humanish 0.104.3 → 0.106.0

package/docs/architecture/oss-lab-poc.md

@@ -1,250 +0,0 @@
-# Maintainer OSS Meta-Lab
-
-Date: 2026-06-01
-
-Status: implemented as an experimental repo-owned lab manifest plus
-compatibility aliases.
-
-Safety amendment (2026-07-14): beginning with `0.15.1`, the bundled `oss`
-manifest is a contract-only dry-run. A direct live OSS meta-lab request fails
-with `HUMANISH_OSS_META_LIVE_ISOLATION_REQUIRED` before callbacks, filesystem
-or network side effects, credential forwarding, or provider launch. The
-historical design and evidence description below is preserved as a record; it
-is not current execution guidance. The separate `oss-smoke` clone/discard lane
-remains available for public repositories.
-
-## Decision
-
-`humanish/labs/oss.yaml` is this repo's authorized-repo meta-simulation
-dogfood loop. It is intentionally a lab manifest, not the canonical consumer
-shape. Consumer projects should author their own `humanish/labs/*.yaml` files
-and run them with `humanish watch <lab>` or `humanish lab run <lab>`.
-
-The lab should feel like `humanish watch`: it opens the Observer and, for human
-output, keeps the shell attached. Its top-level Observer is an
-Observer-of-Observers: each lane represents a headed E2B desktop assigned to a
-GitHub `owner/repo` slug. Inside each desktop, the bootstrap clones the repo,
-gets it into local dev mode where feasible, installs and initializes Humanish,
-runs nested Humanish proof commands, starts the target app when a runnable
-script is present, opens desktop/mobile app windows plus the nested Observer in
-the E2B browser, and starts a nonblocking Codex actor attempt.
-
-The previous clone/discard proof loop remains useful, but it is now explicitly
-named `humanish/labs/oss-smoke.yaml`.
-
-## Commands
-
-Main operator path:
-
-```bash
-humanish watch oss
-humanish lab run oss
-humanish lab run oss --repos CorentinTh/it-tools,drawdb-io/drawdb,maciekt07/TodoApp,lissy93/dashy
-humanish lab run oss --repo CorentinTh/it-tools --repo drawdb-io/drawdb --count 4
-```
-
-Agent/CI contract path:
-
-```bash
-humanish lab run oss --dry-run --json --no-open
-```
-
-Disposable clone smoke path:
-
-```bash
-humanish lab run oss-smoke
-humanish lab run oss-smoke --limit 1 --keep
-```
-
-Local dogfood shortcuts:
-
-```bash
-pnpm humanish:lab:oss
-pnpm humanish:lab:oss:ci
-pnpm humanish:lab:oss:smoke
-```
-
-## Repo Selection
-
-The default public targets are product-like, locally runnable apps/tools:
-
-- `CorentinTh/it-tools`
-- `drawdb-io/drawdb`
-- `maciekt07/TodoApp`
-- `lissy93/dashy`
-
-Target selection should minimize distance from a real user-facing journey.
-Good defaults expose an app, CLI, or agent-facing tool that can be tried out of
-the box, preferably with a local dev script and no account setup. Libraries,
-frameworks, starters, and infrastructure packages belong only in scenarios that
-explicitly test developer experience. They are poor defaults for proving
-Humanish as a user-simulation harness because they add another abstract setup
-layer before any product behavior is visible.
-
-`--repos` accepts a comma-separated list. Repeated `--repo` is also supported.
-If `--count` is larger than the repo list, assignments cycle through the repo
-pool. Inputs must be GitHub `owner/repo` slugs. Arbitrary URLs, local paths,
-tokens, and SSH remotes are rejected. Private repos are maintainer-only and
-require an authorized `GH_TOKEN` or `GITHUB_TOKEN` at runtime; no token value is
-written to committed source or public issue text.
-
-## Private Product Labs
-
-Private products can be used for local maintainer dogfood, but they must stay
-out of the public package surface. Do not commit private repo names as defaults,
-fixtures, screenshots, README examples, npm assets, skill examples, or issue
-draft text.
-
-The safe local shape is:
-
-```bash
-humanish watch .humanish/labs/private-app.yaml --env-file .humanish/local/provider.env
-```
-
-with an authorized runtime GitHub token and default repo redaction enabled.
