mirror of
https://github.com/coleam00/Archon
synced 2026-04-21 13:37:41 +00:00
14 commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Rasmus Widing
|
81859d6842
|
fix(providers): replace Claude SDK embed with explicit binary-path resolver (#1217)
* feat(providers): replace Claude SDK embed with explicit binary-path resolver Drop `@anthropic-ai/claude-agent-sdk/embed` and resolve Claude Code via CLAUDE_BIN_PATH env → assistants.claude.claudeBinaryPath config → throw with install instructions. The embed's silent failure modes on macOS (#1210) and Windows (#1087) become actionable errors with a documented recovery path. Dev mode (bun run) remains auto-resolved via node_modules. The setup wizard auto-detects Claude Code by probing the native installer path (~/.local/bin/claude), npm global cli.js, and PATH, then writes CLAUDE_BIN_PATH to ~/.archon/.env. Dockerfile pre-sets CLAUDE_BIN_PATH so extenders using the compiled binary keep working. Release workflow gets negative and positive resolver smoke tests. Docs, CHANGELOG, README, .env.example, CLAUDE.md, test-release and archon skills all updated to reflect the curl-first install story. Retires #1210, #1087, #1091 (never merged, now obsolete). Implements #1176. * fix(providers): only pass --no-env-file when spawning Claude via Bun/Node `--no-env-file` is a Bun flag that prevents Bun from auto-loading `.env` from the subprocess cwd. It is only meaningful when the Claude Code executable is a `cli.js` file — in which case the SDK spawns it via `bun`/`node` and the flag reaches the runtime. When `CLAUDE_BIN_PATH` points at a native compiled Claude binary (e.g. `~/.local/bin/claude` from the curl installer, which is Anthropic's recommended default), the SDK executes the binary directly. Passing `--no-env-file` then goes straight to the native binary, which rejects it with `error: unknown option '--no-env-file'` and the subprocess exits code 1. Emit `executableArgs` only when the target is a `.js` file (dev mode or explicit cli.js path). Caught by end-to-end smoke testing against the curl-installed native Claude binary. * docs: record env-leak validation result in provider comment Verified end-to-end with sentinel `.env` and `.env.local` files in a workflow CWD that the native Claude binary (curl installer) does not auto-load `.env` files. With Archon's full spawn pathway and parent env stripped, the subprocess saw both sentinels as UNSET. The first-layer protection in `@archon/paths` (#1067) handles the inheritance leak; `--no-env-file` only matters for the Bun-spawned cli.js path, where it is still emitted. * chore(providers): cleanup pass — exports, docs, troubleshooting Final-sweep cleanup tied to the binary-resolver PR: - Mirror Codex's package surface for the new Claude resolver: add `./claude/binary-resolver` subpath export and re-export `resolveClaudeBinaryPath` + `claudeFileExists` from the package index. Renames the previously single `fileExists` re-export to `codexFileExists` for symmetry; nothing outside the providers package was importing it. - Add a "Claude Code not found" entry to the troubleshooting reference doc with platform-specific install snippets and pointers to the AI Assistants binary-path section. - Reframe the example claudeBinaryPath in reference/configuration.md away from cli.js-only language; it accepts either the native binary or cli.js. * test+refactor(providers, cli): address PR review feedback Two test gaps and one doc nit from the PR review (#1217): - Extract the `--no-env-file` decision into a pure exported helper `shouldPassNoEnvFile(cliPath)` so the native-binary branch is unit testable without mocking `BUNDLED_IS_BINARY` or running the full sendQuery pathway. Six new tests cover undefined, cli.js, native binary (Linux + Windows), Homebrew symlink, and suffix-only matching. Also adds a `claude.subprocess_env_file_flag` debug log so the security-adjacent decision is auditable. - Extract the three install-location probes in setup.ts into exported wrappers (`probeFileExists`, `probeNpmRoot`, `probeWhichClaude`) and export `detectClaudeExecutablePath` itself, so the probe order can be spied on. Six new tests cover each tier winning, fall-through ordering, npm-tier skip when not installed, and the which-resolved-but-stale-path edge case. - CLAUDE.md `claudeBinaryPath` placeholder updated to reflect that the field accepts either the native binary or cli.js (the example value was previously `/absolute/path/to/cli.js`, slightly misleading now that the curl-installer native binary is the default). Skipped from the review by deliberate scope decision: - `resolveClaudeBinaryPath` async-with-no-await: matches Codex's resolver signature exactly. Changing only Claude breaks symmetry; if pursued, do both providers in a separate cleanup PR. - `isAbsolute()` validation in parseClaudeConfig: Codex doesn't do it