@iowarp/clio-coder 0.5.0 → 0.5.2

package/.claude-plugin/marketplace.json

@@ -57,16 +57,16 @@
 		{
 			"name": "clio-coder-dev",
 			"source": "./library/skills/meta/clio-coder-dev",
-			"description": "Governs changes to Clio Coder's own source and harness (TUI, skills, agents, tools, prompts, domains): what may change freely, what needs explicit user intent, and how to change it without breaking the architecture. Not for verifying a change against the harness; use clio-coder-test.",
-			"version": "0.4.0",
+			"description": "Develop Clio Coder inside its own repository: trace source contracts, implement bounded changes, coordinate sprint work, and preserve running-agent and session correctness. Use before changing Clio source, prompts, tools, or library; pair with clio-coder-test for validation.",
+			"version": "0.5.0",
 			"category": "meta",
 			"license": "Apache-2.0"
 		},
 		{
 			"name": "clio-coder-test",
 			"source": "./library/skills/meta/clio-coder-test",
-			"description": "Verifies a Clio Coder source change against the current contract, smoke, boundary, and application test lanes, including temporary-state and local-provider harnesses. Not for deciding what may change; use clio-coder-dev.",
-			"version": "0.3.1",
+			"description": "Validate Clio Coder changes using focused source regressions, built CLI and ACP fixtures, skill/library checks, and isolated package or GUI gates. Use when developing or diagnosing Clio itself; select checks from the changed behavior and report evidence without claiming unrun coverage.",
+			"version": "0.4.0",
 			"category": "meta",
 			"license": "Apache-2.0"
 		},

package/CHANGELOG.md

@@ -2,6 +2,41 @@
 
 All notable changes to Clio Coder are documented in this file. The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/) and versions follow Semantic Versioning; pre-1.0 minor releases may include incompatible changes.
 
