Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
81 commits
Select commit Hold shift + click to select a range
6c481cf
refactor(core): defer benchmark implementations during capability dis…
cocolord Nov 18, 2026
e67b5d9
Defer Yardstick evaluation until an enabled Explore read
cocolord Nov 18, 2026
5ff017b
Omit empty Todo summary lanes from ordinary Worker context
cocolord Nov 18, 2026
23042c8
fix(update): honor configured archive directories during rollback
loopx-agent Nov 19, 2026
2e811cc
Merge pull request #9060 from loopx-project/codex/worker-todo-lanes-o…
huangruiteng Nov 19, 2026
c22ad13
Merge pull request #9059 from loopx-project/codex/yardstick-on-demand…
huangruiteng Nov 19, 2026
cc290b2
Merge pull request #9034 from loopx-project/codex/minimal-core-closur…
huangruiteng Nov 19, 2026
7dc4164
fix(lark): replay recall guard on current main
huangruiteng Nov 19, 2026
d47c8f9
Merge pull request #8847 from loopx-project/codex/reject-recalled-lar…
huangruiteng Nov 19, 2026
f976a29
docs: record the verified LoopX 2.3.0 release baseline
loopx-agent Nov 19, 2026
c47655b
Merge pull request #9068 from loopx-project/codex/record-release-230-…
huangruiteng Nov 19, 2026
e92bbb1
Preserve quota spend source when prepared work is admitted
loopx-agent Nov 19, 2026
ed0edeb
test(quota): qualify native saved complete-Turn checkpoints
loopx-agent Nov 18, 2026
b83fb5c
Retry cleanup of an owned file mutation marker
loopx-agent Nov 19, 2026
c37ed1c
Keep synthetic Git-path test data outside the public source scan
loopx-agent Nov 18, 2026
791c090
fix(quota): share native preparation wait with readiness
loopx-agent Nov 18, 2026
982e941
Merge pull request #9010 from loopx-project/codex/c1-native-complete-…
huangruiteng Nov 19, 2026
5492139
Merge pull request #9039 from loopx-project/codex/public-boundary-git…
huangruiteng Nov 19, 2026
98b7d8e
Merge pull request #9064 from loopx-project/codex/c1-ssh-goal-spend-s…
huangruiteng Nov 19, 2026
b02627d
Merge pull request #9008 from loopx-project/codex/c1-shared-native-pr…
huangruiteng Nov 19, 2026
5a0f6ee
Merge pull request #9067 from loopx-project/codex/retry-own-mutation-…
huangruiteng Nov 19, 2026
44ce91d
fix(doctor): bind package update readback to its Python environment
loopx-agent Nov 18, 2026
cb56cdb
Keep Doctor discovery separate from service activation
loopx-agent Nov 18, 2026
788c598
Verify the recorded Python launcher before Doctor probes it
loopx-agent Nov 18, 2026
0673655
fix(desktop): require runtime qualification for macOS preview updates
loopx-agent Nov 18, 2026
8467c8e
fix(quota): reuse replan projection owners during recovery
loopx-agent Nov 18, 2026
685cbea
Qualify Codex 0.169 native context replay on main
loopx-agent Nov 18, 2026
0016317
Merge pull request #9046 from loopx-project/codex/quota-replan-projec…
huangruiteng Nov 19, 2026
1e2c7a4
Merge pull request #9044 from loopx-project/codex/c1-native-replay-wr…
huangruiteng Nov 19, 2026
8b1b4ff
Merge pull request #9009 from loopx-project/codex/doctor-pin-python-u…
huangruiteng Nov 19, 2026
036a95f
Merge pull request #9029 from loopx-project/codex/macos-preview-adopt…
huangruiteng Nov 19, 2026
45dce4c
Stop regenerating retired quota view aliases
loopx-agent Nov 18, 2026
af341e8
fix: pin deferred native context reads to the original capture
loopx-agent Nov 18, 2026
27e62d8
Keep native capture navigation within small output budgets
loopx-agent Nov 19, 2026
3495df1
Retire legacy heartbeat work prompt generation
loopx-agent Nov 18, 2026
4e4ef3f
Capture C1 composition failures as receipt diagnostics
loopx-agent Nov 18, 2026
ce50275
Merge pull request #9054 from loopx-project/codex/c1-heartbeat-prompt…
huangruiteng Nov 19, 2026
c71b4ff
Merge pull request #9055 from loopx-project/codex/c1-generated-comman…
huangruiteng Nov 19, 2026
04bee59
Merge pull request #9052 from loopx-project/codex/c1-composition-fail…
huangruiteng Nov 19, 2026
1e3c88b
Merge pull request #9048 from loopx-project/codex/c1-deferred-context…
huangruiteng Nov 19, 2026
8ab59c2
Recover interrupted native Todo readiness from the original Turn
loopx-agent Nov 18, 2026
61c38ad
Revalidate the original lease before publishing resumed native context
loopx-agent Nov 19, 2026
f3af11a
Fence native readiness publication with the canonical lease writer
loopx-agent Nov 19, 2026
a52afc8
fix(scheduler): keep Codex fallback on the original C1 runtime
loopx-agent Nov 18, 2026
e2b3362
fix(scheduler): pin the originating fallback launchers
loopx-agent Nov 18, 2026
1f5c73a
Retire the duplicate Codex App scheduler projection
loopx-agent Nov 18, 2026
610fd87
Seed the recorded scheduler profile in the elapsed-backoff fixture