either. Resolver throws on non-existence already. - Atomic `.env` writes in setup wizard: pre-existing pattern this PR touched only adjacently. File as separate issue if needed. - classifyError branch in dag-executor for setup errors: scope creep. - `.env.example` "missing #" claim: false positive (verified all CLAUDE_BIN_PATH lines have proper comment prefixes). * fix(test): use path.join in Windows-compatible probe-order test The "tier 2 wins (npm cli.js)" test hardcoded forward-slash path comparisons, but `path.join` produces backslashes on Windows. Caused the Windows CI leg of the test suite to fail while macOS and Linux passed. Use `path.join` for both the mock return value and the expectation so the separator matches whatever the platform produces. |
||
|
Rasmus Widing
|
c1ed76524b
|
refactor: extract providers from @archon/core into @archon/providers (#1137)
* refactor: extract providers from @archon/core into @archon/providers Move Claude and Codex provider implementations, factory, and SDK dependencies into a new @archon/providers package. This establishes a clean boundary: providers own SDK translation, core owns business logic. Key changes: - New @archon/providers package with zero-dep contract layer (types.ts) - @archon/workflows imports from @archon/providers/types — no mirror types - dag-executor delegates option building to providers via nodeConfig - IAgentProvider gains getCapabilities() for provider-agnostic warnings - @archon/core no longer depends on SDK packages directly - UnknownProviderError standardizes error shape across all surfaces Zero user-facing changes — same providers, same config, same behavior. * refactor: remove config type duplication and backward-compat re-exports Address review findings: - Move ClaudeProviderDefaults and CodexProviderDefaults to the @archon/providers/types contract layer as the single source of truth. @archon/core/config/config-types.ts now imports from there. - Remove provider re-exports from @archon/core (index.ts and types/). Consumers should import from @archon/providers directly. - Update @archon/server to depend on @archon/providers for MessageChunk. * refactor: move structured output validation into providers Each provider now normalizes its own structured output semantics: - Claude already yields structuredOutput from the SDK's native field - Codex now parses inline agent_message text as JSON when outputFormat is set, populating structuredOutput on the result chunk This eliminates the last provider === 'codex' branch from dag-executor, making it fully provider-agnostic. The dag-executor checks structuredOutput uniformly regardless of provider. Also removes the ClaudeCodexProviderDefaults deprecated alias — all consumers now use ClaudeProviderDefaults directly. * fix: address PR review — restore warnings, fix loop options, cleanup Critical fixes: - Restore MCP missing env vars user-facing warning (was silently dropped) - Restore Haiku + MCP tool search warning - Fix buildLoopNodeOptions to pass workflow-level nodeConfig (effort, thinking, betas, sandbox were silently lost for loop nodes) - Add TODO(#1135) comments documenting env-leak gate gap Cleanup: - Remove backward-compat type aliases from deps.ts (keep WorkflowTokenUsage) - Remove 26 unnecessary eslint-disable comments from test files - Trim internal helpers from providers barrel (withFirstMessageTimeout, getProcessUid, loadMcpConfig, buildSDKHooksFromYAML) - Add @archon/providers dep to CLI package.json - Fix 8 stale documentation paths pointing to deleted core/src/providers/ - Add E2E smoke test workflows for both Claude and Codex providers * fix: forward provider system warnings to users in dag-executor The dag-executor only forwarded system chunks starting with "MCP server connection failed:" — all other provider warnings (missing env vars, Haiku+MCP, structured output issues) were logged but never reached the user. Now forwards all system chunks starting with ⚠️ (the prefix providers use for user-actionable warnings). * fix: add providers package to Dockerfile and fix CI module resolution - Add packages/providers/ to all three Dockerfile stages (deps, production package.json copy, production source copy) - Replace wildcard export map (./*) with explicit subpath entries to fix module resolution in CI (bun workspace linking) * chore: update bun.lock for providers package exports |
||
|
DIY Smart Code
|
899e993d7a
|
fix(docker): fix Docker build failures and add CI guard (#1022)
