Phantom Secrets

---

▶ Historical v0.4 demo — current behavior differs  ·  🛡 Security model  ·  📋 Threat model  ·  💬 Discussions

[!IMPORTANT] Release boundary: v0.7.3 is the reviewed public distribution. This repository stages 0.7.4; source, changelog entries, workflow definitions, and local test results do not prove that a 0.7.4 artifact or package was published, deployed, provider-enabled, or accepted on any platform. See release readiness and platform support.

Why Phantom?

AI coding agents routinely work in repositories that also contain local credentials. Once a real API key enters an agent context, transcript, tool call, or generated file, you have lost control of where that value may persist.

Traditional secrets managers focus on keys at rest and in transit. Phantom adds a boundary for agent context:

• 🔒 Reduces one credential-exposure path — managed project dotenv files contain sensitive phm_ mappings, MCP responses remain value-blind, and exact proxy routes inject their own configured authentication values. Unmanaged files, broader shell authority, and same-user processes remain in the threat model.
• ⚡ Fast local setup — after installing the reviewed v0.7.3 release, phantom init protects a project without requiring an account, DNS changes, or a custom CA.
• 🧰 Agent-native integrations — setup helpers and value-blind MCP workflows for Claude Code, Cursor, Windsurf, and Codex, plus project instructions for GitHub Copilot.
• 🦀 Open source, local-first, MIT — secrets use the native OS credential store when it is available, with an explicit encrypted-file fallback. Optional cloud sync encrypts vault payloads client-side before the server stores them.

Project status and trust boundary

Phantom's implemented user-facing surfaces are the CLI, vault, authenticated local proxy, MCP server, and optional cloud/team workflows documented below. Cloud and team behavior additionally depends on the deployed service, account plan, and provider configuration; source code alone is not deployment or customer-acceptance evidence. The conversation facade is intentionally narrow:

• phantom_do is proposal-only. It canonicalizes a closed Cargo action and reports its digest, effect, and activation blockers; execute is hard denied.
• phantom_setup_workspace can propose setup, create a bearerless request, and report authenticated status. Applying a request remains a separate trusted-terminal operation.
• Advanced MCP tools remain a compatibility catalog, disabled by default, with separate explicit confirmation and informed terminal-approval gates. They are not governed by the conversation facade's capability card.
• phantom grant retains value-blind lifecycle metadata and design-source foundations, but 0.7.4 hard-denies every live provider issuance/renewal path before credential or network access. A provider grant is not an execution-kernel authority grant, broker lease, or permission for an agent to execute work.
• The authority, broker, runtime, session, and evidence crates are inactive, fail-closed foundations. They do not establish live Locus authority, broker credentials, execute agent actions, or produce externally trusted receipts today.

See the documentation map, architecture, security policy, and threat model for the evidence behind those boundaries.

Quick Start

Install the reviewed v0.7.3 release first. On macOS, the verified Homebrew path is:

$ brew tap ashlrai/phantom
$ brew trust --formula ashlrai/phantom/phantom
$ brew install ashlrai/phantom/phantom

Linux and Windows users should use the exact v0.7.3 assets in Installation. Then protect and verify the project:

$ phantom init
# Auto-detects .env, .env.local, or .env in subdirectories
# Stores real secrets in the native credential store or encrypted vault,
# then rewrites .env with phantom tokens
# Auto-configures Claude Code MCP server if detected

$ phantom agent doctor
# One human-readable readiness check for AI-agent safety

$ phantom exec -- claude
# Authenticated proxy running on an ephemeral 127.0.0.1 port
# App/test processes use phantom tokens; agents use value-blind metadata

For a task contract you can hand to Claude Code, Codex, Cursor, Windsurf, or Copilot, use the safe delegation quickstart and the copyable policy and task templates. Teams evaluating a controlled rollout can start with the enterprise adoption guide.

Windows

The same core commands work on native Windows. Install the exact v0.7.3 Windows ZIP for your architecture from Installation, verify its published .sha256 sidecar, and place both executables on PATH. WSL is a separate Linux environment with its own filesystem and credential-store context.

For an explicitly supervised foreground proxy, run phantom start with stdin, stdout, and stderr each attached to a trusted terminal. Terminal attachment is an admission check, not proof of who controls a PTY. The CLI detects your shell and prints the matching env-var syntax; copy those exports into the terminal that launches the client, keep the owning terminal open, and press Ctrl-C there to stop. Detached --daemon mode and external process control are fail-closed until Phantom has a separately reviewed private cross-platform control channel. phantom stop only authenticates a legacy v0.7.3 .phantom.pid session to report migration guidance; v0.7.3 had no authenticated shutdown endpoint, so the new binary never kills that process or deletes its record. Stop it from its owning v0.7.3 terminal with Ctrl-C, or use the bounded recovery procedure in Troubleshooting. For reference:

PowerShell:

$env:OPENAI_BASE_URL = "http://127.0.0.1:PORT/openai/_phantom/TOKEN/"
$env:PHANTOM_PROXY_PORT = "PORT"
$env:PHANTOM_PROXY_TOKEN = "TOKEN"

cmd.exe:

set OPENAI_BASE_URL=http://127.0.0.1:PORT/openai/_phantom/TOKEN/
set PHANTOM_PROXY_PORT=PORT
set PHANTOM_PROXY_TOKEN=TOKEN

Git Bash / WSL: use the export X=Y syntax from the main quick-start.