+## 0.5.2 - 2026-09-21
+
+### Context continuity and memory
+
+- Add native `self_compact({note_to_self})` with exact assistant-authored notes, whole-batch exclusivity, durable preparation and reduction attempts, summary/eviction checkpoints, continuation delivery, and acknowledgement tied to the persisted successful response. Bound each handoff to two attempts and four summary invocations, and preserve the original operator turn across continuation.
+- Add `/context recover <handoffId> <reduce|deliver>` for explicit recovery, with branch-bound operator authority, preserved transaction identity, bounded deadlines, and no automatic redelivery after an uncertain interruption. Session format v5 retains continuity records through replay, forks, export, and later summaries; inherited notes remain recall only.
+- Enforce input plus reserved output at submission, settled tool continuation, and the final provider boundary. Price pending message framing and images, serialize reductions, retain calibrated accounting through eviction, and inspect the live budget through `context(scope="budget")` and `/context`.
+- Freeze approved durable-memory selection for a prepared turn and its continuations, using bounded content hashes that detect same-size rewrites. Retain existing default record/token limits and keep lexical relevance ranking experimental.
+- Restore private task memory only after successful durable reductions and admitted installation. Revoke pre-commit content jobs and queued reminders while retaining originating usage, deduplicate successful commit notifications, and close repeated-failure episodes on exact successful tool receipts.
+
+### Skills and harness correctness
+
+- Add bounded query/paging for skill discovery with a preserved response footer and visible installed-skill drift. Record exact skill artifacts in evaluation sidecars and independently judge the baseline for non-gating helped/regressed/mixed/no-change/unmeasured attribution.
+- Correct handoff guidance about persisted summaries, repair the first-turn system-prompt fixture, remove checkout-name dependence from the quota footer test, and stabilize retained-memory measurement across stream finalization without relaxing its memory ceiling.
+
+### Library
+
+- Add `wtfp@0.7.3` as a pinned remote catalog package without vendoring its files. Regenerate remote pins from fetched manifests and full-tree digests, preserve the previous pin with a warning when refresh fails, and keep remote packages out of the local Claude Code marketplace projection. Reject fetched package identity changes and report unverified retained pins separately.
+
+### Self-development skills
+
+- Redesign `clio-coder-dev` and `clio-coder-test` around source ownership, bounded sprint work, lifecycle correctness, isolated candidate builds, focused validation, and honest evidence. Update their reference maps, behavioral scenarios, and package metadata.
+- Discover these two source skills inside Clio's own checkout and give task-aware loading guidance through the existing skill interface. Preserve autonomy and worker bindings, no-skills mode, package ownership, hidden skills, and repository containment; keep skill bodies out of the base prompt until loaded.
+
+## 0.5.1 - 2026-09-21
+
+### subscription quota and unified usage
+
+- Read what each connected subscription account has left, before a refusal rather than after it. Adapters for Claude Code, `anthropic-max`, the Codex CLI, and Antigravity read stored credentials and normalize every reported window into one model of consumed percentage, so a provider's own percentages and another's remaining fractions are never interpreted interchangeably. Readings are cached for about five minutes and refreshed lazily when a surface that shows them renders, never on a background timer; a failed refresh keeps the last good reading and marks it stale instead of blanking the account. Reads are read-only: no token refresh, no credential write, no sign-in flow. An expired credential is reported with the provider's own guidance. A provider's own severity word outranks Clio's fallback thresholds, a provider that reports no 5h window gets no 5h row, and two credentials on one Anthropic subscription fold into a single account so the apparent budget is not doubled. Accounts that merely show equal percentages never merge.
+- Replace `/cost` with `/usage`, with no alias period. The overlay keeps the session token and cost accounting `/cost` carried and adds four views over it: Accounts (subscription and credit meters, every reported model group, reset countdowns with local reset times and an explicit time zone, and stale or expired readings), Session (the existing token, cost, cache, reasoning, and background-call breakdown), Models (each model's share of recorded processed tokens beside its detailed breakdown), and Workers (recorded tokens, cost, and context per run beside the shared account limit). Keys 1–4, Tab/Shift+Tab, and the arrows change views; the arrows, PgUp/PgDn, and Home/End scroll, and each view keeps its own position. `clio-coder usage report` is unchanged and remains the cross-session command.
+- Show the same cached readings on the welcome launchpad, the expanded dashboard, worker cards, and fleet islands. The launchpad's `Subscriptions` row sits in the field list beside the wordmark and wraps, so no connected account and not the free local-inference row is clipped. Status puts the session total above the account meters and no longer claims that providers report no quota; Activity shows shared account headroom on worker cards; Context states that request occupancy and account-wide limits are different quantities.
+- Give every percentage a direction in its own label. Detailed meters fill by consumed capacity, label that number `used`, and print the complement as `remaining`; compact badges carry the other direction and say `left`. A weekly window at 9% used and a footer badge reading `weekly 91% left` are the same reading.
+- Fix the compact footer's two-line contract. Line 1 carries active work and model identity, the selected model's weekly headroom beside it when the account and model group are identifiable, and thinking and context detail as width allows. Line 2 always keeps the working directory and Git branch or dirty state; notices, tips, and urgent prompts borrow the rotating hint area beside them rather than replacing the row. Quota never occupies line 2, and the footer never substitutes an unrelated account's busier window for the selected model's. Rendering stays bounded at narrow widths.
+- Present worker quota as shared account headroom, never as a measured per-worker share. A badge appears only for a supported local credential owner and the matching model group, keeping Antigravity's Gemini group distinct from its Claude and GPT group. Remote workers and unknown credential mappings stay unlinked rather than attributed to an account that may not be theirs, and Clio's own `openai-codex` sign-in is treated as separate storage from the Codex CLI's `auth.json`. Local inference is listed at `$0.00` with no subscription window consumed; no saved-quota figure is claimed, because per-target token attribution does not exist yet.
+
 ## 0.5.0 - 2026-09-20
 
 ### public launch and documentation

package/CONTRIBUTING.md

@@ -227,6 +227,15 @@ generated peer-host projection; do not edit it by hand. Core recipes under
 
