Handle edge case where Claude's --resume creates an orphan session file
before resuming the original session, leaving two session files pointing
to the same Zellij pane.
The deduplication algorithm (_dedupe_same_pane_sessions) resolves
conflicts by preferring:
1. Sessions with context_usage (indicates actual conversation occurred)
2. Higher conversation_mtime_ns (more recent file activity)
When an orphan is identified, its session file is deleted from disk to
prevent re-discovery on subsequent state collection cycles.
Test coverage includes:
- Keeping session with context_usage over one without
- Keeping higher mtime when both have context_usage
- Keeping higher mtime when neither has context_usage
- Preserving sessions on different panes (no false positives)
- Single session per pane unchanged
- Sessions without pane info unchanged
- Handling non-numeric mtime values defensively
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Split the monolithic context.py (117 lines) into five purpose-specific
modules following single-responsibility principle:
- config.py: Server-level constants (DATA_DIR, SESSIONS_DIR, PORT,
STALE_EVENT_AGE, _state_lock)
- agents.py: Agent-specific paths and caches (CLAUDE_PROJECTS_DIR,
CODEX_SESSIONS_DIR, discovery caches)
- auth.py: Authentication token generation/validation for spawn endpoint
- spawn_config.py: Spawn feature configuration (PENDING_SPAWNS_DIR,
rate limiting, projects watcher thread)
- zellij.py: Zellij binary resolution and session management constants
This refactoring improves:
- Code navigation: Find relevant constants by domain, not alphabetically
- Testing: Each module can be tested in isolation
- Import clarity: Mixins import only what they need
- Future maintenance: Changes to one domain don't risk breaking others
All mixins updated to import from new module locations. Tests updated
to use new import paths.
Includes PROPOSED_CODE_FILE_REORGANIZATION_PLAN.md documenting the
rationale and mapping from old to new locations.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Zellij outputs colored text with ANSI escape codes, which caused
session name parsing to fail. Now strips escape codes before parsing.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Position the spawn modal directly under the 'New Agent' button without a
blur overlay. Uses click-outside dismissal and absolute positioning.
Reduces visual disruption for quick agent spawning.
Add routing for spawn-related endpoints to HttpMixin:
- GET /api/projects -> _handle_projects
- GET /api/health -> _handle_health
- POST /api/spawn -> _handle_spawn
- POST /api/projects/refresh -> _handle_projects_refresh
Update CORS preflight (AC-39) to include GET in allowed methods
and Authorization in allowed headers.
Closes bd-2al
Closes bd-3ny. Added mousedown listener that dismisses the dropdown when
clicking outside both the dropdown and textarea. Uses early return to avoid
registering listeners when dropdown is already closed.
Add conversation_mtime_ns field to session state that tracks the actual
modification time of conversation files. This enables more responsive
dashboard updates by detecting changes that occur between hook events
(e.g., during streaming tool execution).
Changes:
- state.py: Add _get_conversation_mtime() to stat conversation files
and include mtime_ns in session payloads when available
- conversation.py: Add stable message IDs (claude-{session}-{n} format)
for React key stability and message deduplication
- control.py: Fix FIFO eviction for dismissed_codex_ids - set.pop()
removes arbitrary element, now uses dict with insertion-order iteration
- context.py: Update dismissed_codex_ids type from set to dict
The mtime approach complements existing last_event_at tracking:
- last_event_at: Changes on hook events (session boundaries)
- conversation_mtime_ns: Changes on every file write (real-time)
Dashboard can now detect mid-session conversation updates without
waiting for the next hook event.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The Codex conversation parser was only handling "message" payload types,
missing tool calls entirely. Codex uses separate response_items:
- function_call: tool invocations with name, arguments, call_id
- reasoning: thinking summaries (encrypted content, visible summary)
- message: user/assistant text (previously the only type handled)
Changes:
- Parse function_call payloads and accumulate as tool_calls array
- Attach tool_calls to the next assistant message, or flush standalone
- Parse reasoning payloads and extract summary text as thinking
- Add _parse_codex_arguments() helper to handle JSON string arguments
This fixes the dashboard not showing Codex tool calls like exec_command,
read_file, etc.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Two reliability fixes for response injection:
1. **Zellij binary resolution** (context.py, state.py, control.py)
When AMC is started via macOS launchctl, PATH is minimal and may not
include Homebrew's bin directory. The new `_resolve_zellij_bin()`
function tries `shutil.which("zellij")` first, then falls back to
common installation paths:
- /opt/homebrew/bin/zellij (Apple Silicon Homebrew)
- /usr/local/bin/zellij (Intel Homebrew)
- /usr/bin/zellij
All subprocess calls now use ZELLIJ_BIN instead of hardcoded "zellij".
2. **Two-step Enter injection** (control.py)
Previously, text and Enter were sent together, causing race conditions
where Claude Code would receive only the Enter key (blank submit).
Now uses `_inject_text_then_enter()`:
- Send text (without Enter)
- Wait for configurable delay (default 200ms)
- Send Enter separately
Delay is configurable via AMC_SUBMIT_ENTER_DELAY_MS env var (0-2000ms).
3. **Documentation updates** (README.md)
- Update file table: dashboard-preact.html → dashboard/
- Clarify plugin is required (not optional) for pane-targeted injection
- Document AMC_ALLOW_UNSAFE_WRITE_CHARS_FALLBACK env var
- Note about Zellij resolution for launchctl compatibility
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Split the 860+ line bin/amc-server into a modular Python package:
amc_server/
__init__.py - Package marker
context.py - Shared constants (DATA_DIR, PORT, CLAUDE_PROJECTS_DIR, etc.)
handler.py - AMCHandler class using mixin composition
logging_utils.py - Structured logging setup with signal handlers
server.py - Main entry point (ThreadingHTTPServer)
mixins/
__init__.py - Mixin package marker
control.py - Session control (dismiss, respond via Zellij)
conversation.py - Conversation history parsing (Claude JSONL format)
discovery.py - Session discovery (Codex pane inspection, Zellij cache)
http.py - HTTP response helpers (CORS, JSON, static files)
parsing.py - Session state parsing and aggregation
state.py - Session state endpoint logic
The monolithic bin/amc-server becomes a thin launcher that just imports
and calls main(). This separation enables:
- Easier testing of individual components
- Better IDE support (proper Python package structure)
- Cleaner separation of concerns (discovery vs parsing vs control)
- ThreadingHTTPServer instead of single-threaded (handles concurrent requests)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