Notes:

• PHANTOM_PROXY_TOKEN is the proxy session authenticator. By default, phantom exec and phantom start include it in local *_BASE_URL values as /_phantom/TOKEN/ so unmodified SDKs work. Header-aware clients can set PHANTOM_PROXY_HEADER_AUTH_ONLY=1 and send x-phantom-proxy-token: $PHANTOM_PROXY_TOKEN instead.
• If phantom.exe is blocked by Windows application-control policy, do not automatically remove Mark-of-the-Web. First verify the archive checksum and both binary identities against the release metadata. If local policy permits the verified binaries, a user may then remove the mark explicitly with PowerShell: Get-ChildItem "$env:USERPROFILE\.phantom-secrets\bin\*.exe" | Unblock-File.
• The pre-commit hook installed by phantom init is a #!/bin/sh script. Native git from the command line invokes it via Git for Windows' bundled sh.exe, which is what the official Git for Windows installer ships. GUI clients (GitHub Desktop, some IDE integrations) may run with a stripped-down PATH that lacks sh.exe and silently skip the hook — for these, run commits from a terminal, or use phantom check --staged directly. CI is the durable safety net regardless.
• The release workflow defines x64 and ARM64 Windows ZIPs, and the npm and PowerShell installers map both targets. A workflow definition is not evidence that an exact archive was published, signed, or passed native acceptance. See the platform support matrix.

How It Works

  .env file (AI read denied)       OS Keychain / Vault
  +--------------------------+      +---------------------+
  | OPENAI_API_KEY=phm_a7f3  | ---> | sk-real-secret-key  |
  | STRIPE_KEY=phm_c9d1...   |      | sk_live_real-key... |
  +--------------------------+      +---------------------+
           |                                 |
           v                                 v
  App / test process                Phantom Proxy (127.0.0.1)
  +--------------------------+      +------------------------------+
  | Loads phm_ tokens        |      | Intercepts HTTP requests     |
  | Agent gets MCP metadata  | ---> | Injects route-owned auth key |
  | Makes API calls to proxy |      | Forwards over TLS to real API|
  +--------------------------+      +------------------------------+

1. phantom init reads .env, stores real secrets in the native OS credential store or encrypted-file fallback, and rewrites .env with phm_ tokens
2. phantom exec -- claude starts a local reverse proxy, sets SDK-compatible service base URLs such as OPENAI_BASE_URL=http://127.0.0.1:PORT/openai/_phantom/TOKEN/, exposes PHANTOM_PROXY_TOKEN to the child process, and launches the command
3. API calls hit the proxy, which authenticates the local session, removes its local auth token, injects only the matched route's vault value into that route's exact configured authentication header, and forwards over TLS. Client-controlled headers and bodies never resolve phm_ tokens.
4. When the session ends, the proxy shuts down and its bearer is invalid. phantom exec also replaces project tokens with fresh child-process tokens for that run. Project phm_ values persist until rotation; they are not provider credentials and are never client-resolved, but a process controlling an authenticated active proxy can still invoke its configured provider routes.

Phantom does not grant AI tools permission to read .env or other dotenv files. phantom setup removes legacy Phantom-managed dotenv read grants and preserves deny rules; agents use value-blind MCP inventory instead.

Provider grants

phantom grant retains value-blind lifecycle metadata and provider protocol design foundations. In 0.7.4, all live provider issuance, enrollment exchange, refresh, renewal, and revocation execution is hard-denied before Phantom reads a provider credential or opens a provider network connection. Do not use grant or rotation commands expecting a vendor-side change. Test-only mocks demonstrate local transaction behavior only; they are not live-provider acceptance.

In these docs, provider grant means design-era credential lifecycle metadata; no live grant flow runs in 0.7.4. Authority grant means the inactive, value-free execution authority type in phantom-authority. A provider grant cannot be reinterpreted as an authority grant, Locus credential, broker lease, or execution permit. See the design-era grant lifecycle specification; the issuance contract is the original design contract and retains design-era status language.

MCP Integration (Claude Code, Cursor, Windsurf, Codex)

Phantom ships an MCP server so AI coding tools can inspect value-blind metadata and request gated lifecycle operations. MCP responses do not return real secret values.

