Agent Pulse / Daily decision brief / 2026-10-23

OpenCode can prepare a summary before context fills

Snapshot as of 2026-10-23. Monitoring counts and recommendations refer to that date; Pulse refreshes independently of articles.

OpenCode has merged experimental background compaction: it can prepare a conversation summary while the current conversation continues, then use it when compaction becomes due. The option remains off by default.

One decision-worthy insight. Action, effort, and rollback. 29 / 29 included projects monitored. Methodology ›

Today's Action

Watch / Project-specific claim
watch

OpenCode can prepare a summary before context fills

What changed. The implementation keeps the summary aside until needed, preserving the current conversation prefix. It checks that the summary still fits the conversation and can fall back to ordinary compaction. If the summary is unfinished, compaction waits for it; unused background work still consumes tokens.

Why it matters. For long coding sessions, preparing work early offers a different way to manage the compaction pause. Its value depends on whether the prepared summary is used and preserves the information needed to finish the task.

Teams evaluating long OpenCode sessions One bounded comparison after release containment is checked

Watch for a pinned workload comparison

Before changing a production configuration, confirm a containing release and compare one disposable long task with background_compaction off and on. Record completion, summary fidelity, waiting time and all token usage. Keep the original configuration for rollback.

Reversible

Evidence Chain

3 direct receipts
Confidence / developing

Merged code, subsequent configuration documentation and scripted test definitions inspected. No live task, suite execution or containing release verification was performed for this brief.

Limit

This is an experimental source capability, not a measured reduction in pause time or total cost. A summary can be unused, invalidated or unfinished. No model-quality or workflow result was measured.

What would change this

A pinned containing implementation cannot retain and adopt a compatible prepared summary as described, or matched tasks show the extra work does not improve the intended waiting-time tradeoff.

Watch next. A containing release and matched long-task outcomes with complete token accounting, including discarded summaries.

Secondary Signals

Context, not recommendations

These source movements did not become today's decision brief. They remain visible so readers can inspect the receipts without mistaking activity for consequence.

anomalyco/opencodeOpen receipt ›

anomalyco/opencode

anomalyco/opencode shows 282 public commits this week, up 24 from the prior week and 9.2 below its current 30-day pace, with ux as the strongest subject signal.

NousResearch/hermes-agentOpen receipt ›

NousResearch/hermes-agent

NousResearch/hermes-agent shows 2,565 public commits this week, down 228 from the prior week and 859.2 below its current 30-day pace, with tools as the strongest subject signal.

laude-institute/rev-engineOpen receipt ›

laude-institute/rev-engine

laude-institute/rev-engine shows 49 public commits this week, up 9 from the prior week and 23.1 above its current 30-day pace, with ux as the strongest subject signal.

Tracking Health

29 / 29 monitored today

29 freshly tracked / 13 of 13 core coding agents / 23 excluded with reasons / 0 unavailable / 0 stale. All 52 watchlist repos have an explicit decision.

Read all 52 tracking decisions
openai/codex Tracked

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:45 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:45 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:45 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:45 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:45 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:46 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:48 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:49 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:50 PM GMT+2
git/git Excluded, with reason

Watched for source-control infrastructure stories, but excluded from Agent Pulse project rows because it is a substrate rather than an AI-agent project and its general development volume would distort the editorial filter.

Editorial scope decision

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:51 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:51 PM GMT+2
betagouv/api.gouv.fr Excluded, with reason

Watched as the first API Discovery provider catalog, but excluded from Agent Pulse because it belongs to the daily source-brief lane rather than the AI-agent project shelf.

Editorial scope decision
cbouvat/saracroche-ios Excluded, with reason

Watched for an editor-requested public mobile source profile, but excluded from Agent Pulse because Saracroche is a call-filtering application, not an AI-agent or agent-adjacent project.

Editorial scope decision

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:51 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:51 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:51 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:53 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:55 PM GMT+2
cline/cline Tracked

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:57 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:57 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:57 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:59 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 12:59 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:00 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:00 PM GMT+2
yc-software/qm Tracked

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:00 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:00 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:01 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:04 PM GMT+2

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:04 PM GMT+2
huggingface/dataset-viewer Excluded, with reason

Watched for an editor-requested source audit of agent-driven infrastructure abuse, but excluded from Agent Pulse because Dataset Viewer is a data-processing service rather than an AI-agent project.

