)]}'
{
  "log": [
    {
      "commit": "b8a8f35348326b8daf69e31efc77d37fac4bd276",
      "tree": "fa7a59c5951dbd3ecb5e6bf840ac7b45167a9e74",
      "parents": [
        "b5739403b1b6ec6fac909b258bd47ce5a338940e"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Mon Jun 02 07:39:37 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Mon Jun 02 17:01:19 2025 -0700"
      },
      "message": "loop: implement comprehensive conversation compaction system\n\n\"comprehensive\" is over-stating it. Currently, users get\nthe dreaded:\n\n\terror: failed to continue conversation: status 400 Bad Request:\n\t{\"type\":\"error\",\"error\":{\"type\":\"invalid_request_error\",\"message\":\"input\n\tlength and max_tokens exceed context limit: 197257 + 8192 \u003e 200000,\n\tdecrease input length or max_tokens and try again\"}}\n\nThat\u0027s... annoying. Instead, let\u0027s compact automatically. I was going to\nstart with adding a /compact command or button, but it turns out that\nteasing that through the system is annoying, because the agent state\nmachine is intended to be somewhat single-threaded, and what do you do\nwhen a /compact comes in while other things are going on. It\u0027s possible,\nbut it was genuinely easier to prompt my way into doing it\nautomatically.\n\nI originally set the threshold to 75%, but given that 8192/200000 is 4%,\nI just changed it to 94%.\n\nWe\u0027ll see how well it works!\n\n~~~~\n\nImplement automatic conversation compaction to manage token limits and prevent\ncontext overflow, with enhanced UX feedback and accurate token tracking.\n\nProblem Analysis:\nLarge conversations could exceed model context limits, causing failures\nwhen total tokens approached or exceeded the maximum context window.\nWithout automatic management, users would experience unexpected errors\nand conversation interruptions in long sessions.\n\nImplementation:\n\n1. Automatic Compaction Infrastructure:\n   - Added ShouldCompact() method to detect when compaction is needed\n   - Configurable token thresholds for different compaction triggers\n   - Integration with existing loop state machine for seamless operation\n\n2. Accurate Token Counting:\n   - Enhanced context size estimation using actual token usage from LLM responses\n   - Track real token consumption rather than relying on estimates\n   - Account for tool calls, system prompts, and conversation history\n\n3. Compaction Logic and Timing:\n   - Triggered at 75% of context limit (configurable threshold)\n   - Preserves recent conversation context while compacting older messages\n   - Maintains conversation continuity and coherence\n\n4. Enhanced User Experience:\n   - Visual indicators in webui when compaction occurs\n   - Token count display showing current usage vs limits\n   - Clear messaging about compaction status and reasoning\n   - Timeline updates to reflect compacted conversation state\n\n5. UI Component Updates:\n   - sketch-timeline.ts: Added compaction status display\n   - sketch-timeline-message.ts: Enhanced message rendering for compacted state\n   - sketch-app-shell.ts: Token count integration and status updates\n\nTechnical Details:\n- Thread-safe implementation with proper mutex usage\n- Preserves conversation metadata and essential context\n- Configurable compaction strategies for different use cases\n- Comprehensive error handling and fallback behavior\n- Integration with existing LLM provider implementations (Claude, OpenAI, Gemini)\n\nTesting:\n- Added unit tests for ShouldCompact logic with various scenarios\n- Verified compaction triggers at correct token thresholds\n- Confirmed UI updates reflect compaction status accurately\n- All existing tests continue to pass without regression\n\nBenefits:\n- Prevents context overflow errors in long conversations\n- Maintains conversation quality while managing resource limits\n- Provides clear user feedback about system behavior\n- Enables unlimited conversation length with automatic management\n- Improves overall system reliability and user experience\n\nThis system ensures sketch can handle conversations of any length while\nmaintaining performance and providing transparent feedback to users about\ntoken usage and compaction activities.\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\nChange-ID: s28a53f4e442aa169k\n"