• Conversation facade — phantom_capability reports authority and hard denials for the small facade (not the separately gated advanced compatibility catalog); phantom_do canonicalizes one closed Cargo action and reports the exact activation blockers without executing it; phantom_setup_workspace proposes an exact value-blind plan, creates a bearerless apply request after revalidation, or reads authenticated request status. Provisioning the machine-local seal key and creating a request both require confirm plus an out-of-band approval token; MCP never claims or applies the request.
• Vault — phantom_list_secrets, phantom_status, phantom_init, phantom_add_secret_interactive, phantom_add_secret (deprecated; refuses plaintext), phantom_remove_secret, phantom_rotate, phantom_copy_secret
• Detection + diagnostics — phantom_doctor, phantom_why, phantom_check, phantom_env, phantom_validate_secret, phantom_validate_all
• Local-to-cloud — phantom_wrap, phantom_unwrap, phantom_sync, phantom_cloud_push, phantom_cloud_pull, phantom_cloud_status
• Teams — phantom_team_list, phantom_team_create, phantom_team_members, phantom_team_invite, phantom_team_key_publish, phantom_team_vault_push, phantom_team_vault_pull
• Advanced audit, rotation, expiry, and compliance — phantom_audit_recent, phantom_audit_stats, phantom_audit_analytics, phantom_audit_anomalies, phantom_audit_anomalies_realtime, phantom_audit_hotspot_alerts, phantom_audit_incidents, phantom_leak_incidents_realtime, phantom_audit_alerts, phantom_audit_export_report, phantom_compliance_status, phantom_rotate_with_candidate, phantom_rotate_promote, phantom_rotate_provider, phantom_rotate_with_expiry, phantom_list_with_expiry, phantom_secret_rotation_due, phantom_validation_schedule, phantom_validation_history, phantom_secrets_expiry_check, phantom_secrets_auto_rotate, phantom_expiry_enforce, phantom_rotation_schedule_next, phantom_apply_expiry_policy

phantom_secrets_auto_rotate and phantom_rotate_with_expiry are deprecated compatibility names for approved local phm_ token remaps. They do not rotate a provider credential, renew TTL/rotated_at metadata, clear leak incidents, or sync credentials. phantom_rotate_provider is also hard-denied before credential or network access in 0.7.4. Team invites may assign only member or admin; ownership transfer is not an invite role.

Tools that write state, retrieve or use credentials, or make provider/network requests are disabled by default. They can reach their confirm: true and one-use approval_token gates only when the operator sets PHANTOM_MCP_EFFECTS=trusted-terminal in a server environment outside agent authority. phantom mcp-approve requires attached stdin/stderr, shows the bounded value-blind effect plus exact parameters, and requires a fresh typed challenge before approval. A same-user shell or agent-controlled PTY can defeat that ceremony; leave effects disabled unless the approval command and ~/.phantom approval storage are outside the requesting agent's authority. Conditional tools keep their inspection mode ungated and activate both gates only for the effectful parameters. Real secret values are never accepted as MCP tool arguments; new secrets are entered out-of-band in a trusted terminal. The exact 54-tool names, descriptions, and JSON schemas are mirrored from runtime tools/list into mcp-registry/server.json, and release smoke tests reject any drift.

phantom_rotate_with_candidate and phantom_rotate_promote remain in the 54-tool compatibility catalog as deprecated hard denials. They never create, validate, or promote a candidate and never change vault or shadow metadata; phantom_rotate_provider is also hard-denied before credential or network access in 0.7.4. phantom add creates new names only and refuses existing-name replacement before reading a value. Replacing a credential therefore requires a separately reviewed provider rotation plus an explicit trusted-terminal remove-and-add sequence; those are distinct, non-atomic operations.

Workspace setup is deliberately split across trust boundaries. MCP can call phantom_setup_workspace with phase=propose, then phase=request_apply using the exact returned plan_id and pre_state_id. The first proposal requires dual approval only when it must provision the machine-local plan-seal key; request_apply always requires dual approval because it persists a value-free request outside the repository. Apply it from an attached trusted terminal with phantom workspace apply --request <ID>; MCP has no claim or apply operation.

One command per AI client — Phantom writes the right config file in the right place:

phantom setup --client claude     # .claude/settings.local.json (project)
phantom setup --client cursor     # ~/.cursor/mcp.json
phantom setup --client windsurf   # ~/.codeium/windsurf/mcp_config.json
phantom setup --client codex      # ~/.codex/config.toml
phantom setup --client claude --print   # snippet to stdout for any other client

Install both v0.7.3 release binaries before setup. Released v0.7.3 normally records the running phantom executable with mcp serve; if that executable cannot be resolved, it looks for the local standalone phantom-mcp. Its final legacy fallback is unpinned npx -y phantom-secrets-mcp, which currently resolves an older registry track. Do not rely on that path: keep both verified release binaries installed and inspect the generated entry. Current main removes the registry fallback and fails closed when neither local runtime is executable; that hardening is not part of v0.7.3 and awaits a later release. Restart the AI tool after setup so it picks up the new config.