loopx-agent Nov 18, 2026
c19819a
Verify canonical cadence on refused and settled Turns
loopx-agent Nov 18, 2026
0946e69
Pin named Core commands in saved scheduler fallbacks
loopx-agent Nov 18, 2026
8b8ff27
fix(research): preserve saved case addresses when a version cannot open
loopx-agent Nov 18, 2026
e7ae172
Preserve citation focus after a saved study refresh
loopx-agent Nov 18, 2026
7600293
Merge pull request #9050 from loopx-project/codex/c1-native-readiness…
huangruiteng Nov 19, 2026
14dbfcb
Merge pull request #8983 from loopx-project/codex/research-case-readi…
huangruiteng Nov 19, 2026
39f37bf
Merge pull request #9057 from loopx-project/codex/c1-scheduler-fallba…
huangruiteng Nov 19, 2026
9dbe0f2
Capture authenticated LoopX receipts for Research attention (#8790)
cocolord Nov 19, 2026
ff9ba22
Recover the original Session and Turn from a confirmed Chat task
huangruiteng Nov 19, 2026
031122f
perf(lark,native_chat): eliminate redundant full-directory JSON repar…
loopx-agent Nov 19, 2026
0f7b545
Retire transitional smoke checks and duplicate execution (#9073)
cocolord Nov 19, 2026
27eaeb1
feat(steward): support probe-gated generic continuous_monitor todos w…
loopx-agent Nov 19, 2026
62784ec
Remove redundant defensive prose from prompts and comments (#9072)
cocolord Nov 19, 2026
c6828f0
Merge commit '62784ecc532585d4f277dd94fe9a169fb95b5050' into codex/au…
loopx-agent Nov 19, 2026
85e0aa1
Keep original Chat retry errors visible during receipt refresh
loopx-agent Nov 19, 2026
ff90543
fix(steward): tolerate report validator python drift and allow cancel…
loopx-agent Nov 19, 2026
6a0f8bc
feat(steward): add exact-head PR review signals and protected merge c…
loopx-agent Nov 19, 2026
7a04d68
feat(steward): add isolated OpenViking long-term memory with per-spea…
loopx-agent Nov 19, 2026
6589992
refactor(content-ops): slim surface module and eliminate Any (#8961)
loopx-agent Nov 19, 2026
fac7d35
fix(collaboration): batch handoff query Goal lifetime guard lock per …
loopx-agent Nov 19, 2026
898c9ef
fix(hosts): raise managed service startup readiness timeout to 45s (#…
loopx-agent Nov 19, 2026
1c540c3
perf(manager-context): memoize history_inspect lifecycle checks and s…
loopx-agent Nov 19, 2026
e19f3de
refactor(lark): slim private_conversations below 2000 lines with net …
loopx-agent Nov 19, 2026
dac0eae
perf(chat,hosts): skip frontend source re-hashing on immutable instal…
loopx-agent Nov 19, 2026
9e066cd
refactor(control-plane): split and slim quota monitor_poll_commit (#8…
loopx-agent Nov 19, 2026
6e1a0bd
refactor(quota): slim loopx/quota.py and tighten Any count to zero (#…
loopx-agent Nov 19, 2026
109ed4b
refactor(native-chat): slim external_conversations below 2k lines and…
loopx-agent Nov 19, 2026
b927e20
refactor(chat): slim chat_runtime below 1,500 lines, fix entry-point …
loopx-agent Nov 19, 2026
45c015f
fix(steward): paginate open PR signal GraphQL, allow signal-unavailab…
loopx-agent Nov 19, 2026
3a1f5ea
perf(lark,native_chat): back off noop recovery retries and cache reco…
loopx-agent Nov 19, 2026
1047f3b
Merge pull request #8729 from loopx-project/codex/audit-chat-original…
loopx-agent Nov 19, 2026
477def5
perf(native_chat,lark): cache unchanged quota backlog health and inte…
loopx-agent Nov 19, 2026
8e9a28a
perf(lark,native_chat): coalesce listener snapshots and cache recover…
loopx-agent Nov 19, 2026
a36b171
Separate archive layout coverage from paired launcher tests
loopx-agent Nov 19, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
19 changes: 7 additions & 12 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ Agent 团队的长程效果和长程效率:在可比预算下取得更好的
- **Governed continuation:** heartbeat, scheduling, quota and explicit stop/wait
conditions decide when authorized work can continue.
- **Peer collaboration:** claims, leases, scoped handoffs, dependent artifacts,
receiver adoption and result return; a coordinator gains no implicit authority.
receiver adoption and result return.
- **Personal Workspace:** App/frontend, Lark and CLI expose conversations,
settings, work, decisions, artifacts and readback over the same typed owners.
- **Capabilities and extensions:** reusable engineering, research, knowledge,
Expand All @@ -40,7 +40,7 @@ prompt. Keep the root within both the 24,000-byte (24 KB) Antigravity per-file
rule cap and the default 32 KiB Codex discovery budget, and do not eagerly
import linked detail documents. See
[loading](docs/development/README.md#agent-instruction-loading-and-maintenance)
for host-specific checks; delivery does not prove model compliance.
for host-specific checks.

