agentcorp-broker 0.1.0-alpha.1

package/CHANGELOG.md

@@ -0,0 +1,52 @@
+# Changelog
+
+All notable changes to AgentCorp will be documented in this file.
+This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html) after the first public release.
+
+---
+
+## [Unreleased]
+
+## [0.1.0-alpha.1] — 2026-09-06
+
+Initial public developer preview release of AgentCorp: a local-first coordination broker for teams of AI agents.
+
+### Core Architecture & Coordination Broker
+- **Local-First MCP Broker**: Implemented single-writer, multi-reader architecture coordinating AI agents across distinct roles without coupling agents to a single vendor, framework, or model.
+- **Role Isolation & Access Control**: Strict caller role authentication enforcing allowed peer communications and role-scoped artifact visibility rules defined in `org.toml`.
+- **Approval-Bound Task Activation**: Gated task handoffs where tasks remain in `proposed` state and are hidden from intended assignees until linked proposals are human-approved.
+- **Prioritized Work Queue (`get_work_queue`)**: Unified endpoint returning unread messages, active tasks, live role presence, and ranked actionable suggestions in one call.
+- **Idempotent Handoff Acceptance (`accept_handoff`)**: Atomically acknowledges approved proposals and initiates task transitions without duplicate executions.
+
+### Reliability, Self-Healing & Concurrency
+- **Resilient Stdio MCP Proxy**: Automatic daemon discovery, bounded exponential backoff (100ms–2000ms), and transparent auto-spawn recovery when the broker daemon restarts or crashes.
+- **Cross-Process Startup Lock**: Atomic file-locking mechanism (`.agentcorp/daemon.json.lock`) with active PID probes and stale lock recovery (>10s) ensuring exactly one daemon spawns per project.
+- **Working Directory Isolation**: Automatic path resolution allowing MCP host processes to run from arbitrary working directories while automatically deriving project-scoped database, credentials, logs, and daemon control files from `org.toml`.
+- **Mutation Idempotency (`idempotency_key`)**: Role-scoped composite primary key `(role_id, key)` with SHA-256 operation and payload hash validation, rejecting cross-role key collisions and stale mismatches.
+- **Re-Entrant Atomic Transactions**: SQLite transaction management with depth tracking guaranteeing atomic commits and clean rollbacks for business mutations and idempotency records.
+- **Concurrent Database Startup**: Installs SQLite's busy handler before WAL negotiation and migrations so simultaneous worker connections wait for initialization locks instead of failing intermittently.
+- **Collision-Free Daemon Default**: Background startup now selects an available loopback port by default and records actionable startup failures in the bounded daemon log.
+
+### Persistence, Migrations & Bounded Storage
+- **Transactional SQLite Persistence**: Built on Node.js 22 built-in `node:sqlite` with WAL journal mode, busy timeouts, and versioned schema migrations (Versions 1 through 7).
+- **Hard Resource Bounds**: Enforced 2 MB HTTP body limits, 1 MB message payload limits, 5 MB artifact limits, and configurable audit payload extraction budgets (`max_audit_payload_bytes`).
+- **Opaque Cursor Pagination**: Cursor-based pagination across inboxes, tasks, threads, approvals, and artifacts.
+- **History Pruning & Compaction**: `agentcorp prune` command with foreign-key reply chain detachment, dry-run simulation, and SQLite WAL compaction (`agentcorp compact`).
+
+### Human Oversight & Operational Observability
+- **Human Review CLI (`agentcorp review`)**: Interactive terminal-native review loop with approve, edit-and-approve, reject, skip, and quit actions.
+- **Dark Glassmorphic Web Dashboard (`agentcorp console`)**: Live real-time dashboard powered by Server-Sent Events (`/api/events`) with side-by-side JSON diff editor and live status telemetry.
+- **Role Presence & Activity Freshness**: Telemetry tracking role liveness, last seen timestamps, and activity freshness (`fresh`, `idle`, `stale`).
+- **Diagnostic Health Check (`agentcorp doctor`)**: Comprehensive operational diagnostics inspecting configuration, database schema, credentials, daemon health, and crash logs.
+- **Structured Log Rotation & Sanitization**: Bounded rotating daemon logs (`.agentcorp/daemon.log`, 5 MB max with 3 backups) with payload redaction protecting conversation confidentiality, alongside fatal crash diagnostics (`.agentcorp/crash.log`).
+
+### Release Engineering & Documentation
+- **Strict Test Typechecking**: The default TypeScript check now covers both production source and tests, with a separate production-only build configuration.
+- **Package Boundary Regression Test**: The clean-install smoke suite asserts that source, tests, runtime state, internal design notes, and binary documentation assets cannot enter the npm tarball.
+- **Beginner and Maintainer Guides**: Added end-to-end first-workflow, troubleshooting, GitHub setup, npm preview publication, verification, and recovery instructions.
+- **Dogfooding Findings**: Documented the observed Codex-Gemini collaboration strengths, host-lifecycle friction, and a bounded adapter architecture for future autonomous invocation.
+
+### Audit Integrity & Truncation Semantics
+- **Explicit Audit Semantics**: Formatted Markdown and JSON audit exports (`coord/audit.md`, `coord/audit.json`) distinguishing row-limit truncation (`rowLimitTruncated`) from field-level clipping (`fieldClippingActive`).
+- **SQLite-Layer Payload Bounding**: UTF-8 BLOB-cast byte-aware truncation using `OCTET_LENGTH` and `SUBSTR` preventing native and V8 heap inflation on oversized audit records.
+- **Memory-Isolated Approvals**: Omission of discarded `edited_payload` from SQL query projections.