Phantom's stdio MCP server can be configured in MCP clients that support local command servers. The setup writer currently has reviewed presets for Claude Code, Cursor, Windsurf, and Codex; other clients require their own compatible configuration.

Cloud backup + Dashboard

Back up and restore a vault on the same keychain machine with client-side encryption. Phantom Cloud receives the encrypted vault payload rather than the decrypted secret values; endpoint, client, account, and OS-keychain security remain part of the trust boundary. The cloud encryption key is generated and stored in the local OS keychain. Phantom does not currently ship key transfer or recovery, so account sign-in without that key cannot decrypt this backup. Login, logout, cloud push/pull, and browser open are not headless agent actions: run them with stdin, stdout, and stderr attached to a terminal outside the requesting agent's authority and complete each fresh exact typed challenge.

$ phantom login
# Opens GitHub OAuth (device code flow)

$ phantom cloud push
# Encrypted client-side, uploaded to phm.dev

$ phantom cloud pull   # restore on the machine that holds the original cloud key
# Downloaded and decrypted locally

$ phantom open
# Opens https://phm.dev/dashboard — read-only view of your projects,
# vault sizes, last sync, plan tier, and team membership.

Cloud backup uses ChaCha20-Poly1305 with a random client-side passphrase stored in the OS keychain and derived via Argon2id for encryption. The server stores only ciphertext. Team vaults are separate: they wrap a shared vault key to each registered member's X25519 public key; that does not make personal cloud backups portable.

Team vaults (source-backed pilot)

The repository includes fixed-membership team-vault workflows intended for a planned hosted Pro pilot. Pilot eligibility, team and vault limits, and pricing are TBD; no self-serve checkout is commissioned. The service path stores ciphertext plus per-member encrypted key shares. Team roles gate invitation management, but all members can read and write the shared vault; member removal and atomic offboarding rotation are not shipped. Treat this as a pilot capability until the hosted service and account entitlement are commissioned. CLI team mutations require the same attached trusted-terminal ceremony before credential, keypair, vault-value, or network access. MCP team reads and writes are provider requests and remain disabled by default behind confirm plus a one-use out-of-band approval token.

$ phantom team create "engineering"
# Creates a team; you become the owner.

$ phantom team invite <team_id> <github-username>
# Invites by GitHub login.

$ phantom team key-publish <team_id>
# Registers your X25519 public key on the team.
# (Run once per team; the private key stays in the OS keychain.)

$ phantom team vault-push <team_id>
# Encrypts the current project's vault with a fresh symmetric key,
# wraps that key (X25519 + ChaCha20-Poly1305) for every member that
# has a registered public key, then uploads.

$ phantom team vault-pull <team_id>   # on a teammate's machine
# Pulls, decrypts the per-member share with their private key,
# decrypts the vault, writes secrets locally.

Team memberships and member lists are visible in the read-only dashboard at phm.dev/dashboard/team.