* fix(docker): update Bun base image from 1.2 to 1.3 The lockfile was generated with Bun 1.3.x locally but the Docker image used oven/bun:1.2-slim. Bun 1.3 changed the lockfile format, causing --frozen-lockfile to fail during docker build. * fix(docker): pin Bun to exact version 1.3.9 matching lockfile Floating tag 1.3-slim resolved to 1.3.11 which has a different lockfile format than 1.3.9 used to generate bun.lock. Pin to exact patch version to prevent --frozen-lockfile failures. * fix(docker): add missing docs-web workspace package.json The docs-web package was added as a workspace member but its package.json was never added to the Dockerfile COPY steps. This caused bun install --frozen-lockfile to fail because the workspace layout in Docker didn't match the lockfile. * fix(docker): use hoisted linker for Vite/Rollup compatibility Bun's default "isolated" linker stores packages in node_modules/.bun/ with symlinks that Vite's Rollup bundler cannot resolve during production builds (e.g., remark-gfm → mdast-util-gfm chain). Using --linker=hoisted gives the classic flat node_modules layout that Rollup expects. Local dev is unaffected (Vite dev server handles the isolated layout fine). * ci: pin Bun version to 1.3.9 and add Docker build check - Align CI Bun version (was 1.3.11) with Dockerfile and local dev (1.3.9) to prevent lockfile format mismatches between environments - Add docker-build job to test.yml that builds the Docker image on every PR — catches Dockerfile regressions (missing workspace packages, linker issues, build failures) before they reach deploy * fix(ci): add permissions for GHA cache and tighten Bun engine - Add actions: write permission to docker-build job so GHA layer cache writes succeed on PRs from forks - Tighten package.json engines.bun from >=1.0.0 to >=1.3.9 to document the minimum version that matches the lockfile format * fix(ci): add smoke test, align Bun version across all workflows Review fixes: - Add load: true + health endpoint smoke test to docker-build CI job so we verify the image actually starts, not just compiles - Align Bun 1.3.9 in deploy-docs.yml and release.yml (were still 1.3.11) - Document why docs-web source is intentionally omitted from Docker * chore: float Docker to bun:1.3 and align CI to 1.3.11 - Dockerfile: oven/bun:1.3-slim (auto-tracks latest 1.3.x patches) - CI workflows: bun-version 1.3.11 (current latest, reproducible) - engines.bun: >=1.3.9 (minimum for local devs) Lockfile format is stable across 1.3.x patches, so this is safe. * fix(docker,ci): pin Docker to 1.3.11, loosen engines, harden smoke test - Dockerfile: pin oven/bun:1.3.11-slim (was floating 1.3-slim) so Docker builds are reproducible and match CI exactly. - package.json: loosen engines to ^1.3.0 so end users on any 1.3.x can run the CLI; CI/Docker remain pinned to the canonical latest. - CI smoke test: replace 'sleep 5' with curl --retry-connrefused, and move container cleanup to an 'if: always()' step so a failed health check no longer leaks the named container. --------- Co-authored-by: Rasmus Widing <rasmus.widing@gmail.com> |
||
|
Leex
|
b739fb667a |
fix(docker): CRLF entrypoint + Windows Docker Desktop docs
- .gitattributes: force LF for *.sh, docker-entrypoint.sh, Caddyfile* - docker-entrypoint.sh: normalize to LF - Caddyfile.example: normalize to LF - Dockerfile: add sed CRLF strip as defense-in-depth after COPY - docs/docker.md: add "Local Docker Desktop" section for Windows/macOS Closes #869 |
||
|
Cole Medin
|
ae346c2a67 |
feat: prepare for open-source migration to coleam00/Archon
- Replace all dynamous-community/remote-coding-agent references with coleam00/Archon - Replace all ghcr.io/dynamous-community/remote-coding-agent with ghcr.io/coleam00/archon - Change license from proprietary Dynamous to MIT - Fix cd directory name in docs (remote-coding-agent → Archon) - Remove hardcoded local paths from skills and docs - Add Windows x64 binary to release pipeline (cross-compiled from Linux) - Add --minify --bytecode flags to binary compilation - Create PowerShell install script (scripts/install.ps1) - Fix isBinaryBuild() detection for Bun 1.3.5+ (use import.meta.dir virtual FS check) - Scaffold Astro Starlight docs site at website/ (Astro 6 + Starlight 0.38) - Add deploy-docs.yml workflow for GitHub Pages - Update test.yml branch triggers (develop → dev) - Add install section with curl/PowerShell/Homebrew/Docker to README - Add badges and archon.diy docs link to README - Create SECURITY.md with vulnerability disclosure policy - Update CONTRIBUTING.md for public audience - Add website/ and eslint ignores for Astro-generated files Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
DIY Smart Code
|
380cfef5a8
|
feat(docker): complete Docker deployment setup (#756)
* fix: overhaul Docker setup for working builds and server deployments
Multi-stage Dockerfile: deps → web build → production image. Fixes
missing workspace packages (was 3/9, now all 9), adds Vite web UI
build, removes broken single-file bundle, uses --production install.
Merges docker-compose.yml and docker-compose.cloud.yml into a single
file with composable profiles (with-db, cloud). Fixes health check
path (/api/health), postgres volume (/data), adds Caddyfile.example.