 ## Working with coding agents
 
+In this repository, Clio discovers the local `clio-coder-dev` and
+`clio-coder-test` skills and recommends them for development and validation.
+Load them through the normal skill interface when the task calls for them;
+source discovery does not bypass autonomy, `--no-skills`, installed package
+state, or worker recipe bindings. See the
+[self-development skill behavior](library/skills/README.md#developing-clio-itself).
+When the current checkout's `dist/` backs the running agent, build and test a
+candidate in a separate worktree and restart a separate process to dogfood it.
+
 Read this file, the relevant changelog entry, any local `CLIO-CODER.md`, and the
 owning subsystem's documentation. Prefer `rg` and focused source reads. Preserve
 other work in a shared checkout and coordinate file ownership before editing.

package/README.md

@@ -39,10 +39,11 @@ engineering. Connect a model on your workstation, your institution's gateway,
 or a cloud service. Start with one conversation; add workers and specialized
 workflows when the task calls for them.
 
-**v0.5.0 brings Clio to its public launch:** the terminal is the primary working
-interface, and the bundled browser app provides documentation and an early
-preview of the wider graphical experience. Clio is pre-1.0 software. Review
-changes and validate scientific results with your own reference checks.
+**v0.5.2 adds durable self-compaction, context budgeting, and memory restoration.**
+The terminal is the primary working interface, and the bundled browser app
+provides documentation and an early preview of the wider graphical experience.
+Clio is pre-1.0 software. Review changes and validate scientific results with your
+own reference checks.
 
 ## Get started
 
@@ -101,6 +102,7 @@ For the bootstrap installer, source builds, upgrades, and removal, see
 | **Keep long tasks moving** | Context accounting, compaction, task memory, durable conversation history, and fork/resume navigation. | [Memory](docs/guide/proactive-memory.md) |
 | **Add domain knowledge** | Operator-activated skills, prompt templates, plugins, harness extensions, and local MCP tool servers. | [Library](library/README.md) |
 | **Inspect what happened** | Tool activity, usage, traces, decisions, evidence, and worker receipts. | [Observability](docs/architecture/observability.md) |
+| **Know what you have left** | `/usage` gathers subscription headroom for your connected accounts beside this session's tokens, cost, per-model shares, and worker accounting. | [Quota and usage](docs/guide/commands-and-modes.md#subscription-quota-and-session-usage) |
 
 These capabilities work together. A worker result is a lead you can inspect;
 a successful command is evidence of that command's outcome; neither substitutes
@@ -109,15 +111,16 @@ for scientific validation.
 ### Work in the terminal
 
 The TUI keeps the conversation, tool activity, diffs, and permission decisions
-in one place. Its footer shows the active model and context use; dashboards
-provide more detail without taking over the conversation.
+in one place. Its footer shows the active model, its remaining weekly
+subscription headroom where a connected account reports one, and context use;
+dashboards provide more detail without taking over the conversation.
 
 | Want to… | Use… |
 | --- | --- |
 | Find commands and shortcuts | `/help` |
 | Choose a model or change settings | `/model`, `/settings` |
 | Attach a project file | Type `@` and choose a path |
-| Inspect context, memory, or cost | `/context`, `/memory`, `/cost` |
+| Inspect context, memory, or usage | `/context`, `/memory`, `/usage` |
 | Browse skills and recipes | `/library` or `Alt+L` |
 | Inspect workers | `Alt+W` |
 | Manage the task board | `/tasks` |
@@ -245,12 +248,12 @@ bootstrap installer, upgrades, selective resets, and removal.
 <details>
 <summary><strong>Build and install a published release from source</strong></summary>
 