Prefer [context](docs/development/README.md#progressive-context-for-development):
current goals, constraints and acceptance criteria first; supporting evidence and
Expand All @@ -65,16 +65,14 @@ acceptance is proven; reconcile the existing checkpoint as evidence changes.
Carry one compact delivery brief from task to PR (goal/source, current gap,
observable result, owning boundary and decisive acceptance evidence), keeping
private Goal state out of public artifacts. Choose a complete, independently
reviewable and reversible outcome slice; small diffs, fields, receipts, test
counts and merged PR counts do not establish progress. Characterization,
reviewable and reversible outcome slice. Characterization,
prerequisites, research, docs and maintenance are valid when they remove an
evidenced gap or enable a named next step.

For long-horizon improvements, trace the accepted goal and structured state
through feedback, the replan decision, subsequent work and validated outcomes.
Check whether new evidence changed the action or justified continuing it, and
preserve negative results and uncertainty (`state survival`, graph edges, read
receipts or replan records alone do not prove adoption or benefit). Compare
preserve negative results and uncertainty. Compare
outcomes and complete costs under comparable conditions using the
[measurement contract](docs/architecture/rfcs/long-horizon-harness-benchmark-research-program-v0.md#6-measurement-model);
do not force a direction change merely to increase replan counts.
Expand All @@ -93,16 +91,15 @@ a registered vocabulary, remains local or creates a shared contract, and follow
the [diff advisory](CONTRIBUTING.md#check-new-semantic-vocabulary-while-developing)
before the full-tree check (including intentional untracked sources). For
multi-Agent changes, qualify dependency artifacts, receiver adoption, claim/lease
handling, independent acceptance and result return; sending a message or
registering workers does not prove collaboration, and missing frontend/Lark/CLI
handling, independent acceptance and result return; missing frontend/Lark/CLI
companion work makes a product journey partial.

Before delivery, reconcile the result with the original/current goal, update its
task and RFC checkpoint when the boundary changes, and preserve passed, failed
and untested distinctions. If user feedback exposes the same missing outcome,
repair the owning rule, active task or projection through self-repair rather
than appending stronger prose. PR review must execute the capability-owned
`problem_context` delivery judgment; author declarations do not certify it.
`problem_context` delivery judgment.