* docs: add comprehensive Docker guide and update cloud-deployment.md
New docs/docker.md covers quick start, composable profiles, config,
cloud deployment with HTTPS, pre-built image usage, building, and
troubleshooting. Updates cloud-deployment.md to use the new single
compose file with profiles and fixes stale health endpoint paths.
* docs: restructure docker.md — prerequisites before commands
Moves .env and Caddyfile setup to a Prerequisites section at the top,
before any docker compose commands. Adds troubleshooting entry for the
"not a directory" Caddyfile mount error.
* fix: pass env_file to Caddy container for DOMAIN variable
Caddy needs {$DOMAIN} from .env but the container had no env_file.
Without it, {$DOMAIN} is empty and Caddy parses the site block as
a global options block, causing "unrecognized global option" error.
* docs: rewrite docker.md with server quickstart and fix auth guidance
Restructures around a step-by-step Quick Start that walks through the
full server deployment (Docker install → .env → Caddyfile → DNS → run).
Removes CLAUDE_USE_GLOBAL_AUTH references — Docker has no local claude
CLI, so users must provide CLAUDE_CODE_OAUTH_TOKEN or CLAUDE_API_KEY.
* feat: warn when Docker app falls back to SQLite with postgres running
When ARCHON_DOCKER=true and DATABASE_URL is not set, logs a warning
with the exact connection string to add to .env. Prevents users from
running --profile with-db and unknowingly using SQLite instead.
* feat: configurable data directory via ARCHON_DATA env var
Users can set ARCHON_DATA=/opt/archon-data in .env to control where
Archon stores workspaces, worktrees, artifacts, and logs on the host.
Defaults to a Docker-managed volume when not set.
* fix: fix volume permission errors with entrypoint script
Docker volume mounts create /.archon/ as root, but the app runs as
appuser (UID 1001). New docker-entrypoint.sh runs as root to fix
permissions, then drops to appuser via gosu. Works both when running
as root (default) and as non-root (--user flag, Kubernetes).
* fix: configure git credentials from GH_TOKEN in Docker entrypoint
Git inside the container can't authenticate for HTTPS clones without
credentials. The entrypoint now configures git url.insteadOf to inject
GH_TOKEN into GitHub HTTPS URLs automatically.
* security: use credential helper for GH_TOKEN instead of url.insteadOf
The url.insteadOf approach stored the raw token in ~/.gitconfig as a
key name, visible to any process. Credential helper keeps the token
in the environment only. Also fixes: chown -Rh (no symlink follow),
signal propagation (exec bun directly as PID 1), error diagnostics,
and deduplicates root/non-root branches via RUNNER variable.
* security: scope SSE flush_interval to /api/stream/*, harden headers
flush_interval -1 was global, disabling buffering for all endpoints.
Now scoped to @sse path matcher. Also adds HSTS, changes X-Frame-Options
to DENY, and trims the comment header.
* security: use env-var for postgres password, bind port to localhost
Hardcoded postgres:postgres with port exposed to 0.0.0.0 is a risk
on servers with permissive firewalls. Now uses POSTGRES_PASSWORD env
var with fallback, and binds to 127.0.0.1 only.
* fix: caddy depends_on app with service_healthy condition
Without the health condition, Caddy starts proxying before the app
is ready, returning 502s on first boot.
* fix: remove hardcoded container_name from caddy service
Hardcoded name prevents running multiple instances on the same host.
Other services already use Compose default naming.
* security: exclude .claude/ from Docker image
Skills, commands, rules, and prompt engineering details are not needed
at runtime and expose internal architecture in the production image.
* fix: assert web build produces index.html in Dockerfile
A silent Vite failure could produce an empty dist/ — the container
would start with a healthy backend but a broken UI serving 404s.
* chore: remove redundant WORKDIR in Dockerfile Stage 2
WORKDIR /app is inherited from Stage 1 (deps). Re-declaring it adds
a no-op layer and implies something changed.
* feat: add cloud-init config for automated server setup
New deploy/cloud-init.yml for VPS providers — paste into User Data
field to auto-install Docker, clone repo, build image, and configure
firewall. User only needs to edit .env and run docker compose up.
* feat: add optional Caddy basic auth for cloud deployments
Single env var (CADDY_BASIC_AUTH) expands to the full basicauth directive
or nothing when unset — no app changes needed. Webhooks and health check
are excluded. Documented in .env.example, deploy config, and docker.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(docker): add agent-browser + Chromium for E2E testing workflows
Enables E2E validation workflows (archon-validate-pr, validate-ui,
replicate-issue) to run inside Docker containers out of the box.