Command Reference

Command	Description
phantom init	Import .env secrets into vault and rewrite with phantom tokens. Use --empty to initialize a new project before its first phantom add; add does not auto-create project state. --all <DIR> processes eligible repositories found by the bounded five-level scan; discovery stops below the first matching repository. Use --dry-run to inspect the exact set and --jobs N / -j N to control parallelism.
phantom exec -- <cmd>	Start an authenticated proxy and run a command with secret injection
phantom start	Run an explicitly supervised foreground proxy; keep its trusted terminal open and press Ctrl-C there to stop
phantom start --daemon / phantom stop	Detached start fails closed; stop is a TTY-only legacy-state diagnostic that never kills or deletes, not current process control
phantom list	Show secret names stored in vault (never values; --json for machine-readable output)
phantom add <KEY>	Create a new secret name transactionally in an initialized project through a hidden terminal prompt or trusted --stdin producer; existing names are denied before any value read and are never replaced
phantom remove <KEY>	After exact trusted-terminal confirmation, transactionally remove the vault value, lifecycle config, and exact managed-dotenv mapping; headless use fails before value access or mutation
phantom reveal <KEY>	From an attached trusted terminal, review and type the exact challenge before printing one value or copying it for an auto-cleared 30-second clipboard window
phantom status	Show vault/mapping state and whether the machine-local lifecycle lock is held; a held lock does not authenticate or identify a listener
phantom rotate	After an exact attached-terminal challenge bound to the project, config, managed dotenv, and sorted protected-name digest, regenerate every local phantom token (old mappings become invalid). Headless use fails before vault access or mutation. Provider-backed --name/--provider execution is hard-denied before credential or network access in 0.7.4.
phantom grant add <provider>	Reserved provider-enrollment surface; hard-denied before credential or network access in 0.7.4. See Provider grants.
phantom grant list / status	Read provider-grant names, providers, lifecycle state, and expiry metadata without returning credential values.
phantom grant revoke <provider>	Reserved remote-revocation surface; currently fails closed before local mutation because provider revocation is not wired.
phantom doctor	Check configuration and vault health (--fix to auto-repair). Reports install source, vault backend, audit-log status, Argon2 params, and MCP wiring per client
phantom agent report	Emit a read-only AI-agent readiness report (--json for automation). Reports unsafe, protected, verified, team-ready, or compliance-ready
phantom agent doctor	Human-readable agent readiness view backed by the same policy engine
phantom agent setup	Preview or apply safe defaults for agent use (--dry-run first, --apply to write changes)
phantom workspace plan [--json]	Build an exact sealed setup plan and create a value-free pending request; does not change the workspace or vault
phantom workspace apply --request <ID>	Recompute and claim the exact request in an attached trusted terminal, require typed confirmation, then apply transactionally with rollback on failure
phantom workspace status --request <ID> [--json]	Read authenticated, value-free request state
phantom check	Scan for unprotected secrets (pre-commit hook, --staged, --runtime)
phantom sync	Push secrets to Vercel / Railway (--dry-run --json previews safely; --only PATTERN filters by glob, repeatable)
phantom pull	Pull secrets from Vercel / Railway into vault
phantom setup	Wire Phantom into an AI client. --client claude (default), cursor, windsurf, or codex. Add --print to emit the config snippet to stdout
phantom env	Generate .env.example for team onboarding
phantom export	From an attached trusted terminal, review an exact value-blind plan, type its fresh challenge, and enter a dedicated passphrase through the hidden prompt. Export --passphrase-file, plaintext export, argv passphrases, existing targets, symlinks, and paths outside the project fail closed.
phantom import	From an attached trusted terminal, review and type the exact source/target/name/overwrite challenge before restoring a backup or importing Doppler, Infisical, dotenvx, 1Password, or env data. --force selects the reviewed overwrite set but never bypasses consent. A bounded private passphrase file is accepted only on non-Windows platforms and still requires the terminal ceremony.
phantom audit show	Print recent audit events (--last N, --op OP, --name NAME, --json). Requires PHANTOM_AUDIT=1
phantom audit tail	Follow the audit log live (--op, --name filters)
phantom audit path	Print the absolute path to the audit log file
phantom audit verify	Verify HMAC-SHA256 chain integrity; exits 1 if tampering detected
phantom login	From an attached trusted terminal, approve an exact network plan, then separately approve browser opening/polling and keychain persistence
phantom logout	From an attached trusted terminal, approve deletion of persistent cloud authorization from the OS keychain
phantom cloud push	After exact trusted-terminal confirmation, push a client-encrypted vault. A remote success followed by local reconciliation failure is partial success and must not be retried automatically.
phantom cloud pull	After exact trusted-terminal confirmation, pull and decrypt a vault. With force=false, skipped existing entries preserve the prior merge base and block later push until a fully reconciled approved pull.
phantom cloud status	Authenticated provider read; requires attached trusted terminals and an exact challenge before stored-bearer or network access
phantom wrap	Wrap package.json scripts with phantom exec automatically
phantom unwrap	Restore original package.json scripts
phantom watch	Watch managed dotenv files and report new unprotected secrets. --auto hard-denies before mutation in 0.7.4; use transactional phantom init.
phantom why <KEY>	Explain why a key is or is not protected
phantom copy <KEY>	Copy a secret to an initialized target after exact trusted-terminal confirmation; refuses existing target vault, config, or managed-dotenv ownership rather than overwriting
phantom team list/members	Authenticated, value-blind provider reads. CLI requires attached trusted terminals and an exact challenge before bearer/network access; MCP requires confirm plus out-of-band approval.
phantom team create/invite/key-publish/vault-push/vault-pull/rotate-vault	Run CLI team effects only from an attached trusted terminal and complete the exact challenge before credential, keypair, vault-value, or network access. Invites assign only member or admin.
phantom validate / --watch	Send selected credentials to configured validators only after exact trusted-terminal consent; watch authorization ends when config or the vault name set changes
phantom validate schedule	Read status without mutation; setting or disabling persistent scheduler policy requires attached terminals and an exact typed challenge
phantom expiry set	Persist lifecycle policy only after attached-terminal confirmation; expiry enforce is read-only, while expiry rotate is only a deprecated local token remap
phantom open [page]	From an attached trusted terminal, open only the reviewed aliases dashboard, billing, team, docs, pricing, github, issues, or site; arbitrary URLs, paths, and unknown aliases are rejected
phantom upgrade	--check-only performs a read-only version lookup only for an eligible standalone install. Live standalone replacement requires attached terminals and two exact challenges; managed installs route to their owner, ambiguous installs fail closed, and --force is rejected.
phantom completion <shell>	Print a shell-completion script (bash, zsh, fish, powershell, elvish)

Rotating real provider credentials

Phantom 0.7.4 does not execute live provider issuance or rotation. Every provider path—single-provider CLI, batch CLI, MCP, grant enrollment, additive issuance, and destructive/rolling refresh—is hard-denied before provider credential access and before network I/O. Operators must rotate at the vendor, then use Phantom's trusted local secret-entry path. Source adapters and exact cfg(test) mocks are transaction-design evidence, not provider activation, commissioning, or acceptance.

