Skip to main content

Instalação portátil: docs/installation/quickstart.md — clone → setup → install. Delegação agêntica: forge.agentic.json (SpecialistAgenticManifest/v1). Docs: commands · skills · agents · tutorial · economy · troubleshooting

Platform Forge

Platform Forge

CI License: MIT Python

Agentic Platform Engineering intelligence platform — deterministic, offline-first, evidence-first, graph-aware, provider-neutral in the core.

Platform Forge reads repositories, Terraform/OpenTofu, Kubernetes, GitOps, CI/CD, observability data, SLOs, security posture, cost and ownership — and builds one coherent, evidence-backed model of a platform. Facts have provenance; findings cite evidence; unanswered questions come back as named refusals, never hedged prose.

Versão em Português: README.pt-BR.md

Install

pip install platformforge   # from PyPI — offline-capable

From a checkout (development):

pip install -e .            # core — offline-capable
pip install -e ".[dev]"     # + tests/lint
# or: uv sync && uv pip install -e .

Quick start

platformforge doctor                       # environment health
platformforge init --repo .                # scaffold .platformforge/
platformforge inspect --repo .             # inventory analyzable artifacts
platformforge analyze iac ./infra          # HCL → facts
platformforge analyze k8s ./manifests      # manifests → facts
platformforge analyze gha --repo .         # workflows → facts
platformforge analyze kyverno ./policies   # kyverno/gatekeeper → facts (version-aware)
platformforge analyze cosign bundle.json   # sigstore shape — claimed ≠ verified
platformforge analyze slsa prov.json       # SLSA requirement/evidence/gap
platformforge analyze cloud-aws ./dumps    # AWS CLI dumps → T1 facts
platformforge judge facts.json             # rules → findings
platformforge judge facts.json --versions '{"kubernetes":"1.29"}'  # version-gated rules
platformforge graph build facts.json       # facts → provenanced graph
platformforge graph blast --node X         # blast radius (per impact class)
platformforge graph identity-become --node role/x   # who can become this role
platformforge observe slo contract.yaml --sli '{"total_events":1e6,"bad_events":400}'
platformforge observe postmortem --incident i.json # unresolved root cause stays named
platformforge finops costs billing.json    # cost facts + summary
platformforge finops ingest cur.csv        # CUR/Azure/GCP/OpenCost/Kubecost → rows
platformforge finops focus-validate rows.json # FOCUS 1.0 column compliance
platformforge finops unit costs.json --denominators d.json # cost/request etc.
platformforge change review --repo . --patch diff.patch   # sandbox → graph delta → risk
platformforge product maturity --signals s.json
platformforge mcp tools                    # bounded MCP capability surface
platformforge mcp serve                    # stdio server (same core as CLI)
platformforge lab run-all                  # Forge Lab scenario suite
platformforge lab chaos lab/chaos-pod-kill # fault injection on the graph
platformforge evals run                    # eval framework (§94–95)
platformforge evals coverage               # rule ↔ case coverage matrix
platformforge evals precision              # measured false-positive rate
platformforge store gc                     # §151 — dry-run GC of stale blobs
platformforge collect ./dumps              # sniff dumps → facts
platformforge fleet report lab/fleets/acme # fleet north-star receipt
platformforge optimize scan lab/fleets/acme  # opportunities → ChangeIntent
platformforge agents list                  # canonical 41-agent roster
platformforge agents check                 # host-mirror drift gate
platformforge agents playbook "task"       # zero-subagent fallback
platformforge route '{"task":"...","risk":"high"}'  # Router V2 decision
platformforge bench scale                  # measured 10k-node/500k-edge bench
platformforge diagnose <node> --findings j.json --facts f.json
platformforge plan findings.json           # ordered remediation plan
platformforge policy check facts.json      # catalog as policy
platformforge security ./dumps             # secrets+iam+sbom+supply bundle
platformforge sdd status --feature F       # spec lifecycle + hash cascade
platformforge forge manifest               # capability manifest (interop)

Pipeline

artifacts → analyze → facts ──► judge → findings ──► scorecards / gates
               │                                   ▲ rules/catalog/
               └──────► graph build → graphfy ──────┘ (deps, blast, diff)

Facts carry evidence tiers (measured → provider → plan → repo → doc → declared → inferred). Graph edges inherit provenance. Every mutation path goes through inspect → propose → sandbox → verify → approve → apply; the core itself never mutates anything.