- Install system Chromium via apt-get (~200MB vs ~500MB Chrome for Testing)
- Install agent-browser@0.22.1 via npm (postinstall downloads Rust binary)
- Purge nodejs/npm after install to keep image lean
- Set AGENT_BROWSER_EXECUTABLE_PATH=/usr/bin/chromium
- agent-browser auto-detects Docker and adds --no-sandbox
Closes #787
* fix(docker): symlink agent-browser native binary before purging nodejs
The npm entry point (bin/agent-browser.js) is a Node.js wrapper that
launches the Rust binary. After purging nodejs/npm to save ~60MB, the
wrapper can't execute. Fix by copying the native Rust binary directly
to /usr/local/bin and symlinking agent-browser to it.
* feat(docker): add cookie-based form auth sidecar for Caddy
- Add auth-service/ Node.js sidecar (/verify, /login GET/POST, /logout)
- Use bcryptjs for password hashing, HMAC-SHA256 signed HttpOnly cookies
- Add auth-service to docker-compose.yml under ["auth"] profile (expose: not ports:)
- Restructure Caddyfile.example with handle blocks for Option A (form auth), Option B (basic auth), None
- Add AUTH_USERNAME, AUTH_PASSWORD_HASH, COOKIE_SECRET env vars to .env.example and deploy/.env.example
- Add Form-Based Authentication section to docs/docker.md
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: address review findings for auth-service (HIGH/MEDIUM)
Fixes applied:
- HIGH: validate AUTH_PASSWORD_HASH is a valid bcrypt hash at startup
(bcrypt.getRounds() guard — prevents silent lockout on placeholder hash)
- HIGH: add request method/URL context to unhandled error log + non-empty 500 body
- HIGH: add server.on('error') handler for port bind failures (EADDRINUSE/EACCES)
- HIGH: document AUTH_PORT/AUTH_SERVICE_PORT indirection in server.js comment
- HIGH: add auth-service/test.js with isSafeRedirect and cookie sign/verify tests
- MEDIUM: add escapeHtml() helper; apply to loginPage error param (latent XSS)
- MEDIUM: add 4 KB body size limit in readBody (prevents memory exhaustion)
- MEDIUM: export helpers + require.main guard (enables clean import-level testing)
- MEDIUM: fix docs/docker.md Step 4 instruction — clarify which handle block to comment out
Tests added:
- auth-service/test.js: 12 assertions for isSafeRedirect (safe paths + open redirect vectors)
- auth-service/test.js: 5 assertions for signCookie/verifyCookie round-trip and edge cases
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: escape $ in AUTH_PASSWORD_HASH example to prevent Docker Compose variable substitution
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* refactor(core): break up god function in command-handler (#742)
* refactor(core): break up god function in command-handler
Extract handleWorktreeCommand, handleWorkflowCommand, handleRepoCommand,
and handleRepoRemoveCommand from the 1300-line handleCommand switch
statement. Add resolveRepoArg helper to eliminate duplication between
repo and repo-remove cases. handleCommand now contains ~200 lines of
routing logic only.
* fix: address review findings from PR #742
command-handler.ts:
- Replace fragile 'success' in discriminator with proper ResolveRepoArgResult
discriminated union (ok: true/false) and fix misleading JSDoc
- Add missing error handling to worktree orphans, workflow cancel, workflow reload
- Fix isolation_env_id UUID used as filesystem path in worktree create/list/orphans
(look up working_path from DB instead)
- Add cmd. domain prefix to all log events per CLAUDE.md convention
- Add identifier/isolationEnvId context to repo_switch_failed and worktree_remove_failed logs
- Capture isCurrentCodebase before mutation in handleRepoRemoveCommand
- Hoist duplicated workflowCwd computation in handleWorkflowCommand
- Remove stale (Phase 3D) comment marker
docs:
- Remove all /command-invoke references from CLAUDE.md, README.md,
docs/architecture.md, and .claude/rules/orchestrator.md
- Update command list to match actual handleCommand cases
- Replace outdated routing examples with current AI router pattern
* refactor: remove MAX_WORKTREES_PER_CODEBASE limit
Worktree count is no longer restricted. Remove the constant, the
limit field from WorktreeStatusBreakdown, the limit_reached block
reason, formatWorktreeLimitMessage, and all associated tests.