    },
    {
      "commit": "f28729932fdf9ecc67d3fabbfca297178a323a14",
      "tree": "5ae2eca28c1c139a95dde0648a17eeaba3e60bce",
      "parents": [
        "716bfee93847e19465723af2d7c553d23c9984df"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 22 10:35:28 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 22 11:28:35 2025 -0700"
      },
      "message": "Refactor agent git state into its own struct to tease apart its locking a bit.\n\nI want to invoke calling the git state when editing files, and that\nrequires separating it somewhat from the agent\u0027s messy and coarse\nlocking.\n"
    },
    {
      "commit": "bc8c8dc5bc7abca6fa523c5ca45c9fd2c09384c2",
      "tree": "d59b7b3ef6965949376481cfa6108fee7312f6fd",
      "parents": [
        "1a648f34a53216a9ed88792d583fad50246f3d4b"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Wed May 21 13:19:13 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Wed May 21 13:40:03 2025 -0700"
      },
      "message": "sketch main: migrating things from /init into cmdline flags\n\nAs much as possible, I want NewAgent() to take as many arguments as it\ncan, and try to avoid doing different things with Init().\n\nYou can\u0027t quite do everything because the port that Docker has\nopen for forwarding starts as \"0\" and we call \"docker port\" to find it,\nbut besides that, almost everything is knowable up front.\n"
    },
    {
      "commit": "64f2aa8db137ee801120fe38b19f60103a2326dd",
      "tree": "b322e422ac17af0e9ed0e59ea9f6d91d40291f5c",
      "parents": [
        "b81d7d476ff2d104d34d3a637f2bd826a2a89eaf"
      ],
      "author": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Wed May 14 18:31:05 2025 +0000"
      },
      "committer": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Wed May 14 13:53:29 2025 -0700"
      },
      "message": "loop: make multiplechoice tool calls end the turn\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\nChange-ID: s8d507faf9c095824sk\n"
    },
    {
      "commit": "72252cbcb97840d724133be67c4c69cc69ebb2d3",
      "tree": "a361499dc3fa6b9af2be3e74cfd59fd8ba34690e",
      "parents": [
        "7ce5fb76d8748ebf73c5adf9d6cd8eb67716fba8"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Sat May 10 17:00:08 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Sat May 10 17:00:08 2025 -0700"
      },
      "message": "llm and everything: Update ToolResult to use []Content instead of string for multimodal support\n\nThis was a journey. The sketch-generated summary below is acceptable,\nbut I want to tell you about it in my voice too. The goal was to send\nscreenshots to Claude, so that it could... look at them. Currently\nthe take screenshot and read screenshot tools are different, and they\u0027ll\nneed to be renamed/prompt-engineered a bit, but that\u0027s all fine.\n\nThe miserable part was that we had to change the return value\nof tool from string to Content[], and this crosses several layers:\n - llm.Tool\n - llm.Content\n - ant.Content \u0026 openai and gemini friends\n - AgentMessage [we left this alone]\n\nExtra fun is that Claude\u0027s API for sending images has nested Content\nfields, and empty string and missing needs to be distinguished for the\nText field (because lots of shell commands return the empty string!).\n\nFor the UI, I made us transform the results into a string, dropping\nimages. This would have been yet more churn for not much obvious\nbenefit. Plus, it was going to break skaband\u0027s compatibility, and ...\nyet more work.\n\nOpenAI and Gemini don\u0027t obviously support images in this same way,\nso they just don\u0027t get the tools.