Editorial scope decision
TangleML/tangle Excluded, with reason

Watched for source-grounded reporting on machine-learning pipeline orchestration, but excluded from Agent Pulse because Tangle is a pipeline application rather than an AI-agent or agent-adjacent project.

Editorial scope decision
TangleML/tangle-ui Excluded, with reason

Watched for source-grounded reporting on Tangle’s machine-learning pipeline interface, but excluded from Agent Pulse because it is the companion UI for a non-agent pipeline application and should not widen or double-count the AI-agent shelf.

Editorial scope decision
ollama/ollama Excluded, with reason

Commissioned source; monitored for reporting but excluded from Agent Pulse pending editor scope decision.

Editorial scope decision
algorithmicsuperintelligence/optillm Excluded, with reason

Commissioned source; monitored for reporting but excluded from Agent Pulse pending editor scope decision.

Editorial scope decision

Commissioned source; monitored for reporting but excluded from Agent Pulse pending editor scope decision.

Editorial scope decision
NandhaKishorM/laya Excluded, with reason

Watched as a local decision-model alternative for source-grounded reporting; excluded from the coding-agent Pulse roster pending an editor scope decision.

Editorial scope decision
ickma2311/jev-baselines-eval Excluded, with reason

Watched as public evaluation evidence for decision-model comparisons; excluded from coding-agent Pulse because a benchmark repository is not an agent runtime.

Editorial scope decision
nicobailon/pi-mcp-adapter Excluded, with reason

A Pi integration extension tracked for compatibility reporting, not a standalone agent runtime; its releases are supplemental evidence outside the included Agent Pulse runtime comparisons.

Editorial scope decision

Protocol design incubation repository, not a coding-agent runtime; tracked for event interoperability reporting outside Agent Pulse activity comparisons.

Editorial scope decision
huggingface/datasets Excluded, with reason

Dataset loading and benchmark inspection infrastructure is tracked for reporting; it is not an agent runtime competitor and remains outside Agent Pulse comparison.

Editorial scope decision
mikekelly/pi-mcp-events Excluded, with reason

A focused MCP event companion, tracked as an integration source rather than a standalone coding-agent runtime in the Pulse comparison.

Editorial scope decision
mikekelly/figma-listen Excluded, with reason

A Figma event-source server, tracked for integration evidence rather than as a standalone coding-agent runtime in the Pulse comparison.

Editorial scope decision
strands-labs/strands-decider Excluded, with reason

A decision-model training and inference library tracked for the commissioned System 1/System 2 comparison, rather than a standalone coding-agent runtime in the Pulse comparison.

Editorial scope decision
charmbracelet/bubbletea Excluded, with reason

A general terminal UI library tracked as the dependency carrying program-status reports, not an autonomous coding-agent runtime; excluded from Agent Pulse activity comparisons.

Editorial scope decision
Parcha-ai/agentrun Excluded, with reason

Tracked for the commissioned AgentRun workflow and decision-model deep dive; excluded from the standalone coding-agent Pulse comparison because this repository supplies a workflow layer and integrations for existing agents.

Editorial scope decision

A port-scheduling evaluation environment, tracked for benchmark validity reporting rather than included in the coding-agent activity comparison.

Editorial scope decision

Included in Agent Pulse with a readable, recently refreshed public-git source.

Oct 23, 01:06 PM GMT+2
deepseek-ai/DeepInfer Excluded, with reason

Tracked as model-serving infrastructure for the commissioned DeepSeek production and local-inference investigation; it is not a coding-agent runtime and is outside the current Agent Pulse activity comparison.

Editorial scope decision
deepseek-ai/deepseek-recipe Excluded, with reason

Tracked as a protocol-conversion and prompt-encoding library for the commissioned DeepSeek investigation; it does not execute agent tools or provide an agent runtime, so remains outside the current Agent Pulse comparison.

Editorial scope decision
gradium-ai/xn-ptts Excluded, with reason

Tracked as local speech synthesis infrastructure and model distribution for source reporting; it does not run agent reasoning or coding workflows and is outside the Agent Pulse coding-agent comparison.

Editorial scope decision

Project Ledger

Source map ›
Open the full measured ledger29 included projects
QwenLM/qwen-code Coding agent CLI / Tracked
456 7d commits

QwenLM/qwen-code shows 456 public commits this week, up 152 from the prior week and 151.5 above its current 30-day pace, with tools as the strongest subject signal.

