelope
graq compliance switch off       # symmetric disable

# Per-subsystem CLI surface
graq compliance status                                      # legacy + new subsystems block
graq compliance export --since 2026-08-01 --sha256-sidecar  # Article 12 evidence
graq compliance baseline-doc generate --output baseline.jsonl  # Q16.1 baseline
graq compliance periodic-assessment run --period-start ... --period-end ...  # Q16.3
graq compliance feedback record --rating 5 --note "..."     # Q16.5 observation
graq compliance eur-lex-check                               # weekly drift guard

Article	What GraQle provides	Where
Art 4 — AI literacy	Integration guidance for providers + deployers	Art 4 doc
Art 9 — Risk management	Periodic-assessment artefacts with auto-remediation triggers	graq compliance periodic-assessment run
Art 11 — Technical documentation	Dated, content-addressed baseline document at deployment	graq compliance baseline-doc generate
Art 12 — Record-keeping	JSONL audit export + SHA-256 tamper-detection sidecar	graq compliance export
Art 13 — Deployer transparency	graph_health + confidence on every reasoning envelope	every graq_reason call
Art 14 — Human oversight	Confidence-gated refusal of auto-apply + claim-limits vocabulary	GRAQLE_EU_AI_ACT_MODE=on + graq edit/apply/auto
Art 15 — Accuracy / robustness / cybersecurity	17 named defences + 7 measurable claims	graq compliance status --include-robustness
Art 25 — Value-chain responsibility	Intended-purpose declarations + PCT (Proof-Claims Token) x-ai-eu extension (11 fields)	Art 25 doc + graq pct issue/validate
Art 43 — Conformity assessment	Substrate evidence inputs (baseline-doc + audit log + periodic assessment + robustness + Article 14 gate) for the deployer's Annex VI internal-control file	Art 43 doc
Art 50 — Transparency for users	Auto banner + ai_disclosure machine field	GRAQLE_EU_AI_ACT_MODE=on
Art 72 — Post-market monitoring	graqle govern serve continuous anchoring + graqle govern health snapshot	v0.62.0

Three substantive non-claims kept legally clean:

• GraQle is NOT itself a high-risk AI system (no Annex III category applies).
• GraQle is NOT a GPAI provider under Article 51 (we use third-party LLMs, we don't place one on the EU market).
• We provide signals, audit primitives, and conformity-assessment evidence inputs. We never say compliant or certified. The discipline is enforced in code — TestNonClaimsInvariants blocks any release that introduces a compliant/certified field.

→ Full Article-by-Article mapping in docs/compliance/eu-ai-act/

Contributions welcome on the compliance docs

The EU AI Act docs are deliberately open to contribution — corrections, translations (DE/FR/ES/IT have highest demand), compliance gap reports from deployers building Annex VI internal-control files, and cross-framework mappings (NIST AI RMF, ISO 42001, ENISA, etc.) are all welcome. See CONTRIBUTING-COMPLIANCE.md for the contribution guide, the vocabulary discipline the CI enforces, and what kinds of changes go through which review path.

Security & integrity

No telemetry	GraQle does not phone home, collect usage data, or send analytics.
No code upload	Source never leaves your machine unless you opt in to cloud sync.
Secret scanning	200+ regex patterns + Shannon-entropy detection + AST scan on every output candidate.
PyPI Trusted Publishing	OIDC-only — no long-lived API tokens in our pipeline.
Sigstore signatures	Every wheel signed by our GitHub Actions identity. Verify with graq trustctl verify --version <v>.
CycloneDX SBOM	Attached to every GitHub Release.
.pth-file guard	Publish pipeline rejects any wheel containing .pth files (the LiteLLM-class attack vector).
Reproducible builds	SOURCE_DATE_EPOCH-pinned, rebuild from tagged source and compare checksums.
Survive-disappearance	Production audit records anchor to public Sigstore Rekor — verifiable even if Quantamix disappears.

→ Full disclosure policy: SECURITY.md · Report vulnerabilities to [e-mail]

What's new in v0.75.0

The EU AI Act layer is complete. GraQle now offers an optional, configurable EU AI Act (Reg. (EU) 2024/1689) compliance layer — off by default, enforced through a tamper-evident, irreversible latch.