\n\n~~~~~~~~~~ Sketch said:\n\nThis architectural change transforms tool results from plain strings to []Content arrays, enabling multimodal interaction in the system. Key changes include:\n\n- Core structural changes:\n  - Modified ToolResult type from string to []Content across all packages\n  - Added MediaType field to Content struct for MIME type support\n  - Created TextContent and ImageContent helper functions\n  - Updated all tool.Run implementations to return []Content\n\n- Image handling:\n  - Implemented base64 image support in Anthropic adapter\n  - Added proper media type detection and content formatting\n  - Created browser_read_image tool for displaying screenshots\n  - Updated browser_screenshot to provide usable image paths\n\n- Adapter improvements:\n  - Updated all LLM adapters (ANT, OAI, GEM) to handle content arrays\n  - Added specialized image content handling in the Anthropic adapter\n  - Ensured proper JSON serialization/deserialization for all content types\n  - Improved test coverage for content arrays\n\n- UI enhancements:\n  - Added omitempty tags to reduce JSON response size\n  - Updated TypeScript types to handle array content\n  - Made field naming consistent (tool_error vs is_error)\n  - Preserved backward compatibility for existing consumers\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\nChange-ID: s1a2b3c4d5e6f7g8h\n"
    },
    {
      "commit": "d2f54c2e43ad3ad2c9b389068c1555c3e7f231f2",
      "tree": "a57204539b914c7109b5a53456148970b1293357",
      "parents": [
        "f392251fe06355e73d80eae868916410ab18b08b"
      ],
      "author": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Wed May 07 18:38:07 2025 -0700"
      },
      "committer": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Wed May 07 19:17:50 2025 -0700"
      },
      "message": "loop: improve update_tests.sh reliability\n\nEncouraging the agent to call the title tool immediately\nhas its drawbacks!"
    },
    {
      "commit": "3659d8714d31bad050d36043b67efd76a8d53b77",
      "tree": "2803c54c63628e397e63bdbbe4fa2647a4d5f556",
      "parents": [
        "021557a7d33dacd4fedb9a4677fc93c48569d57a"
      ],
      "author": {
        "name": "David Crawshaw",
        "email": "david@zentus.com",
        "time": "Mon May 05 17:52:23 2025 -0700"
      },
      "committer": {
        "name": "David Crawshaw",
        "email": "david@zentus.com",
        "time": "Mon May 05 17:52:27 2025 -0700"
      },
      "message": "all: more gemini key plumbing\n"
    },
    {
      "commit": "4f84ab729ddbf428b0e891940f08f70b4edee05c",
      "tree": "f2e52e4a01c188ada1f5acf8b2a013029b999495",
      "parents": [
        "44f9b4cec11e269a52fbfc099989ab425b8e125f"
      ],
      "author": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Tue Apr 22 16:40:54 2025 -0700"
      },
      "committer": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Fri May 02 12:57:44 2025 -0700"
      },
      "message": "all: support openai-compatible models\n\nThe support is rather minimal at this point:\nOnly hard-coded models, only -unsafe, only -skabandaddr\u003d\"\".\n\nThe \"shared\" LLM package is strongly Claude-flavored.\n\nWe can fix all of this and more over time, if we are inspired to.\n(Maybe we\u0027ll switch to https://github.com/maruel/genai?)\n\nThe goal for now is to get the rough structure in place.\nI\u0027ve rebased and rebuilt this more times than I care to remember.\n"
    },
    {
      "commit": "4d5e9978fce2feb262663b9d0455c06546723ddc",
      "tree": "dfa4824a78b33a01c5c0d72d4077cfec78010a62",
      "parents": [
        "2772f63c725cdc47f644f68a9e684e798e54b543"
      ],
      "author": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Thu May 01 15:56:37 2025 -0700"
      },
      "committer": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Thu May 01 16:09:09 2025 -0700"
      },
      "message": "all: make tests accept OUTER_SKETCH_ANTHROPIC_API_KEY\n\nInner sketch already has an ANTHROPIC_API_KEY...but it is the\nsketch anthropic key, not the anthropic anthropic key. Blarg!\n\n"
    },
    {
      "commit": "a3dcd86d1c05e95ee9c058a43262458a92160898",
      "tree": "86a49759e93940a80c940703a0d03a7084a90dd1",
      "parents": [
        "8da3d456505ec29a27d73720731a929cfab46b3f"
      ],
      "author": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Wed Apr 30 19:47:16 2025 +0000"