* fix: address review findings — error handling, log prefixes, tests, docs
- Wrap workflow list discoverWorkflowsWithConfig in try/catch (was the
only unprotected async call among workflow subcommands)
- Cast error to Error before logging in workflow cancel/status catch blocks
- Add cmd. domain prefix to all command-handler log events (12 events)
- Update worktree create test to use UUID isolation_env_id with DB lookup
- Add resolveRepoArg boundary tests (/repo 0, /repo N > count)
- Add worktree cleanup subcommand tests (merged, stale, invalid type)
- Add updateConversation assertion to repo-remove session test
- Fix stale docs: architecture.md command handler section, .claude → .archon
paths, remove /command-invoke from commands-reference, fix github.md example
* feat(workflows)!: replace standalone loop with DAG loop node (#785)
* feat(workflows): add loop node type to DAG workflows
Add LoopNode as a fourth DAG node type alongside command, prompt, and
bash. Loop nodes run an AI prompt repeatedly until a completion signal
is detected (LLM-decided via <promise>SIGNAL</promise>) or a
deterministic bash condition succeeds (until_bash exit 0).
This enables Ralph-style autonomous iteration as a composable node
within DAG workflows — upstream nodes can produce plans/task lists
that feed into the loop, and downstream nodes can act on the loop's
output via $nodeId.output substitution.
Changes:
- Add LoopNodeConfig, LoopNode interface, isLoopNode type guard
- Add loop branch in parseDagNode with full validation
- Extract detectCompletionSignal/stripCompletionTags to executor-shared
- Add executeLoopNode function in dag-executor with iteration logic
- Add nodeId field to loop iteration event interfaces
- Add 17 new tests (9 loader + 8 executor)
- Add archon-test-loop-dag and archon-ralph-dag default workflows
The standalone loop: workflow type is preserved but deprecated.
* refactor(workflows): rewrite archon-ralph-dag prompt to match command quality bar
Expand the loop prompt from ~75 lines to ~430 lines with:
- 7 numbered phases with checkpoints (matching archon-implement.md pattern)
- Environment setup: dependency install, CLAUDE.md reading, git state check
- Explicit DO/DON'T implementation rules
- Per-failure-type validation handling (type-check, lint, tests, format)
- Acceptance criteria verification before commit
- Exact commit message template with heredoc format
- Edge case handling (validation loops, blocked stories, dirty state, large stories)
- File format specs for prd.json schema and progress.txt structure
- Critical fix: "context is stale — re-read from disk" for fresh_context loops
Also improved bash setup node (dep install, structured output delimiters,
story counts) and report node (git log/diff stats, PR status check).
* feat(workflows)!: remove standalone loop workflow type
BREAKING: Standalone `loop:` workflows are no longer supported.
Loop iteration is now exclusively a DAG node type (LoopNode).
Existing loop workflows should be migrated to DAG workflows
with loop nodes — see archon-ralph-dag.yaml for the pattern.
Removed:
- LoopConfig type and LoopWorkflow from WorkflowDefinition union
- executeLoopWorkflow function (~600 lines) from executor.ts
- Loop dispatch in executeWorkflow
- Top-level loop: parsing in loader (now returns clear error message)
- archon-ralph-fresh.yaml, archon-ralph-stateful.yaml, archon-test-loop.yaml
- LoopEditor.tsx and loop mode from WorkflowBuilder UI
- ~900 lines of standalone loop tests
Kept (for DAG loop nodes):
- LoopNodeConfig, LoopNode, isLoopNode
- executeLoopNode in dag-executor.ts
- Loop iteration events in store/event-emitter
- isLoop tracking in web UI workflow store (fires for DAG loop nodes)
* fix: address all review findings for loop-dag-node PR
- Fix missing isDagWorkflow import in command-handler.ts (shipping bug)
- Wrap substituteWorkflowVariables and getAssistantClient in try-catch
with structured error output in executeLoopNode
- Add onTimeout callback for idle timeout (log + user notification + abort)
- Add cancellation user notification before returning failed state
- Differentiate until_bash ENOENT/system errors from expected non-zero exit
- Use logDir for per-iteration AI output logging (logAssistant, logTool,
logStepComplete, tool_called/tool_completed events, sendStructuredEvent)
- Reject retry: on loop nodes at load time (executor doesn't apply it)
- Remove dead isLoop field from WorkflowStartedEvent
- Fix stale error message "DAG/loop dispatch" -> "DAG dispatch"
- Fix stale commitWorkflowArtifacts doc referencing "loop-based"
- Fix archon-ralph-dag.yaml referencing deleted workflows
- Update CLAUDE.md: "Two execution modes", add loop node to DAG description
- Extract parseIdleTimeout helper (3 copies -> 1 in loader.ts)
- Use isLoopNode() type guard in validateDagStructure
- Simplify buildLoopNodeOptions with conditional spread
- Restore loop?: never on StepWorkflow for type safety
- Add tests: AI error mid-iteration, plain signal detection, false positive
- Fix stale test assertion for standalone loop rejection message
* feat: refactor Gitea adapter to community forge structure + tea CLI
Moves the Gitea platform adapter from the old location
(packages/server/src/adapters/gitea.ts) to the proper community
forge adapter structure:
packages/adapters/src/community/forge/gitea/
├── adapter.ts # Main adapter class