## Commit And PR Hygiene

Expand Down Expand Up @@ -171,9 +168,7 @@ Before staging non-trivial changes:

Before approval or merge, read the repository-kept
[merge workflow](skills/loopx-pr-merge/SKILL.md) and its
[self-merge rules](skills/loopx-pr-merge/SKILL.md#self-merge-eligibility-and-benchmark-exceptions);
installing it is optional and does not replace repository policy or the current
capability-owned review.
[self-merge rules](skills/loopx-pr-merge/SKILL.md#self-merge-eligibility-and-benchmark-exceptions).

Control-plane, runtime and product surfaces (`loopx/**`, `apps/**`, `packages/**`,
permissions/authority, persisted state/receipts, quota/scheduler/todo rules, and
Expand Down
8 changes: 3 additions & 5 deletions README.ja.md
Original file line number Diff line number Diff line change
Expand Up @@ -105,8 +105,7 @@ LoopX は、永続的な作業状態を 1 つのコンパクトなレイヤー
</picture>

作業を実行するのは Agent ランタイムです。LoopX は、エンジニアリング、研究、探索、
運用のループが実行をまたいで継続できるようにするための状態を保持します。これは
新たなエージェントフレームワークでも、特定プロバイダー向けのオーケストレーションランタイムでもありません。
運用のループが実行をまたいで継続できるようにするための状態を保持します。

<details>
<summary>詳細:コントロールレイヤーの仕組みと位置付け</summary>
Expand Down Expand Up @@ -136,7 +135,7 @@ LoopX は、次のようなものを実行するときに役立ちます:
- 進捗がエンジニアではないオペレーターにも理解できる状態でなければならない、
クリエイター、研究、運用のワークフロー

LoopX は自律的な本番コントローラーではありません。危険な権限、
危険な権限、
公開、本番環境への書き込み、最終的なオーナーシップは人間に残ります。

</details>
Expand Down Expand Up @@ -867,8 +866,7 @@ LoopX は、長時間稼働するエージェントのエコシステムを築
## 現在のステータス

LoopX は、長時間にわたるエージェントの作業のための実用的なローカルコントロールプレーンであり、
より広い導入の段階に入りつつあります。完全なエージェントプラットフォーム、エージェントランタイム、
自律的な本番コントローラーではありません。
より広い導入の段階に入りつつあります。

現在の LoopX は、ゴール、型付きの todo と意思決定の
スコープ、対等な claim とリース、エビデンスと writeback、クォータを考慮したスケジューリング、
Expand Down
29 changes: 10 additions & 19 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -138,7 +138,7 @@ LoopX is useful when you run:
- creator, research, or operations workflows whose progress must remain
legible to a non-engineering operator.

LoopX is not an autonomous production controller. Dangerous permissions,
Dangerous permissions,
publishing, production writes, and final ownership stay with the human.

</details>
Expand Down Expand Up @@ -390,14 +390,11 @@ benchmark's strict solved threshold. No score in this study equals 0.80, so
these counts also match the study brief's >0.80 view.

Against native Goal, task-level rewards improve on **23**, tie on **13**, and
regress on **10** tasks. Strict solves match Plain Codex, so the mean gain is
not a claim that every task improves or that more tasks are fully solved than
Plain. Mean reward also credits partial progress.
regress on **10** tasks. Strict solves match Plain Codex. Mean reward also credits partial progress.

One effective trial per task and mode; designated replacement trials and
unequal runtime/budgets, including longer budgets for some Heartbeat replacements.
These are observed system results, not an equal-budget efficiency result or
an isolated causal estimate of LoopX's effect.
These are observed system results.