      },
      "committer": {
        "name": "Josh Bleecher Snyder",
        "email": "josharian@gmail.com",
        "time": "Thu May 01 12:39:18 2025 -0700"
      },
      "message": "ant: improve encapsulation\n\n- Replace string literals with package constants for message roles and content types.\n- Create UserStringMessage helper function to simplify user message creation\n- Replace manual Content creation with ant.StringContent()\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "9373c07a6d2fc97c2fad41e8228a7f2e8bc10ccb",
      "tree": "f7485f9715aee87c185ba2113b2eab6c41b1398d",
      "parents": [
        "b7c5875548da5057eac0405bf4e0ae8bbc43667c"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 01 10:27:01 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 01 12:13:31 2025 -0700"
      },
      "message": "Update tests for agent\u0027s new iterator implementation\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n\nThe commit 4eb6bb6c9ac switched from outbox channels to a subscriber model\nwith message iterators. This commit updates all existing tests to use the\nnew iterator implementation and adds new tests specifically for the iterator\nfunctionality.\n\n- Modified agent_test.go to use new iterator approach\n- Modified agent_git_test.go to check history array directly\n- Added new iterator_test.go to test iterator-specific functionality\n\nFix issues found by code review\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n\n- Fix context leak by properly using cancel function\n- Modernize for loops using range over int\n- Use fmt.Appendf instead of []byte(fmt.Sprintf)\n\nOptimize slow tests to improve test execution time\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n\nThis commit optimizes the test suite to reduce execution time:\n\n1. Reduces TestGitCommitTracking test time by using fewer git commits\n   (20 instead of 110) which cuts the test time by 77%.\n\n2. Improves iterator tests by using channels for synchronization instead\n   of time.Sleep, reducing TestIteratorWithNewMessages from 200ms to nearly 0ms.\n\n3. Reduces timeouts and sleeps in other iterator tests.\n\nTotal test execution time reduced by more than 60% (from 2.3s to 0.9s).\n"
    },
    {
      "commit": "b7c5875548da5057eac0405bf4e0ae8bbc43667c",
      "tree": "a0b9eb0f68de826d7b51a5e8f5eaa7976fb48de3",
      "parents": [
        "5228b5850d8453c225e5611918fcd00b747c65cc"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 01 10:10:17 2025 -0700"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Thu May 01 12:13:31 2025 -0700"
      },
      "message": "agent plumbing: convert outbox to subscribers and an iterator\n\nWaitForMessage() could only work for one thread, because it was using a\nsingular channel for outboxes. This was fine when we only had one user,\nbut WaitForMessageCount() was kinda similar, and had its own thing, and\nI want to rework how polling works and need another user.\n\nAnyway, this one is hand-coded, because Sketch really struggled\nwith getting the iterator convincingly safe. In a follow-up commit,\nI\u0027ll try to get Sketch to write some tests.\n"
    },
    {
      "commit": "96b60dd7e71c7db55dfe4cff86265eab02e54e9f",
      "tree": "3ee9ddbbe4b260b7000f194e369b527ce658cd66",
      "parents": [
        "bce3a13a7c16ffdb602b66550e6b3479d34fb9b0"
      ],
      "author": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 09:49:10 2025 -0700"
      },
      "committer": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 15:20:35 2025 -0700"
      },
      "message": "Add StateMachine tracking to Agent control flow\n\nThis commit integrates the existing StateMachine type with the Agent\nto provide state tracking and validation throughout the conversation\ncontrol flow. This allows for better monitoring and debugging of the\nAgent\u0027s behavior during execution.\n\nThis commit adds tests to verify the correct behavior of the state\nmachine integration with the Agent type:\n\n1. A test for basic state transitions\n2. A test for a complete processTurn flow with a simple response\n3. A test for a processTurn flow with tool use\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "2c4db099e6763cf18f0c57876298b9f5c70ed925",