├── auth.ts # parseAllowedUsers, isGiteaUserAuthorized
├── types.ts # WebhookEvent interface
├── index.ts # Barrel export
└── adapter.test.ts # 43 passing tests
Key changes:
- Fix imports: createLogger, getArchonWorkspacesPath,
getCommandFolderSearchPaths now from @archon/paths
- Fix imports: cloneRepository, syncRepository, addSafeDirectory,
toRepoPath, toBranchName, isWorktreePath now from @archon/git
- Remove execAsync / child_process / promisify — use @archon/git
functions for all git operations
- auth.ts extracted from @archon/core into adapter package (mirrors
GitHub adapter's auth.ts pattern)
- types.ts extracted: WebhookEvent interface now standalone
- Replace gh CLI hints with tea CLI in context strings:
'tea issue view N' and 'tea pr view N'
- Register GiteaAdapter in packages/server/src/index.ts via
@archon/adapters/community/forge/gitea import
- Document GITEA_* env vars in .env.example
Tests: 43 pass, 0 fail
Co-authored-by: John Fitzpatrick <john@cyberfitz.org>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Archon <archon@dynamous.ai>
Co-authored-by: Thomas <info@smartcode.diy>
Co-authored-by: Rasmus Widing <152263317+Wirasm@users.noreply.github.com>
Co-authored-by: Fitzy <fitzy@cyberfitz.org>
Co-authored-by: John Fitzpatrick <john@cyberfitz.org>
|
||
|
Rasmus Widing
|
3338e80ff9 |
fix: Fix Dockerfile for monorepo workspace structure
- Copy all workspace package.json files before bun install (lockfile depends on all workspace definitions) - Fix build path to packages/server/src/index.ts (not src/) |
||
|
Rasmus Widing
|
68e7db0466
|
feat: Phase 5 - CLI binary distribution (#325)
* docs: Add Phase 5 CLI binary distribution plan - Create detailed implementation plan for binary distribution - Add Phase 5.0: Bundle defaults for binary (depends on #322) - Add Phase 5.1-5.7: Build scripts, GitHub Actions, curl install, Homebrew formula, Windows docs, version command, release guide - Update research doc with Phase 6 (auto-update command) - Renumber dashboard to Phase 7, workflow builder to Phase 8 - Mark Phases 1-4 as complete in research doc * feat: Phase 5 - CLI binary distribution Implement standalone binary distribution for Archon CLI: - Bundle default commands and workflows into binaries at compile time - Add build scripts for cross-platform compilation (macOS/Linux, ARM64/x64) - Create GitHub Actions release workflow triggered on version tags - Add curl install script with checksum verification - Create Homebrew formula for macOS/Linux installation - Update version command to show platform, build type, and database info - Add developer release guide documentation - Update README with CLI installation instructions Binary compilation uses Bun's --compile flag to create standalone executables that include the Bun runtime and all dependencies. Default workflows and commands are imported as text at compile time and embedded directly into the binary. * fix: Pin Dockerfile to Bun 1.3.4 to match lockfile version The Docker build was failing because oven/bun:1-slim resolved to 1.3.6 while the lockfile was created with 1.3.4, causing --frozen-lockfile to fail. * docs: Clarify binary vs source builds for default commands/workflows * fix: Address PR review issues for CLI binary distribution Security fixes: - install.sh: Require SKIP_CHECKSUM=true to bypass checksum verification instead of silently skipping (addresses security vulnerability) - install.sh: Show actual error output when version check fails instead of falsely reporting success Validation improvements: - checksums.sh: Validate all 4 expected binaries exist before generating checksums to prevent releasing incomplete builds - build-binaries.sh: Verify binary exists and has reasonable size (>1MB) after each build step - update-homebrew.sh: Validate extracted checksums are non-empty and look like valid SHA256 hashes (64 hex chars) - update-homebrew.sh: Fix sed patterns to use URL context for updating checksums on subsequent runs Bug fixes: - homebrew/archon.rb: Fix test to expect exit code 0 (success) instead of 1 for `archon version` - loader.ts: Log error when bundled workflow fails to parse (indicates build-time corruption) Test coverage: - Add bundled-defaults.test.ts for isBinaryBuild() and content validation - Add connection.test.ts for getDatabaseType() function - Add binary build bundled workflow tests to loader.test.ts - Add binary build bundled command tests to executor.test.ts All 959 tests pass. |
||
|
Rasmus Widing
|
3026a6445d
|
Add Archon distribution config and directory structure (#101)
* Add Archon distribution config and directory structure
- Create centralized path resolution in src/utils/archon-paths.ts
- Add YAML configuration system (src/config/) with layered loading
- Update Dockerfile and docker-compose for /.archon/ directory
- Add GHCR publish workflow for multi-arch Docker builds
- Create deploy/ directory with end-user docker-compose
- Add /init command to create .archon structure in repos
- Add docs/configuration.md reference guide
- Update README with Quick Start section
- Add bun run validate script
- Update tests for new path defaults (~/.archon/)
Directory structure:
- Local: ~/.archon/{workspaces,worktrees,config.yaml}
- Docker: /.archon/{workspaces,worktrees}
- Repo: .archon/{commands,workflows,config.yaml}
Legacy WORKSPACE_PATH and WORKTREE_BASE env vars still supported.