• Configurable + irreversible latch (governance.eu_ai_act in graqle.yaml): once enabled, the layer cannot be silently disabled and blocking cannot be downgraded to advisory. The latch is an ed25519-signed, hash-chained record (.graqle/eu_ai_act_latch.jsonl) — not a hand-editable flag — so a tamper attempt fails closed and can never turn it off.
• Enforced compliance phase (CG-EU-AIA) wired into the gate: when enabled, AIA-relevant write tools pass an Article-14 human-oversight check. blocking + low confidence → refused with an audited per-action override path; advisory → recorded + advised, never blocked. Reads, planning, and reasoning are never gated.
• Light-touch + honest by design — a record-keeping / traceability aid (supports Art. 12 / 72), not a hard wall, and not a substitute for human compliance judgement. The latch supports the Act's expectations; it is not "required by the Act."

This completes GraQle's universal-governance arc (constitution-as-code → every client → universal server gate → cost-is-observability → EU AI Act layer).

→ Full v0.75.0 changelog

What's new in v0.73.0

Cost is observability, never a quality gate. GraQle never cuts reasoning or debate quality to save money. Every cost path is now advisory: it measures and surfaces spend (the cost-savings story) but never halts still-valuable work.

• Reasoning continues past budget to natural convergence / max_rounds; the cost of continuing is measured (continuation_cost_usd in result metadata).
• Multi-backend debate no longer stops on budget — it runs to max_rounds and reports over-budget rounds instead.
• Advisory per-session cost meter in the MCP server surfaces session_cost_usd and a one-time over-budget note — purely observational, never blocks a tool, and hardened against malformed cost values.
• Runaway protection stays value-based (max_rounds + the absolute LLM-call ceiling), never price-based.

→ Full v0.73.0 changelog

What's new in v0.72.0

One constitution, every AI client. The governance rulebook now renders into every supported client from a single source — including OpenAI Codex via AGENTS.md, which previously had no instruction file. Run graq init and your AI tool pair-programs with a disciplined senior engineer from the first command, whichever tool you use.

• The constitution (graqle/data/constitution/) — governed-tools-only rules, the 9-phase workflow, the full MCP tool inventory, token-cost rules, learned-behaviour workarounds, and a configurable (off-by-default) EU AI Act section. Modular Markdown; edit once, every client stays in sync.
• Per-client rendering: Claude Code → CLAUDE.md, OpenAI Codex → AGENTS.md (new), Cursor → .cursorrules, Windsurf → .windsurfrules. Append-under-marker and idempotent — an existing file is never clobbered.
• graq gate-install adds a non-destructive permissions backstop to .claude/settings.json (deny native write/exec, allow the governed graq_* tools) behind the existing PreToolUse hook.

→ Full v0.72.0 changelog

Recent releases

• v0.62.0 — Runtime R2: graqle govern serve continuous anchoring worker + govern health Article-72 monitoring snapshot.
• v0.61.0 — Runtime R1: FastAPI middleware + @governed decorator. Drop-in governance for any FastAPI app.
• v0.60.0 — Runtime R0 Mode A: GovernedRuntime.attest() and PII-safe pseudonymize_ref().
• v0.59.0 — Layer 5 cryptographic substrate GA: RFC 8785 canonicalisation + RFC 6962 Merkle commitments + ed25519 signatures + Sigstore Rekor anchoring + local replay queue.
• v0.58.0 — EU AI Act Wave 3 substrate (Article 43 conformity-assessment evidence) + OPSF PCT alignment + GRAQLE_WORKTREE_ROOT for parallel-worktree dev.
• v0.57.0 — EU AI Act Wave 2: graq compliance switch single entry-point, Article 14 confidence-gated refusal, claim-limits vocabulary, EUR-Lex drift guard.

→ Full changelog

Pricing

Tier	What you get
Free	Local-only graphs · core SDK · governance gates · EU AI Act surfaces · attest() runtime · govern serve anchoring (self-hosted, anchored to public Rekor)
Pro — $19/mo	Cloud sync · priority models · hosted Rekor relay
Team — $29/dev/mo	Shared KGs · team-wide lessons · audit log retention · SOC 2 evidence pack
Enterprise	On-prem · custom backends · dedicated support · regulated-deployment SLAs · contact us

The free tier is real: the verifier, the runtime attestation path, and the continuous anchoring worker are all in the open-source SDK. Paid tiers add operational scale, team features, and a managed Rekor relay.

Patent & license

Core methods are patent-pending: EP26167849.4 (filed 2026-03-25), EP26162901.8 (CIP), and EP26166054.2 (CogniGraph divisional). The SDK source is fully auditable under the GraQle License — see LICENSE. Reimplementation of the patented methods outside this SDK requires a separate patent license.

→ github.com/quantamixsol/graqle — issues, discussions, contributions welcome.

GraQle is built by Quantamix Solutions. Query your architecture. Prove your AI's decisions.