      "tree": "ae4408992dd74fa0b5392523d2e6713e388326e7",
      "parents": [
        "b74b526b579320c08ad85cc8b4bdd719c9d6d7c8"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip@bold.dev",
        "time": "Mon Apr 28 16:57:50 2025 -0700"
      },
      "committer": {
        "name": "Autoformatter",
        "email": "bot@sketch.dev",
        "time": "Wed Apr 30 19:59:45 2025 +0000"
      },
      "message": "Restart conversation support.\n\nThe idea here is to let the user restart the conversation, possibly with\na better prompt. This is a common manual workflow, and I\u0027d like to make\nit easier.\n\nI hand wrote the agent.go stuff, but Sketch wrote the rest.\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "9f4b8089f93f39300b819ccf83cf64a7a867c0dc",
      "tree": "c1f00c2b00892efd2656196ebbc37c433d138658",
      "parents": [
        "bc6b629676e97d1789cff9a1d945b5faf680372e"
      ],
      "author": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 17:34:07 2025 +0000"
      },
      "committer": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 11:28:34 2025 -0700"
      },
      "message": "agent.go: don\u0027t panic on nil initialResp\n\na.processUserMessage(ctx) appears to return nil, nil under\ncertain circumstances, and this causes a panic in processTurn().\n\nThis change makes it log the error instead of panicing.\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "885a16a8e37fe82dac66a7db729be202f6ce4bea",
      "tree": "eff84390d70dda2a09c79faffad2ba0850c1b763",
      "parents": [
        "0d95d3a268ffad82b7c645200804fb43ed088a81"
      ],
      "author": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 02:49:25 2025 +0000"
      },
      "committer": {
        "name": "Sean McCullough",
        "email": "banksean@gmail.com",
        "time": "Wed Apr 30 09:41:22 2025 -0700"
      },
      "message": "Refactor loop/agent.go to reduce complexity\n\n- Restructured Agent.InnerLoop into smaller, more focused functions\n- Renamed InnerLoop to processTurn to better reflect its purpose\n- Extracted helper methods for different responsibilities\n- Improved code organization and testability\n- Each extracted function now handles a single responsibility\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "99a9a0285fe449f047fbc93766f37e6ea7af21df",
      "tree": "ba255c9b05449c09d137cd6cf81edf58469542a1",
      "parents": [
        "f98d7305e73ecec9a78f94f527693d5adfb89159"
      ],
      "author": {
        "name": "Philip Zeyliger",
        "email": "philip.zeyliger@gmail.com",
        "time": "Sun Apr 27 15:15:25 2025 +0000"
      },
      "committer": {
        "name": "Philip Zeyliger",
        "email": "philip.zeyliger@gmail.com",
        "time": "Sun Apr 27 20:00:22 2025 -0700"
      },
      "message": "Implement tracking of outstanding LLM and Tool calls\n\nThis commit implements a listener pattern between ant.convo and the Agent for tracking outstanding calls.\n\n* Added fields to the Agent struct to track outstanding LLM calls and Tool calls\n* Implemented the listener methods to properly track and update these fields\n* Added methods to retrieve the counts and names\n* Updated the State struct in loophttp.go to expose this information\n* Added a unit test to verify the tracking functionality\n* Created UI components with lightbulb and wrench icons to display call status\n* Added numerical indicators that always show when there are active calls\n\nCo-Authored-By: sketch \u003chello@sketch.dev\u003e\n"
    },
    {
      "commit": "2e463fb649fcff14d4025ddb91f630a98e7da526",
      "tree": "0e86854d80d2759a913870655f13226c31f9d30c",
      "parents": [],
      "author": {
        "name": "Earl Lee",
        "email": "earl.lee@sketch.dev",
        "time": "Thu Apr 17 11:22:22 2025 -0700"
      },
      "committer": {
        "name": "Earl Lee",
        "email": "earl.lee@sketch.dev",
        "time": "Thu Apr 17 11:35:33 2025 -0700"
      },
      "message": "Initial commit\n"
    }
  ]
}