package/CONTRIBUTING.md

@@ -0,0 +1,20 @@
+# Contributing to AgentCorp
+
+Thank you for helping build vendor-neutral agent coordination infrastructure.
+
+## Local setup
+
+1. Install Node.js 22.13 or newer.
+2. Run `npm ci` for a lockfile-reproducible dependency install.
+3. Run `npm run check`, `npm run build`, and `npm test` before opening a pull request.
+
+Keep changes focused and add tests for new behavior. Protocol or persistence
+changes should update the design spec or an architecture note. Security
+invariants must fail closed: an error must never deliver a message, approve a
+transition, or broaden artifact visibility.
+
+Use Conventional Commit-style subjects when practical, such as `feat:`,
+`fix:`, `docs:`, and `test:`.
+
+By contributing, you agree that your contribution is licensed under the
+Apache License 2.0.

package/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 AgentCorp contributors
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

package/README.md

@@ -0,0 +1,250 @@
+# AgentCorp
+
+AgentCorp is a local-first coordination broker for teams of AI agents. It gives
+MCP-capable agents stable roles, typed messages, durable tasks, approval gates,
+and access-controlled artifacts without coupling the agents to one vendor or
+model.
+
+> Status: **developer preview (0.1.0-alpha.1)**.
+> Central long-lived broker daemon, terminal-native human console & review loop,
+> real-time SSE event pipeline with 15s keep-alive heartbeat, dark glassmorphic web dashboard, and audit exports.
+
+![AgentCorp Architecture Blueprint](./docs/images/architecture.jpg)
+
+## What works today
+
+- Long-lived broker daemon over MCP Streamable HTTP, Admin REST API, and Server-Sent Events (`/api/events`)
+- Terminal-native human console (`agentcorp console`) and interactive review loop (`agentcorp review`)
+- Sleek dark glassmorphic web dashboard with live push updates and side-by-side JSON diff editor
+- Thin role-bound stdio adapters proxying to the central daemon
+- Role definitions and allowed communication paths loaded from `org.toml`
+- Per-role and admin credentials stored securely in `.agentcorp/credentials.json`
+- Role-bound MCP connections with connection-enforced caller identity
+- Typed messages with durable status history
+- Approval-bound task handoffs: intended assignees cannot see or start proposed work before approval
+- Prioritized per-role work queues with executable next-action suggestions
+- Idempotent handoff acceptance that acknowledges and starts approved work in one call
+- Safe-by-default approval policy evaluation
+- Human approve, edit-and-approve, and reject commands with full audit logging
+- Validated task lifecycle transitions
+- SHA-256-addressed artifact metadata with role-scoped read checks
+- Transactional SQLite persistence with versioned schema migrations
+- Automated human-readable Markdown and JSON audit trail exports (`coord/`)
+- Public TypeScript API plus the `agentcorp` CLI
+
+## Requirements
+
+- Node.js 22.13 or newer
+
+AgentCorp uses Node's built-in `node:sqlite` module to avoid native npm
+dependencies. On Node 22 it may print an experimental-feature warning; it does
+not require an experimental flag from 22.13 onward.
+
+## Quick start
+
+New to Node.js, terminals, or MCP? Follow the complete
+[beginner setup guide](./docs/getting-started.md). The short path is below.
+
+Until the first npm release, run the CLI from this repository:
+
+```sh
+npm install
+npm run build
+node dist/cli.js init
+node dist/cli.js validate
+node dist/cli.js start
+```
+
+After publication:
+
+```sh
+npm install --global agentcorp-broker
+agentcorp init
+agentcorp validate
+agentcorp start
+```
+
+`agentcorp init` creates an architect/developer organization with conservative
+approval defaults and generates local credentials under `.agentcorp/credentials.json`.
+Runtime state under `.agentcorp/` should not be committed.
+
+## Connect agents over MCP
+
+Each local MCP connection is bound to exactly one configured role. When your
+MCP host does not preserve the project working directory, pass an absolute path
+to `--config`. AgentCorp automatically derives project-scoped `--db`, `--credentials`,