* Complete Archon distribution config implementation
- Wire up config system in src/index.ts (Task 3.5)
- Remove legacy WORKSPACE_PATH and WORKTREE_BASE support
- Add logConfig() function to config-loader.ts
- Update docker-compose.yml to use ARCHON_DOCKER env var
- Remove legacy env vars from .env.example
- Update all documentation to reference ARCHON_HOME
- Create scripts/validate-setup.sh for setup validation
- Add setup:check script to package.json
- Create docs/getting-started.md guide
- Create docs/archon-architecture.md technical docs
- Update tests to use ARCHON_HOME instead of legacy vars
- Fix validate.md command template for new paths
All plan phases now complete:
- Phase 1: Archon Directory Structure
- Phase 2: Docker Distribution
- Phase 3: YAML Configuration System
- Phase 4: Developer Experience
- Phase 5: Documentation
|
||
|
Rasmus Widing
|
c85622fbbb
|
Migrate from Node.js/npm/Jest to Bun runtime (#85)
* Migrate from Node.js/npm/Jest to Bun runtime - Replace npm with bun for package management (bun.lock) - Replace Jest with bun:test for testing - Update tsconfig for Bun (ESNext module, bundler resolution) - Update Dockerfile to use oven/bun:1-slim - Update CI workflow to use oven-sh/setup-bun@v2 - Remove dynamic import hack from codex.ts (direct ESM imports) - Fix test mocking for Bun (export execFileAsync, use spyOn) - Update all documentation (CLAUDE.md, README.md, CONTRIBUTING.md) All 395 tests pass, type-check passes, E2E validated with curl. * ci: retrigger CI build * fix: make execFileAsync a function for better Bun mockability * fix: ensure execFileAsync returns string not Buffer * fix: rename _execFileAsync to comply with naming convention * fix: make mkdirAsync mockable for Bun tests * fix: update engines to bun>=1.0.0 and add mkdirAsync mock * fix: pin Bun to 1.3.4 in CI to fix mock.module test failures Newer Bun versions have different mock.module() behavior that causes cross-test module pollution, resulting in 71 test failures in CI while tests pass locally. Pinning to 1.3.4 ensures consistent behavior. * fix: run orchestrator tests last to avoid mock.module pollution Bun's mock.module() pollutes the global module cache, causing tests to fail when orchestrator.test.ts (which mocks command-handler and factory) runs before those modules' own test files. Fix by running tests in two batches: 1. All tests except orchestrator 2. Orchestrator tests last This ensures orchestrator's mocks don't affect other test files. |
||
|
Cole Medin
|
c136c0e598 | Fixing up Dockerfile | ||
|
Cole Medin
|
ca74774cac | Codex Implementation | ||
|
Cole Medin
|
3bf65f617a | Optimizing permissions and code environment | ||
|
Cole Medin
|
41415d7e7f |
feat: implement telegram + claude mvp with generic architecture
- Add generic IPlatformAdapter and IAssistantClient interfaces for extensibility - Implement TelegramAdapter with streaming/batch modes - Implement ClaudeClient with session persistence and resume capability - Create TestAdapter for autonomous validation via HTTP endpoints - Add PostgreSQL database with 3-table schema (conversations, codebases, sessions) - Implement slash command system (/clone, /status, /getcwd, /setcwd, /reset, /help) - Add Docker containerization with docker-compose (with-db profile for local PostgreSQL) - Fix Claude Agent SDK spawn error (install bash, pass PATH environment variable) - Fix workspace volume mount to use /workspace in container - Add comprehensive documentation and health check endpoints Architecture highlights: - Platform-agnostic design allows adding Slack, GitHub, etc. via IPlatformAdapter - AI-agnostic design allows adding Codex, etc. via IAssistantClient - Orchestrator uses dependency injection with interface types - Session persistence survives container restarts - Working directory + codebase context determine Claude behavior 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> |