Provider support matrix

Provider	Support	Notes
vercel	Hard denied	Additive issuance can orphan a live successor after a local failure; no provider call in 0.7.4
google	Hard denied	Secret Manager version creation remains disabled pending durable successor recovery and verified abort
github	Hard denied	GitHub App installation-token issuance remains disabled pending the same recovery contract
stripe	Hard denied	Raw keys are manual; rolling OAuth refresh can invalidate the predecessor before successor recovery
aws	Hard denied	SigV4 access-key-pair rotation is not implemented
sentry	Hard denied	Installation-token issuance remains disabled pending durable successor recovery
supabase	Hard denied	PAT rotation is manual; OAuth/additive issuance paths are not commissioned

phantom rotate --batch remains metadata-only discovery/manual guidance and fails closed before vendor execution. No single-provider exception exists.

Features

• Encrypted vault -- macOS Keychain, Linux Secret Service, or Windows Credential Manager, with a ChaCha20-Poly1305 encrypted-file fallback for CI and headless environments. Phantom does not claim Secure Enclave hardware binding. Argon2id uses m=64 MiB, t=3, p=1.
• Phantom tokens -- 256-bit CSPRNG phm_ placeholders in .env, rotatable on demand
• Authenticated proxy sessions -- each proxy run generates a fresh PHANTOM_PROXY_TOKEN; CLI-generated SDK URLs include it for compatibility, and header-aware clients can opt into x-phantom-proxy-token with PHANTOM_PROXY_HEADER_AUTH_ONLY=1
• Route-owned credential injection -- Client headers and bodies never resolve phm_ tokens. After authenticating and matching an exact built-in route, the proxy injects that route's configured vault secret only into its fixed authentication header; a missing mapping fails before any upstream call. Request bodies are still collected under a hard byte cap before forwarding.
• Full SSE/streaming support -- Response streaming preserved end-to-end for OpenAI, Anthropic, and other streaming APIs
• Smart detection -- Heuristic engine distinguishes secrets (*_KEY, *_TOKEN, sk-*, ghp_*) from config (NODE_ENV, PORT)
• Platform sync -- Push/pull secrets to Vercel and Railway
• Pre-commit hook -- Runs phantom check --staged when Git invokes the hook; it checks staged dotenv content plus a bounded set of hardcoded-key prefixes. Hooks can be bypassed or skipped, so CI and a broader secret scanner remain necessary.
• MCP server -- value-blind vault, diagnostics, cloud, team, audit, rotation, validation, expiry, and compliance responses for Claude Code, Cursor, Windsurf, and Codex; effectful compatibility tools remain separately gated and disabled by default
• Cloud backup -- client-encrypted same-keychain-machine backup and restore; key transfer and recovery are not shipped, and deployed-service and account configuration remain separate operational gates
• Export/import -- Encrypted export requires an attached terminal, exact challenge, and hidden passphrase; export passphrase files, plaintext, and argv passphrases are disabled. Import uses its own exact terminal ceremony; only non-Windows encrypted-backup import may read a bounded private passphrase file. Competitor imports support Doppler, Infisical, dotenvx, 1Password, and plain .env.
• Tamper-evident audit log -- PHANTOM_AUDIT=1 writes vault events as JSONL to ~/.phantom/audit.log. Each entry is chained with HMAC-SHA256; phantom audit verify detects tampering. phantom audit show/tail/path for log access.
• Response scrubbing -- Scrubs configured secret values from supported API response paths before returning data to the caller
• Script wrapping -- phantom wrap wraps selected runtime/build scripts (dev, start, serve, build, deploy, preview) and deliberately leaves test, lint, type, and format scripts alone
• Watch mode -- phantom watch reports new unprotected secrets; legacy --auto hard-denies before mutation, so protection stays a reviewed transactional phantom init
• Multi-project scanner -- phantom init --all <DIR> processes eligible repositories found within a five-level bounded scan and stops below the first matching repository; use --dry-run to verify the exact set and --jobs N to control parallelism
• Multi-IDE setup -- phantom setup --client claude|cursor|windsurf|codex writes the right MCP config for each AI tool, or --print for a generic snippet
• Agent readiness -- phantom agent doctor and phantom agent report --json report bounded protection findings and activation gaps; they do not certify a repository or agent as safe
• Enriched diagnostics -- phantom doctor reports install source, vault backend, audit-log status, Argon2 params, and MCP wiring per client
• Secret explainer -- phantom why <KEY> explains detection heuristics
• Cross-project copy -- phantom copy shares secrets between project vaults
• Team vaults -- Fixed-membership encrypted sharing; owner/admin roles gate invitations, while current vault access is member-wide and offboarding rotation is not shipped
• Fail-closed service routing -- agentic proxy sessions accept Phantom's exact built-in OpenAI, Anthropic, Stripe, Supabase, and other reviewed routes; repository-defined destinations are rejected pending trusted-terminal approval support
• Threat model -- See THREAT_MODEL.md for assets, actors, mitigations, and known gaps