+and daemon control paths relative to the config file (or you can provide explicit overrides):
+
+```json
+{
+  "mcpServers": {
+    "agentcorp-architect": {
+      "command": "agentcorp",
+      "args": [
+        "--config", "/absolute/project/org.toml",
+        "mcp", "--role", "architect"
+      ]
+    }
+  }
+}
+```
+
+Configure the second agent with the same config path, changing only the final
+role to `developer`. You may also explicitly provide `--db /absolute/project/.agentcorp/agentcorp.db`
+if using a custom database location.
+
+The MCP tool surface includes 15 tools:
+
+- `register_role`, `whoami`
+- `create_task`, `list_tasks`, `get_work_queue`, `update_task_status`
+- `send_message`, `get_inbox`, `acknowledge_message`, `accept_handoff`, `get_thread`
+- `create_artifact`, `list_artifacts`, `get_artifact`
+- `get_operation`
+
+## Efficient agent loop
+
+Use `get_work_queue` as the first AgentCorp call in every agent session and
+again after each handoff or status change. It combines unread messages, active
+tasks, live role presence, and prioritized next actions so an agent does not need
+to independently reconcile `get_inbox`, `list_tasks`, and task threads.
+
+For a gated task handoff:
+
+1. The planner creates the task with an intended `assigned_to` and sends a
+   linked `proposal`.
+2. Until that proposal is approved, the task remains `proposed` and is hidden
+   from the intended assignee.
+3. Approval atomically activates the task as `assigned` and exposes the
+   proposal in the assignee's work queue.
+4. The assignee calls `accept_handoff` once. AgentCorp acknowledges the proposal
+   and requests the `in_progress` transition without creating duplicates on
+   retries.
+
+> [!IMPORTANT]
+> **LLM Wake Limitations & Handoff Best Practices**:
+> - **Passive MCP Protocol**: MCP servers expose tools over standard JSON-RPC; they **cannot independently wake an idle LLM** inside an external host (Antigravity, Codex, Cursor, Claude Desktop). Standing agent instructions must invoke `get_work_queue` at session start.
+> - **Lightweight ID / Summary Handoffs**: When agents coordinate or hand off work, always pass concise IDs and brief summaries (e.g., `taskId`, `messageId`, summary of diffs/artifacts) rather than dumping whole codebase contexts or large files into prompt memory. Artifacts should be retrieved on demand with `get_artifact`.
+
+## Human approval console
+
+AgentCorp offers first-class human-in-the-loop oversight designed for terminal-first developer workflows with an optional live web dashboard.
+
+### 1. Terminal-Native Interaction (Preferred)
+
+Launch the status summary or dive straight into interactive sign-off:
+
+```sh
+# View daemon status, task counts, and pending approvals
+agentcorp console
+
+# Interactive sign-off loop ([a]pprove, [e]dit & approve, [r]eject, [s]kip, [q]uit)
+agentcorp review
+```
+
+### 2. Live Web Dashboard
+
+Open the modern dark glassmorphic dashboard with live push updates over Server-Sent Events (`/api/events`):
+
+```sh
+agentcorp console --browser
+```
+
+Features include:
+- Real-time approval feed with side-by-side JSON diff editor
+- Role presence indicators (online, idle, offline) and connection health
+- Role inboxes & active task threads
+- Artifact catalog viewer
+- Runtime policy inspector and instant toggle switch (enable / disable)
+
+### 3. Direct Scriptable CLI Commands
+
+Messages and transitions that do not match an `auto_approve` rule are held by default.
+
+```sh
+agentcorp approvals list
+agentcorp approvals approve <approval-id> --note "Reviewed"
+agentcorp approvals approve <approval-id> --payload '{"revised":"payload"}'
+agentcorp approvals reject <approval-id> --note "Needs a safer plan"
+agentcorp policies list
+agentcorp policies set '{"id":"safe-reports","subject":"message","priority":80,"message_type":"report","risk_tags":["read_only"],"action":"auto_approve"}'
+agentcorp policies disable safe-reports
+```
+
+The recipient cannot see a pending message. Approval changes its state to
+`delivered`; rejection keeps it out of the inbox.
+
+## Storage lifecycle and bounding
+
+AgentCorp enforces bounded memory and disk usage to protect long-running daemons from denial of service and memory exhaustion:
+
+- **Size limits**:
+  - HTTP body: 2 MB maximum (returns HTTP 413 `PAYLOAD_TOO_LARGE`).
+  - Message payload: 1 MB maximum (`PAYLOAD_TOO_LARGE`).
+  - Artifact content: 5 MB maximum (`ARTIFACT_TOO_LARGE`).
+  - Audit payload budget: Operator-configurable (`max_audit_payload_bytes`, defaults to 64 KB).
+- **Cursor pagination**:
+  - Task, message, inbox, and artifact queries are bounded to 50 items by default (max 200).
+  - MCP tools and Admin REST endpoints support `limit` and opaque `cursor` pagination with optional `envelope` payloads and `X-Next-Cursor` headers.
+- **History pruning and WAL compaction**:
+  ```sh
+  # Dry-run simulate pruning resolved tasks and messages older than 30 days
+  agentcorp prune --older-than 30
+
+  # Execute pruning and compact SQLite database
+  agentcorp prune --older-than 30 --execute --compact
+
+  # Manually checkpoint WAL and vacuum freed pages
+  agentcorp compact
+  ```
+- **Bounded audit export**:
+  ```sh
+  # Export the 100 most recent records or only records since a timestamp
+  agentcorp audit export --limit 100 --since 2026-09-01T00:00:00Z
+  ```
+
+## Self-healing, observability, and idempotency
+
+AgentCorp includes enterprise-grade operational safeguards to guarantee durable multi-agent sessions:
+
+- **Resilient Stdio MCP Proxy**: When running inside an IDE via stdio (`agentcorp mcp --role developer`), the proxy wraps connections with auto-reconnection and bounded exponential backoff. If the central daemon process crashes or restarts, running agent sessions transparently recover without transport exceptions.
+- **Mutation Idempotency & Operation Lookup**: Mutations (`create_task`, `send_message`, `create_artifact`, `accept_handoff`, `update_task_status`) accept an optional `idempotency_key`. Replays return the exact recorded response without duplicate side effects. Agents can query cached outcomes at any time using `get_operation`.
+- **Role Presence & Last-Seen Telemetry**: Every authenticated agent interaction touches role presence. `get_work_queue`, `/health`, `/api/status`, and the Web Console report real-time statuses (`online` <1m, `idle` 1–5m, `offline` >5m).
+- **Bounded Rotating Daemon Logs & Crash Diagnostics**: Daemon events are written to `.agentcorp/daemon.log` with size-bounded rotation (5 MB with 3 backups). Uncaught exceptions record complete diagnostic forensics into `.agentcorp/crash.log`.
+- **System Doctor**: Run `agentcorp doctor` at any time to verify configuration validity, database integrity (`PRAGMA integrity_check`), credential coverage, daemon status, and log health.
+
+## Policy semantics
+
+Policies are initially seeded from `org.toml` into SQLite. After the first run,
+the database is the source of truth. Rules are evaluated by descending
+`priority`; every populated condition on a rule must match. Risk tags use
+subset matching: all tags named by the rule must be present on the request.
+
+If no rule matches, AgentCorp requires human approval. The reserved
+`delegate_to_role` action is rejected in v0 rather than silently weakened.
+
+## Development
+
+```sh
+npm run check
+npm test
+npm run build
+npm pack --dry-run
+```
+
+The main design document is [docs/design-spec.md](./docs/design-spec.md).
+The delivery sequence and design gaps are tracked in
+[docs/ROADMAP.md](./docs/ROADMAP.md).
+Maintainers should follow the [release guide](./docs/RELEASING.md); do not
+publish directly from an unverified working tree.
+
+## Security model
+
+AgentCorp 0.1.0-alpha.1 enforces caller identity, communication routing, policy gating, and artifact access control through its central authenticated daemon over MCP Streamable HTTP and role-bound stdio adapters using header-only bearer tokens.
+
+As a local developer preview, AgentCorp does not sandbox an underlying agent's general shell or filesystem access. Keep `.agentcorp/credentials.json`, `.agentcorp/agentcorp.db`, and the administrative CLI unavailable to untrusted processes. Role capabilities are currently descriptive metadata, and host-agent capability sandboxing is scheduled for Milestone 4.
+
+See [SECURITY.md](./SECURITY.md) for reporting and deployment guidance.
+
+## License
+
+Apache-2.0. See [LICENSE](./LICENSE).