Design docs

Index: docs/README.md — ADRs, agent docs, cycle reports.

Architecture & model: ARCHITECTURE.md · GRAPHFY.md · DOMAIN-MAP.md · RUNTIME-TOPOLOGY.md · RECONCILIATION.md · CLOUD.md · OBSERVATION-MODEL.md · COMPATIBILITY.md · DEPRECATION.md

Scale & ops: FLEET.md · MULTI-CLUSTER.md · PLATFORM-ANALYTICS.md · OPTIMIZATION.md · FEDERATION.md · OPS.md · APPROVALS.md · ROLLBACK.md · LIVE.md · COLLECTORS.md · CAPACITY.md · AI-PLATFORM.md · ENTERPRISE.md · GOLDEN-PATHS.md · GOLDEN-PATH-ANALYTICS.md · POLICY-INTELLIGENCE.md · PLATFORM-MEASUREMENT.md · INCIDENTS.md · RUNBOOKS.md · EXECUTION.md

Agentic & economy: AGENTIC-OS.md · docs/agents/ (agentic runtime) · ECONOMY.md · QUALITY-PER-TOKEN.md · KNOWLEDGE.md · SDD.md · VERIFICATION.md

Rules & evidence: RULES.md · EVALS.md · SECURITY.md · SOURCES.md · CAPABILITIES.md · RESEARCH.md · docs/freeze/ (architecture freeze)

Reference: CLI-REFERENCE.md (curated) · docs/CLI-SURFACE.md (generated index) · GLOSSARY.md · MCP.md · ROADMAP.md · CHANGELOG.md · CONTRIBUTING.md · AGENTS.md · CODE_OF_CONDUCT.md · README.pt-BR.md

Support matrix

Axis Supported
Python 3.10+ (CI tests 3.10–3.13)
OS Linux (primary); macOS/Windows run the core — host adapters are platform-dependent
Runtime deps PyYAML, jsonschema, python-hcl2 — offline core only
Optional extras aws (read-only dumps), mcp (host surface), dev (pytest, ruff)
Host agent mirrors .claude/, .agents/, .codex/, .devin/ — generated, agents check gated
Architecture state Frozen — see docs/freeze/; new capability requires an unfreeze RFC

License

MIT — see LICENSE, SOURCES.md, CREDITS.md.

Graph Studio

platformforge graph view|ui projects the built graph through forge/ForgeGraphView/v1 and serves the embedded local explorer (declinable at install via --components). Full docs: the-forge/docs/graph-studio/.

Metadata

Release files for platformforge 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for platformforge 0.1.2
File Size Uploaded Provenance
platformforge-0.1.2.tar.gz 3.5 MB With provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for platformforge 0.1.2
File Interpreter ABI Platform
platformforge-0.1.2-py3-none-any.whl Python 3 none any Details

Total release size: 4.8 MB

Release files / platformforge-0.1.2.tar.gz

Download URL platformforge-0.1.2.tar.gz
Size 3.5 MB
Tags Source
SHA-256 checksum
How to use checksums
fb1e7803f07a2b535e39bcef1c6d6cb9ac0686ae509117118b0334772bbdaecc
BLAKE2b-256 checksum
How to use checksums
337e0faba27c606b4fd744ebe1817250ba9bd71c857c2b0391ab5c2e72099176
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

Release files / platformforge-0.1.2-py3-none-any.whl

Download URL platformforge-0.1.2-py3-none-any.whl
Size 1.3 MB
Tags Python 3
SHA-256 checksum
How to use checksums
4bb30e0bec810ee2c2a82b8dfe74cf0372c9055ca07d8950579f702ae5538417
BLAKE2b-256 checksum
How to use checksums
45e3b77f634a025d3d84da08f197715d017a6c19ba077c3d43da39a22a005046
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

Security for platformforge 0.1.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 2 files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for platformforge-0.1.2.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page