-From source, pin the release you intend to run. The following selects v0.5.0
+From source, pin the release you intend to run. The following selects v0.5.2
 once its release tag is published; for development before the tag, use the
 [contributor checkout instructions](CONTRIBUTING.md#set-up-a-checkout).
 
 ```bash
-git clone --branch v0.5.0 https://github.com/iowarp/clio-coder.git
+git clone --branch v0.5.2 https://github.com/iowarp/clio-coder.git
 cd clio-coder
 corepack enable pnpm
 pnpm run install:local

package/ROADMAP.md

@@ -1,16 +1,41 @@
 # Clio Coder roadmap
 
-Clio Coder is preparing v0.5.0 for its public launch. The terminal is the primary
-coding interface; bundled browser documentation and the early graphical app
-introduce the next part of the product.
+Clio Coder v0.5.1 is released. The next release focus is context management,
+compaction, and memory: preserving task continuity, making context decisions
+understandable, and reducing the cost of long sessions. The terminal remains
+the primary coding interface.
 
 This roadmap describes priorities, not available features or delivery promises.
 For implemented behavior, use the [documentation](docs/README.md). Historical
 release changes are in [CHANGELOG.md](CHANGELOG.md); older development plans in
 git history are not current contracts.
 
+## v0.5.2 context and memory priorities
+
+Development starts on the local `v052` branch with solution design and
+evaluation planning. The implementation sprint follows maintainer approval of
+the solution shape. These priorities describe intended work, not shipped features.
+
+- **Task continuity through compaction.** Let the active agent save a bounded
+  handoff with its next action, preserve it verbatim, and recover interrupted
+  transitions from the session ledger.
+- **Earlier, shared context awareness.** Give the agent and operator the same
+  pressure and budget facts while preserving automatic eviction and overflow
+  recovery.
+- **Clear memory lifetimes.** Distinguish the active working set, session
+  continuity, and reviewed durable lessons; keep provenance and scope explicit.
+- **Measured speed and usefulness.** Prefer reversible eviction, avoid redundant
+  restoration and model calls, and evaluate task completion, context cost,
+  cache effects, and recovery latency together.
+
+Threshold changes and wider memory automation depend on comparative evaluation.
+The first implementation slice is the shared context snapshot and deterministic
+pressure policy, followed by the recoverable handoff transaction.
+
 ## v0.5.0 launch priorities
 
+These priorities are retained as the historical scope of the public launch.
+
 - **A dependable terminal workflow.** Responsive startup, clear tool activity,
   honest progress and timing, useful permissions, and recoverable sessions.
 - **Models chosen by the operator.** Local inference, institutional gateways,
@@ -25,8 +50,8 @@ git history are not current contracts.
 - **A reproducible distribution.** Package contents, installed CLI/browser
   checks, component notices, and qualification of the exact release artifact.
 
-The candidate includes these systems; final publication remains a maintainer
-action after the [release checklist](docs/process/release-cut-checklist.md).
+These systems formed the public launch. Future releases still follow the
+[release checklist](docs/process/release-cut-checklist.md).
 A completed test campaign establishes its recorded coverage, not universal
 correctness across every model, platform, or scientific workload.
 

package/dist/{acp-KS7ARGV6.js → acp-F6PGOFZ4.js}