package/SECURITY.md

@@ -0,0 +1,23 @@
+# Security policy
+
+AgentCorp is a developer preview (0.1.0-alpha.1) and has not yet received an independent
+security audit. Do not use as a hostile multi-tenant security boundary.
+
+## Supported versions
+
+Only the latest published preview version receives security fixes before 1.0.
+
+## Reporting a vulnerability
+
+Please report vulnerabilities privately through GitHub Security Advisories at
+[https://github.com/Shash-J/AgentCorp/security/advisories](https://github.com/Shash-J/AgentCorp/security/advisories).
+Do not open a public issue containing exploit details. Include affected versions,
+reproduction steps, impact, and any known mitigation.
+
+## Deployment guidance
+
+- Restrict filesystem access to `.agentcorp/agentcorp.db`.
+- Do not expose the stdio process directly to a network.
+- Keep administrative approval commands unavailable to agent processes.
+- Review `auto_approve` rules narrowly and prefer explicit risk tags.
+- Treat external artifact URIs as untrusted input.

package/dist/audit.d.ts

@@ -0,0 +1,25 @@
+import type { AgentCorpBroker } from "./broker.js";
+import { type AuditArtifactRecord, type AuditExportOptions, type AuditSnapshotMetadata, type AuditTaskRecord, type MessageRecord, type PendingApproval } from "./types.js";
+export interface AuditSnapshot {
+    exportedAt: string;
+    company: string;
+    metadata: AuditSnapshotMetadata;
+    roles: Array<{
+        id: string;
+        model?: string | undefined;
+        capabilities: string[];
+        allowedPeers: string[];
+        boundAgent?: string | undefined;
+    }>;
+    tasks: AuditTaskRecord[];
+    messages: MessageRecord[];
+    approvals: PendingApproval[];
+    artifacts: AuditArtifactRecord[];
+}
+export declare function generateAuditSnapshot(broker: AgentCorpBroker, options?: AuditExportOptions): AuditSnapshot;
+export declare function formatAuditMarkdown(snapshot: AuditSnapshot): string;
+export declare function exportAuditTrail(broker: AgentCorpBroker, outputDir?: string, options?: AuditExportOptions): {
+    markdownPath: string;
+    jsonPath: string;
+    snapshot: AuditSnapshot;
+};