30d 1,305 vs prior 7d +152 contributors 77 coverage 17 confidence High

Watch whether this burst becomes a sustained baseline or a one-week spike.

NousResearch/hermes-agent Agent framework / Tracked
2,565 7d commits

NousResearch/hermes-agent shows 2,565 public commits this week, down 228 from the prior week and 859.2 below its current 30-day pace, with tools as the strongest subject signal.

30d 14,675 vs prior 7d -228 contributors 1,035 coverage 44 confidence High

Watch whether the cooldown is a pause after heavy work or the start of a quieter cycle.

langchain-ai/langchain Agent framework / Tracked
70 7d commits

langchain-ai/langchain shows 70 public commits this week, up 48 from the prior week and 20.8 above its current 30-day pace, with tools as the strongest subject signal.

30d 211 vs prior 7d +48 contributors 17 coverage 15 confidence High

Watch whether the subject-proxy work turns into visible release or user-facing control changes.

anomalyco/opencode Coding agent / Tracked
282 7d commits

anomalyco/opencode shows 282 public commits this week, up 24 from the prior week and 9.2 below its current 30-day pace, with ux as the strongest subject signal.

30d 1,248 vs prior 7d +24 contributors 75 coverage 20 confidence High

Watch whether the subject-proxy work turns into visible release or user-facing control changes.

google-gemini/gemini-cli Coding agent CLI / Tracked
20 7d commits

google-gemini/gemini-cli shows 20 public commits this week, down 5 from the prior week and 2.9 below its current 30-day pace, with tools as the strongest subject signal.

30d 98 vs prior 7d -5 contributors 13 coverage 60 confidence High

Watch whether the subject-proxy work turns into visible release or user-facing control changes.

microsoft/agent-framework Agent framework / Tracked
52 7d commits

microsoft/agent-framework shows 52 public commits this week, down 67 from the prior week and 38.1 below its current 30-day pace, with ux as the strongest subject signal.

30d 386 vs prior 7d -67 contributors 73 coverage 9 confidence High

Watch whether the cooldown is a pause after heavy work or the start of a quieter cycle.

andrewyng/openworker Desktop agent runtime / Tracked
42 7d commits

andrewyng/openworker shows 42 public commits this week, up 26 from the prior week and 14.9 below its current 30-day pace, with permissions as the strongest subject signal.

30d 244 vs prior 7d +26 contributors 16 coverage 1 confidence High

Watch whether the subject-proxy work turns into visible release or user-facing control changes.

browser-use/browser-use Browser agent framework / Tracked
32 7d commits

browser-use/browser-use shows 32 public commits this week, down 23 from the prior week and 0 above its current 30-day pace, with tools as the strongest subject signal.

30d 137 vs prior 7d -23 contributors 7 coverage 1 confidence Medium

Watch the next tools commits for evidence that the theme is becoming product behavior.

6 7d commits

google-antigravity/antigravity-cli shows 6 public commits this week, up 1 from the prior week and 0.9 above its current 30-day pace, with observability as the strongest subject signal.

30d 22 vs prior 7d +1 contributors 1 coverage 4 confidence Medium

Watch the next observability commits for evidence that the theme is becoming product behavior.

mistralai/mistral-vibe Coding agent CLI / Tracked
5 7d commits

mistralai/mistral-vibe shows 5 public commits this week, up 5 from the prior week and 3.1 above its current 30-day pace, with ux as the strongest subject signal.

30d 8 vs prior 7d +5 contributors 5 coverage 5 confidence Medium

Watch the next ux commits for evidence that the theme is becoming product behavior.

laude-institute/headlong Persistent agent microharness / Tracked
2 7d commits

laude-institute/headlong shows 2 public commits this week, down 15 from the prior week and 14.3 below its current 30-day pace, with tools as the strongest subject signal.

30d 70 vs prior 7d -15 contributors 7 coverage 3 confidence Medium

Watch the next tools commits for evidence that the theme is becoming product behavior.

Coverage Boundary

Read methodology ›

V0.2 reports one dated decision brief from direct public source evidence and keeps git motion as a secondary measurement layer. Community feedback metrics are absent until a reproducible feedback collector ships.

local git commit historycommit subjectsrecent touched file pathspublished article coveragereported direct-source daily insight
not collected: GitHub issue flownot collected: GitHub PR review timingnot collected: Discord feedbacknot collected: X/Twitter feedbacknot collected: Reddit or Hacker News feedbacknot collected: private maintainer backlog