Installation

Homebrew (macOS, reviewed v0.7.3 release)

Homebrew 6 requires explicit formula trust for third-party taps:

$ brew tap ashlrai/phantom
$ brew trust --formula ashlrai/phantom/phantom
$ brew install ashlrai/phantom/phantom

This formula installs both phantom and phantom-mcp from the immutable v0.7.3 GitHub release.

Exact GitHub assets (Linux and Windows)

Use the asset matching your OS and architecture. Download its adjacent .sha256 file from the release, verify it before extraction, and place both phantom and phantom-mcp (.exe on Windows) on PATH.

Platform	v0.7.3 archive	Published checksum
Linux x86_64	phantom-x86_64-unknown-linux-gnu.tar.gz	sha256
Linux ARM64	phantom-aarch64-unknown-linux-gnu.tar.gz	sha256
Windows x64	phantom-x86_64-pc-windows-msvc.zip	sha256
Windows ARM64	phantom-aarch64-pc-windows-msvc.zip	sha256

On Linux, verify with sha256sum -c <archive>.sha256. On Windows, compare Get-FileHash -Algorithm SHA256 <archive> with the published sidecar.

Build the exact release source

The v0.7.3 tag resolves to source commit cffd0f29ab85a45358f011fdcfd40667d576c420:

$ git clone https://github.com/ashlrai/phantom-secrets.git
$ cd phantom-secrets
$ git checkout cffd0f29ab85a45358f011fdcfd40667d576c420
$ cargo build --release --locked --bin phantom --bin phantom-mcp

Do not treat unpinned package-manager commands as v0.7.3. As verified on 2026-08-31, npm and the MCP Registry publish 0.6.0, while crates.io publishes 0.5.1; those are older distribution tracks, not the reviewed current release.

Connect an MCP client

$ phantom setup --client claude   # or cursor, windsurf, codex

Architecture

The Rust workspace is organized as product crates plus fail-closed execution-kernel foundations. Presence in the workspace does not mean a foundation is activated in production.

Layer	Crate	Role and current status
Product	phantom-core	Config, dotenv parsing/rewriting, tokens, auth, cloud client, audit, validation, and shared policy.
Product	phantom-vault	VaultBackend trait, OS keychain and encrypted-file backends, and shared cryptography.
Product	phantom-proxy	Authenticated loopback reverse proxy with fixed route-owned auth-header injection, inert client tokens, response scrubbing, and streaming support.
Product	phantom-cli	Operator CLI for initialization, proxy lifecycle, readiness, audit, import/export, sync, team, and workspace workflows.
Product	phantom-mcp	Stdio MCP server. The governed conversation facade is narrow; the advanced compatibility catalog uses separate legacy gates.
Product	phantom-core/src/issuance, CLI grant	Provider protocol/design foundations and value-free grant metadata. All live provider issuance/enrollment execution is hard-denied before credential or network access in 0.7.4.
Local effect layer	phantom-core::fs, phantom-vault transactions	Retained project/config directory capabilities, exact identity/content/permission before-images, no-follow targets, single-link checks, and typed durable, committed-verified-with-durability-warning, or CommittedButUncertain (Partial) effects.
Setup kernel	phantom-workspace	Value-blind discovery, sealed planning, and recoverable trusted-terminal setup transactions. Non-Unix durable mutation fails closed.
Inactive foundation	phantom-authority	Closed authority contracts and deny-all production verification boundary. No live Locus verifier.
Inactive foundation	phantom-locus-contract	Value-free compatibility contract describing requirements for a future Phantom/Locus integration.
Inactive foundation	phantom-broker	Bounded broker protocol and durable replay/accounting primitives. No active transport, lease issuer, or runtime connection.
Inactive foundation	phantom-runtime	Closed engineering action schemas with a deny-all production executor.
Inactive foundation	phantom-session	Crash-explicit session journal. Not wired into active execution.
Inactive foundation	phantom-evidence	Value-free evidence and receipt primitives. Not externally anchored or wired into active execution.