@@ -1,7 +1,7 @@
 import { createRequire as __clioCreateRequire } from "node:module"; const require = __clioCreateRequire(import.meta.url);
 import {
   runClioCommand
-} from "./chunk-X4CEFOSG.js";
+} from "./chunk-ADO3OQIC.js";
 import {
   restoreStdout,
   takeOverStdout,
@@ -129,4 +129,4 @@ async function runAcpCommand(args, options = {}) {
 export {
   runAcpCommand
 };
-//# sourceMappingURL=acp-KS7ARGV6.js.map
+//# sourceMappingURL=acp-F6PGOFZ4.js.map

package/dist/{agents-YOGJ2YEJ.js → agents-ZHP4R6EX.js}

@@ -1,20 +1,20 @@
 import { createRequire as __clioCreateRequire } from "node:module"; const require = __clioCreateRequire(import.meta.url);
-import "./chunk-4VDHLCF5.js";
+import "./chunk-4WJBWKMI.js";
 import "./chunk-HOYZUKG7.js";
 import "./chunk-QHSBROEW.js";
 import "./chunk-NASYVWDB.js";
-import "./chunk-IN7DGBVS.js";
+import "./chunk-VLX5VZ35.js";
 import {
   ensureClioState
 } from "./chunk-NOCSM22Y.js";
 import "./chunk-KW5HXKLJ.js";
 import "./chunk-URCXX7WJ.js";
 import "./chunk-U26XXPJS.js";
-import "./chunk-T2MDZES3.js";
+import "./chunk-EBMTBE76.js";
 import "./chunk-4A6YRGV4.js";
 import {
   AgentsDomainModule
-} from "./chunk-UCINIRHW.js";
+} from "./chunk-TPCZXWLT.js";
 import "./chunk-GGPF6LLE.js";
 import {
   ConfigDomainModule
@@ -23,7 +23,7 @@ import "./chunk-3UW3KU4A.js";
 import "./chunk-GN7QBEEI.js";
 import {
   SafetyDomainModule
-} from "./chunk-XMDBTYCH.js";
+} from "./chunk-N3CWOGVF.js";
 import "./chunk-TKDRT2DF.js";
 import "./chunk-LQTKDAUO.js";
 import "./chunk-ZU4JPCPR.js";
@@ -31,21 +31,21 @@ import {
   printError
 } from "./chunk-NBXFPI7U.js";
 import "./chunk-SQZHXUB3.js";
-import "./chunk-ZARZ4POR.js";
-import "./chunk-UX2L5CUD.js";
-import "./chunk-LAR7DX5E.js";
+import "./chunk-G5ZDQQE3.js";
+import "./chunk-AI36V3ON.js";
+import "./chunk-B5TASTCK.js";
 import "./chunk-DNCE23JJ.js";
 import "./chunk-CDQ2UC6W.js";
 import "./chunk-5LP5FNNR.js";
 import "./chunk-NVPXXW2H.js";
 import "./chunk-ZNUZ3EXD.js";
-import "./chunk-DOUISWEE.js";
-import "./chunk-VKHLUZNO.js";
+import "./chunk-QOJ4ERMU.js";
+import "./chunk-UZ7YBL43.js";
 import "./chunk-7QXCO3AF.js";
-import "./chunk-T33IYTZM.js";
-import "./chunk-BRCOSE7O.js";
-import "./chunk-DE2LU267.js";
-import "./chunk-733ZBT5P.js";
+import "./chunk-SAASVMIY.js";
+import "./chunk-SZUO5BSE.js";
+import "./chunk-KKAJGT3B.js";
+import "./chunk-E74MJF6G.js";
 import "./chunk-6CTHLZFB.js";
 import "./chunk-2KUBKL63.js";
 import "./chunk-MYI32N4R.js";
@@ -55,23 +55,24 @@ import "./chunk-Z4GLJ6J4.js";
 import "./chunk-VNBICSS2.js";
 import "./chunk-2WVPJF3Z.js";
 import "./chunk-ALJWT6O6.js";
-import "./chunk-I4Y4WKZR.js";
+import "./chunk-F6KCJO3U.js";
 import "./chunk-UZVYEF57.js";
-import "./chunk-EKV4UBCM.js";
+import "./chunk-MMZMM6SW.js";
+import "./chunk-PZBQUJ2F.js";
 import "./chunk-UHW5V7HC.js";
 import {
   isUserVisibleAgent
-} from "./chunk-LNOXCGN2.js";
+} from "./chunk-7UFDIOBT.js";
 import "./chunk-BJH6WFIB.js";
 import "./chunk-5EQO5NWP.js";
-import "./chunk-VYRQHORQ.js";
-import "./chunk-P4GNZPXO.js";
+import "./chunk-NMQSO6Z6.js";
+import "./chunk-OYUS3UZP.js";
 import "./chunk-R2P2U3ZS.js";
 import "./chunk-NTTS3O53.js";
 import "./chunk-C2CPOIH2.js";
 import "./chunk-MGLRTOCK.js";
 import "./chunk-M4KFUXS7.js";
-import "./chunk-V4XJWDHR.js";
+import "./chunk-6UKFRTXI.js";
 import "./chunk-56XP4FJU.js";
 import "./chunk-QZTTRSJE.js";
 import "./chunk-U2MSJ57E.js";
@@ -160,4 +161,4 @@ function renderLine(spec) {
 export {
   runAgentsCommand
 };
-//# sourceMappingURL=agents-YOGJ2YEJ.js.map
+//# sourceMappingURL=agents-ZHP4R6EX.js.map