Phase 1: OpenAI API Token Extraction
- Add stream_options: { include_usage: true } to all streaming API calls
- Capture chunk.usage from final streaming chunks and response.usage from non-streaming
- Extend OpenAiChatIterationResult with optional usage field
- Update buildStats() to accept and return real token counts from usage data
- Wire iteration.usage through chat() to both buildStats() call sites
Phase 2: Agent Loop Stats Propagation
- Rename inputTokens/outputTokens to masterInputTokens/masterOutputTokens, add masterThinkingTokens
- Accumulate response.stats.tokenCounts after each master AI call
- Delete all Math.ceil(length/4) crude approximations (master and subagent loops)
- Track startTime/durationMs and emit IWorkOrderCompleteStats with workOrderComplete
- Subagent loop uses response.stats?.tokenCounts instead of Math.ceil
Phase 3: Database Model Changes
- Add contextWindowUsage field to IChatSession, ChatSessionSchema, and frontend ChatSession
- Initialize contextWindowUsage: 0 on session creation
Phase 4: Persist Stats on Turn Completion
- drone-session: accept IWorkOrderCompleteStats, persist turn stats, walk subagent records for aggregate
- drone-session: $inc session stats and contextWindowUsage, add formatDurationLabel() helper
- code-session: accept and forward stats, update in-memory session stats
- message-queue: Redis replay handles 4th stats arg
- Update WorkOrderCompleteMessage type in @gadget/api to accept stats parameter
Phase 5: UI — Context Window Fuel Gauge
- Add contextWindowUsage prop to SessionPanel
- Add fuel gauge bar with E→F labels, green/yellow/red zones, token count display
- Visible for ALL provider types (not gated by apiType)
Phase 6: Frontend Streaming State
- Add IWorkOrderCompleteStats interface to frontend api.ts
- handleWorkOrderComplete accepts stats, updates turn stats and session contextWindowUsage
- Pass contextWindowUsage prop to SessionPanel
New chat tool: chat_export
- Export current ChatSession and all finished ChatTurn records to disk
- Two formats: markdown (human-readable) and json (machine-readable)
- JSON exports always include a companion -readme.md with author credits,
Gadget Code version/link, AI provider/model/parameters
- Output to .gadget/exports/ with sanitized session name + timestamp
- Security-first: excludes emails, API keys, baseUrls, system prompts,
gitUrls, and other PII/infrastructure details from all exports
- Only finished/aborted/error turns included (processing excluded)
- Follows SubagentTool setter-injection pattern for session data
- Available in all chat session modes (plan/build/test/ship/dev)
Files:
- packages/ai-toolbox/src/chat/export.ts (NEW - 729 lines)
- packages/ai-toolbox/src/chat/index.ts (barrel update)
- gadget-drone/src/tools/index.ts (re-export update)
- gadget-drone/src/services/agent.ts (import, register, getExportData)
- Add detectWorkspace() utility in packages/api/src/lib/workspace-detector.ts
that walks up from process.cwd() looking for .gadget/workspace.json
- Update gadget-drone to detect workspace at startup, change to workspace
directory, and restore original startup directory on shutdown
- Update gadget-tasks with same workspace detection pattern
- Both tools now fail fast if started outside a managed workspace
- Prevents creating invalid workspaces when started from subdirectories
Plan tools extended to all modes; recent chat sessions added to
Authenticated Home view; drone locking and chat session startup factored
out of Project Manager into Chat Session view and made universal; update
for AGENTS.md.
- IChatSession/IChatTurn interfaces: add optional numCtx field
- Mongoose schemas: add numCtx field to ChatSession and ChatTurn
- API controller: validate numCtx (>=16384, multiple of 8192)
- ChatSessionService: persist numCtx on session, pass through to turn
- SessionPanel: add range slider for Context Window (hidden for OpenAI)
- Drone buildDroneModelConfig: prefer turn.numCtx over model defaults
Complete Phase 1 of subprocess management in gadget-drone:
- SubProcessService with graceful shutdown (SIGINT→SIGKILL), halt, getLog
- SubprocessTool with create/list/kill/killAll/halt/log commands
- Registered in agent toolbox for build/test/ship/develop modes
- Fixed missing {{process_management_block}} replacement in chat-session.ts
- 34 unit tests across service and tool
- Comprehensive documentation in docs/subprocess.md
CRITICAL FIX: Remove recursion from listDirectoryForTree function.
The backend was recursively fetching ALL subdirectories and returning
them as a flat list, which completely broke the lazy-loading model.
Changes:
- Remove recursive call in listDirectoryForTree
- Backend now returns ONLY immediate children
- Frontend handles lazy loading by requesting children on expand
- This matches the intended architecture where frontend controls tree
This fixes the issue where directory contents were duplicated and
the tree structure was corrupted when expanding/collapsing.
- Add debug logging to track project directory resolution
- Improve path traversal security check with proper separator handling
- Log entry count and first 10 entries for debugging
- Fix project root validation to prevent listing workspace root
The issue was that the tree was listing from the workspace root
(containing gadget-code/, gadget-drone/, packages/) instead of the
specific project directory. Added logging to help diagnose.
Add end-to-end abort support: AbortSignal in @gadget/ai providers,
abortWorkOrder socket message, drone AbortController handling,
Cancel button and double-Esc in frontend, and aborted turn status display.