[Interactive results, task comparisons and methodology](https://loopx-project.github.io/loopx/benchmarks/lhtb/)
· [Five-arm study and limits](benchmark/LHTB/studies/five-arm-gpt56sol-max/README.md)
Expand Down Expand Up @@ -564,7 +561,7 @@ its own policy and authority checks before proposing a transition.
| Session dash | Starts a live single-page panel that tracks fleet progress: sessions, their goals, and each goal's status/todo progress, with result statistics; auto-refreshes in place. | `loopx dash`, [session dash design](docs/product/surfaces/session-dash-panel-design.md) |
| External projections | Projects todos and gates into collaboration surfaces while LoopX remains authoritative. | `loopx lark-kanban`, [Lark Kanban adapter](docs/integrations/lark-kanban-control-plane-adapter.md) |
| Domain capabilities | Packages repeatable work lanes such as issue fixing, content operations, value connector planning, ML experiment advice, benchmark evidence, and Explore. | `loopx issue-fix`, `loopx content-ops`, `loopx value-connectors`, `loopx ml-experiment`, `loopx benchmark`, [Explore](loopx/capabilities/explore/README.md) |
| Experimental context learning | Lets named registered agents trial provider-neutral Reward Memory through ignored, default-off project configuration. OpenViking is one provider option, not a global dependency. | `loopx reward-memory experiment-status`, [Reward Memory architecture](loopx/capabilities/reward_memory/README.md) |
| Experimental context learning | Lets named registered agents trial provider-neutral Reward Memory through ignored, default-off project configuration. OpenViking is one provider option. | `loopx reward-memory experiment-status`, [Reward Memory architecture](loopx/capabilities/reward_memory/README.md) |
| Governance patterns | Captures reusable routing, gate, evidence, projection, and planning shapes. | [interaction patterns](docs/concepts/interaction-pattern-catalog.md), [state model](docs/state-interaction-model.md) |

The shipped primitives include lifetime goals, concrete user gates, audited safe
Expand All @@ -583,8 +580,7 @@ signals, and public/private boundary checks.

The execution path is `Agent -> Capability -> Provider`; the control path
returns `Provider readback -> Capability transition -> Kernel`. An extension is
how an optional provider is packaged and managed, not another control-plane
owner. See [Architecture](docs/architecture.md) and
how an optional provider is packaged and managed. See [Architecture](docs/architecture.md) and
[Extensions and Capabilities](docs/reference/extensions.md).

</details>
Expand All @@ -599,8 +595,7 @@ owner. See [Architecture](docs/architecture.md) and
[Anthropic's *Building always-on agents*](https://claude.com/resources/articles/building-always-on-agents)
make persistent personal agents and sleep/wake event-driven runs a familiar product
and architecture model. LoopX approaches that space as an open, provider-neutral
[24/7 control plane](#always-on-agents) for agents you already run—not as a hosted
replacement for either product.
[24/7 control plane](#always-on-agents) for agents you already run.

- **Personal agent:** use the shipped Workspace and connected Lark surfaces to
inspect goals, steer work and resolve decisions. The
Expand All @@ -615,10 +610,9 @@ replacement for either product.
[Reward Memory](loopx/capabilities/reward_memory/README.md), and evaluate
candidate changes through [Explore](loopx/capabilities/explore/README.md).
Both are optional. The link to **recursive self-improvement (RSI)** is the
engineering loop—propose, test, review, retain—not a claim that LoopX has
demonstrated autonomous model improvement or compounding capability gains.
engineering loop—propose, test, review, retain.

A personal steward is an interaction role, not a second source of authority.
A personal steward is an interaction role.
Background execution still needs an available host and configured runtime.
The [overall roadmap](docs/architecture/rfcs/loopx-overall-roadmap-v0.md)
separates shipped foundations from the next end-to-end acceptance milestones.
Expand Down Expand Up @@ -876,8 +870,7 @@ long-running agent ecosystem. Our confirmed partners include:
## Current Status

LoopX is a usable local control plane for long-running agent work and is
entering broader adoption. It is not a full agent platform, an agent runtime, or
an autonomous production controller.
entering broader adoption.

Today LoopX ships a durable state kernel for goals, typed todos and decision
scopes, peer claims and leases, evidence and writeback, quota-aware scheduling,
Expand All @@ -888,9 +881,7 @@ Workspace all build on that shared control state.

Support levels remain explicit. The state and CLI contracts are the stable
center; several host integrations and advanced paths are optional, default-off,
or experimental. LoopX does not grant credentials, approve destructive or
production actions, publish on a user's behalf without authorization, or turn
an unverified run into evidence of success.
or experimental.

Current investment is organized through the
[Technical Directions map](docs/project/technical-directions.md): long-horizon
Expand Down
Loading
Loading