chore: bump version to 0.5.0

Author: iamvirul

CHANGELOG.md

@@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.5.0] - 2026-04-20
+
 ### Added
 - **Supervisor evaluation loop** — the Supervisor is now an active quality gate, not advisory only. When it rejects an agent output (`approved: false`), the orchestrator re-invokes the agent with the flags and recommendation appended as feedback, up to `COUNCIL_EVAL_RETRIES` times (default: 2 → 3 total attempts). If the retry budget is exhausted and the output is still flagged, the result is surfaced anyway with the flags visible — no silent passes.
 - `COUNCIL_EVAL_RETRIES` env var — clamped to `[0, 5]`. Set to `0` to restore pre-0.5 advisory-only behaviour. Non-integer values fall back to the default.
@@ -125,7 +127,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Executor runs with explicit `permissionMode: 'acceptEdits'` rather than relying on inherited default
 - `@anthropic-ai/claude-agent-sdk` pinned to `^0.2.101` (no `latest` in production)
 
-[Unreleased]: https://github.com/iamvirul/the-council/compare/v0.4.0...HEAD
+[Unreleased]: https://github.com/iamvirul/the-council/compare/v0.5.0...HEAD
+[0.5.0]: https://github.com/iamvirul/the-council/compare/v0.4.0...v0.5.0
 [0.4.0]: https://github.com/iamvirul/the-council/compare/v0.3.0...v0.4.0
 [0.3.0]: https://github.com/iamvirul/the-council/compare/v0.2.3...v0.3.0
 [0.2.3]: https://github.com/iamvirul/the-council/compare/v0.2.2...v0.2.3

README.md

@@ -7,7 +7,7 @@
 
 A four-tier AI agent orchestration system built as a Claude Code MCP server.
 
-The Council is a TypeScript [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server with four Claude agents. When you give it a problem, it figures out the complexity and sends it to the right agents. A formatting task goes straight to the fast Aide (Haiku). A coding task goes to the Executor (Sonnet). A design or architecture problem first goes through the Chancellor (Opus) for a plan, then the Executor runs each step, delegating simple sub-tasks to the Aide. After each agent produces output, the Supervisor (Haiku) reviews it for quality and flags any issues before results surface to the caller.
+The Council is a TypeScript [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server with four Claude agents. When you give it a problem, it figures out the complexity and sends it to the right agents. A formatting task goes straight to the fast Aide (Haiku). A coding task goes to the Executor (Sonnet). A design or architecture problem first goes through the Chancellor (Opus) for a plan, then the Executor runs each step, delegating simple sub-tasks to the Aide. After each agent produces output, the Supervisor (Haiku) acts as an active quality gate — rejecting outputs that miss the mark and triggering a retry with its feedback attached, up to a configurable limit.
 
 Sub-agents run via the `claude` CLI — the same one Claude Code uses. **If you already have Claude Code installed, no separate API key or extra cost is needed.** The install script finds the `claude` binary and wires it up automatically. Alternatively, set `ANTHROPIC_API_KEY` in the MCP server env for CI or API-key-based setups.
 
@@ -250,7 +250,7 @@ Use `get_council_state` at any point to inspect a session. Each session tracks:
 - Executor step results
 - Aide task results
 - Supervisor verdicts (one per Executor step and Aide task)
-- Metrics: total agent calls, agents invoked, duration, caveman mode
+- Metrics: total agent calls, agents invoked, duration, caveman mode, eval retries
 
 ---
 
@@ -281,15 +281,18 @@ src/
     chancellor/     # Chancellor agent wrapper
     executor/       # Executor agent wrapper
     aide/           # Aide agent wrapper
-    supervisor/     # Supervisor agent wrapper (non-blocking quality review)
+    supervisor/     # Supervisor agent wrapper (active quality gate with eval loop)
   infra/            # External dependencies
-    agent-sdk/      # runner.ts - spawns claude CLI subprocess for each agent
-    config/         # caveman.ts - token compression mode config
+    agent-sdk/      # runner.ts (subprocess), parse.ts (JSON extractor), run-with-validation.ts (retry wrapper)
+    config/         # caveman.ts (token compression), eval.ts (Supervisor retry budget)
     state/          # Session store: memory / file / SQLite backends
     logging/        # pino structured logger (stderr only)
   mcp/
     server/         # MCP server setup, tool registration, lifecycle
     tools/          # Zod schemas for all tool inputs
+tests/
+  unit/             # Fast isolated tests: config, schemas, stores, agent invokers, orchestrator
+  integration/      # Real filesystem tests for FileStore and SQLiteStore
 ```
 
 ---

package.json

@@ -1,6 +1,6 @@
 {
   "name": "council-mcp",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "description": "Three-tier AI agent MCP orchestration system: Chancellor (Opus), Executor (Sonnet), Aide (Haiku)",
   "type": "module",
   "main": "dist/index.js",