Project-local effects resolve from the retained root acquired with the project transaction lock; global editor configuration is separately bounded beneath a retained user-config authority. A renamed ambient path and replacement decoy do not redirect those in-progress governed writes. These controls are not a same-user sandbox: local locks coordinate Phantom writers, and another process with equivalent user authority remains inside the threat model. See the architecture, threat model, and platform evidence matrix.

Initialization retains the reviewed project-root identity and exact dotenv/config leaf snapshots before vault provisioning, then revalidates root and leaf identity, bytes, and permissions under the project lock before mutation. On Windows, new private anchored files/directories establish a protected current-user DACL before content bytes, while replacements preserve the reviewed exact DACL and inheritance state before writing. Those Windows properties are source contracts; protected native Windows CI acceptance remains pending.

CommittedVerifiedButDurabilityUncertain is committed, exactly verified success with a value-free warning/receipt and must not be rolled back or retried. CommittedButUncertain is the distinct Partial case requiring operator reconciliation because verification or durability remains unresolved.

Vault-backed mutations resolve machine-local vault/application authority before the project transaction lock, then compare the acquired root identity and reread the exact config through that lock. This avoids the shared process-environment/project-lock inversion without trusting project bytes during the gap.

Phantom also tracks a pinned Rama-derived engineering standard for explicit network stacks, modular crates, runnable examples, and tiered platform CI. Rama is a benchmark, not a dependency, and the comparison is not a feature or platform-parity claim.

apps/web contains the Next.js site and backend routes for cloud vault sync, GitHub device authentication, and Stripe billing. The repository source and local tests are separate evidence from the currently deployed state at phm.dev.

Older registry tracks: npm currently publishes phantom-secrets and phantom-secrets-mcp at 0.6.0; these are not the reviewed v0.7.3 release.

CI runs locked, all-target workspace builds and tests on macOS, Linux, and Windows runner environments, plus formatting, Clippy, and npm release-mapping checks. Release builds and native end-to-end acceptance are separate evidence layers; see Platform support.

Security

• Managed dotenv replacement -- after successful initialization, Phantom-managed dotenv values are tokens; unmanaged files, backups, logs, and external tools remain outside this claim
• ChaCha20-Poly1305 encryption for file vault and cloud sync, Argon2id key derivation
• Client-encrypted cloud vaults -- the cloud vault API stores ciphertext; decryption happens in the authenticated client. This claim does not cover plaintext sent intentionally to deployment providers during phantom sync.
• 256-bit CSPRNG tokens -- phm_ prefix distinguishes Phantom tokens from supported real-key formats; random collisions are cryptographically negligible, not mathematically impossible
• Proxy binds 127.0.0.1 only -- not bound to a non-loopback interface; same-user local-process and bearer theft remain in the threat model
• Targeted memory zeroization for major vault retrieval, serialization, and decrypted-file buffers; some proxy lookup copies remain ordinary strings, so zeroization is defense in depth rather than a complete memory-erasure guarantee
• Allowlist model -- proxy only injects secrets for explicitly configured service patterns

See SECURITY.md for the responsible disclosure policy and THREAT_MODEL.md for the full threat model (assets, actors, mitigations, known gaps, cryptography summary).

Pricing

Packaging direction	Local open source	Pro (planned hosted pilot)	Enterprise (planned contract)
Local vault	Available without a hosted entitlement	Available	Available
Hosted cloud vaults	No commissioned entitlement represented here	Eligibility and limits TBD	Scope defined by written agreement
MCP server	Yes	Yes	Yes
Personal cloud backup	Same-keychain-machine restore only	Key portability evaluation planned	Scope defined by written agreement
Team features	--	Fixed-membership pilot; eligibility and limits TBD	Planned enterprise controls; scope TBD
Price	No charge for local open-source use	Pricing TBD; no self-serve checkout commissioned	Written agreement

This table is packaging direction, not evidence that a hosted entitlement, enterprise control, support term, or price is active for a particular account. The planned Pro pilot has no commissioned self-serve checkout; eligibility, team and vault limits, and pricing remain TBD. Verify the deployed service and written plan terms; see the enterprise adoption guide for shipped-versus-planned gates.

Links

• phm.dev -- Hosted site; deployment and authenticated acceptance remain separate from source
• Documentation map
• Getting Started Guide
• Safe delegation quickstart
• Enterprise adoption guide
• Agent delegation templates
• Examples index
• Security Model
• Threat Model
• Roadmap
• Troubleshooting
• Changelog
• Contributing
• Support
• Governance
• Citation

Contributing

Focused contributions are welcome. Start with CONTRIBUTING.md, the examples index, or a discussion. The project does not assume that a good first issue label is populated. Follow CODE_OF_CONDUCT.md, and use SECURITY.md rather than a public issue for vulnerabilities.

Star history

If Phantom saves you from leaking a key — or even just from worrying about it — please star the repo ⭐. It's the single biggest signal we use to know what to build next.

License

MIT — see LICENSE.