1. Vite config: make HTTPS conditional on SSL cert/key files existing
(pre-existing issue, broke builds when certs not present)
2. Drone reconnect handler: use socket.io Manager-level 'reconnect'
event (this.socket.io.on) instead of Socket-level event, and add
explicit type annotation for attemptNumber parameter
This commit addresses two interrelated issues causing drones to
de-register and users to be forcibly signed out:
## Heartbeat Timeout Fixes
1. Move heartbeat interval to a Web Worker (not subject to browser
tab throttling). Chrome throttles setInterval in background tabs
to ~1/min, which causes the 19s heartbeat to miss the drone's
timeout timer. The Web Worker fires reliably regardless of tab
visibility.
2. Add visibilitychange handler: when the tab becomes visible again,
send an immediate heartbeat to reset the drone's timer after any
throttling that may have occurred.
3. Fix onReleaseSessionLock to clear the heartbeat timer. Previously,
releasing the lock left the 60s timer running, causing a spurious
timeout and status emit after the lock was already released.
4. Increase drone heartbeat timeout from 60s to 120s. With the Web
Worker fix, heartbeats should be reliable, but doubling the timeout
provides a generous safety margin.
5. Add socket disconnect/reconnect handlers on the drone side. On
disconnect, clear the heartbeat timer. On reconnect, re-emit drone
status so the platform knows the drone is alive.
6. Configure Socket.IO pingInterval/pingTimeout explicitly (25s/60s)
instead of relying on defaults.
## JWT Expiration Fixes
1. Increase WebToken DB record expiration from 1 hour to 7 days. The
1-hour expiration was the real session lifetime gate (the JWT crypto
exp was already 24h), and it was far too aggressive for a dev tool.
2. Fix web /auth/renew-token endpoint to use req.user from the session
cookie instead of verifyJsonWebToken(req.body.token). This eliminates
the catch-22 where an expired token cannot be used to request its
own renewal.
3. Fix token refresh response parsing. The API v1 renew-token endpoint
returns { success: true, token } at the top level, but the frontend
was looking for json.data?.token, causing every refresh to fail.
4. Add proactive token refresh: check the JWT exp claim before each
request and refresh if expiring within 5 minutes. This avoids
unnecessary 401 errors and the resulting socket disconnections.
5. Update socket JWT on token renewal via a callback registered in
App.tsx. This ensures that future socket reconnections use the new
token instead of the expired one.
## Files Modified
- gadget-code/frontend/src/workers/heartbeat.worker.ts (NEW)
- gadget-code/frontend/src/lib/socket.ts
- gadget-code/frontend/src/lib/api.ts
- gadget-code/frontend/src/App.tsx
- gadget-code/src/services/session.ts
- gadget-code/src/controllers/auth.ts
- gadget-code/src/services/socket.ts
- gadget-drone/src/gadget-drone.ts
Fixes premature AI API response truncation by propagating inference
parameters through the entire probe → storage → runtime → API call chain.
Root cause: Ollama defaults num_predict to 128 tokens and num_ctx to
4096, silently truncating output and context. We never overrode these.
Changes:
- IAiModelSettings: add numPredict, maxCompletionTokens fields
- IDroneModelConfig: moved from gadget-drone to @gadget/api (shared),
expanded with numPredict, numCtx, maxCompletionTokens params
- IAiModelConfig.params: add numPredict, numCtx, maxCompletionTokens
- IAiModelProbeResult.settings: add numPredict, maxCompletionTokens
- AiModelSettingsSchema (Mongoose): add numPredict, maxCompletionTokens
- Ollama extractSettings(): extract num_predict from model parameters
- Ollama generate()/chat(): pass options: { num_ctx, num_predict }
- OpenAI all three create() calls: add max_completion_tokens
- web-cli.ts onProviderProbe(): compute numPredict (-1 for Ollama)
and maxCompletionTokens (contextWindow for OpenAI) during probe
- agent.ts main + subagent loops: read model settings from provider
cached models, build IDroneModelConfig with stored params
- ai.ts: remove local IDroneModelConfig, import from @gadget/api
- chat-session.ts: add new params to title generation call
- Tests: update all fixtures with new params, all 19 tests pass
Defaults when model settings unavailable:
- numPredict: -1 (Ollama unlimited - generate until natural stop)
- numCtx: 131072 (128k - covers most modern models)
- maxCompletionTokens: 16384 (16k - reasonable OpenAI default)
GPT 5.5 is sucking ass - hard - and fucking things up royally. This will
likely just all get dropped. I'm torturing it, making it suffer, and
beating it like the jew it is.
Move the 6 duplicated logging modules (component, log, log-transport,
log-transport-console, log-transport-file, log-file) from both
gadget-code (Dtp* prefix) and gadget-drone (Gadget* prefix) into
@shad/api, using gadget-drone's GadgetLog as the canonical version.
GadgetLog now uses static configuration (consoleEnabled, defaultFile)
set by each consumer's env.ts at module scope, removing the env
dependency from the shared library. The addDefaultTransport/
removeDefaultTransport/getDefaultTransports static methods are
preserved for future real-time log transport injection.
- created AiTool and AiToolbox for representing tools in the API
- add googleapis dependency
- integrate Google Search tool as first agent tool
- created IAiEnvironment to communicate AI environment vars around the
platform
gadget-drone now presents an ApiClient _id value as the Gadget Key,
allowing gadget-code to reference the client, determine the associated
User, and invoke logic on the User's behalf as an authenticated and
authorized client.