-Do not pass `--no-redact-repos` for a private target. Public receipts for
-those runs should say `authorized private app target` and point only to ignored
-local artifact paths, redacted statuses, and verifier results. Never publish
-private screenshots, logs, app URLs, source snippets, branch names, issue
-names, stream URLs, or operational details.
-
-## Runtime Shape
-
-The meta-lab writes ignored local Observer evidence:
-
-```text
-.humanish/
-  runs/<oss-meta-run-id>/
-    run.json
-    review.json
-    review.md
-    events.ndjson
-    observer/
-      index.html
-      observer-data.json
-```
-
-Each stream lane records:
-
-- assigned repo slug for public runs, or a redacted lane label for token-backed
-  maintainer/private runs;
-- whether a live E2B desktop stream exists; auth-bearing stream URLs are
-  runtime-only for the attached Observer server and are not persisted;
-- target app URL/status when a runnable script becomes HTTP-ready inside the
-  sandbox;
-- nested Observer presence and nested verification status;
-- headed desktop visual-window status and browser window count;
-- Codex actor status, optionally moved before deterministic setup with
-  `HUMANISH_OSS_META_ACTOR_FIRST=1` and required through
-  `HUMANISH_OSS_META_REQUIRE_ACTOR=1`;
-- setup-quality filesystem evidence: shallow tree, Humanish setup checks,
-  package scripts, study-quality rating, and allowlisted previews for public
-  runs;
-- public-safe remote bootstrap log tail;
-- public-safe gaps and events.
-
-The current implementation launches live E2B desktop streams when
-`E2B_API_KEY` and `OPENAI_API_KEY` are present, overlays those stream URLs only
-in the attached Observer server, and marks missing key or launch failures
-in-lane. It also packs the local Humanish package, uploads it into each sandbox,
-raises a visible bootstrap terminal, clones the assigned repo, runs nested
-Humanish setup and proof commands, starts the target app, opens desktop/mobile
-app windows, opens the nested Observer in the sandbox browser, arranges visible
-browser windows for screenshot proof, and starts the Codex actor attempt. By
-default the actor remains nonblocking and runs after deterministic readback; with
-`HUMANISH_OSS_META_ACTOR_FIRST=1`, the actor attempts setup/use before
-deterministic validation; with `HUMANISH_OSS_META_REQUIRE_ACTOR=1`, the bootstrap
-waits up to `HUMANISH_OSS_META_ACTOR_TIMEOUT_MS` for terminal actor readback and
-will not mark the lane passed unless the actor exits cleanly.
-
-When the remote bootstrap completes, the host persists a local
-`setup-quality/<stream>-setup-quality.json` artifact. The Observer Files tab can
-render it inline from the served Observer. Static `file://` observers keep the
-artifact link openable but do not hydrate it inline. Token-backed/private runs
-suppress raw file previews by default while preserving the setup checks and
-tree shape. They also preserve `studyQuality` structural signals so an actor
-that merely installs Humanish receives a `ceremonial` rating instead of being
-treated as successful user-study leverage.
-
-The live desktop substrate is an optional peer dependency. Install it in the
-project that runs live labs:
-
-```bash
-npm i -D @e2b/desktop
-```
-
-Remaining substrate work: upgrade the nested `--app-url` browser proof into
-provider-backed personas that actually drive multi-step target-app journeys,
-and live-prove actor-first setup/use, not only deterministic bootstrap readback.
-
-## Smoke Harness Runtime
-
-`humanish lab run oss-smoke` shallow clones lightweight public GitHub repos into
-ignored runtime state, applies Humanish setup inside each throwaway clone, runs
-the synthetic four-lane proof path, verifies the generated bundle, records
-git-status evidence, writes an ignored report, and removes cloned repos by
-default.
-
-```text
-.humanish/
-  lab/oss/<run-id>/
-    report.json
-    report.md
-  tmp/oss-lab/<run-id>/
-    repos...      # removed by default
-```
-
-Each cloned repo receives disposable uncommitted changes:
-
-- `humanish/` source starter files;
-- `.humanish/` runtime state;
-- `.gitignore` updates;
-- `package.json` script updates;
-- synthetic run/Observer evidence under the clone's ignored `.humanish/`.
-
-The clone is removed unless `--keep` is passed. The host report remains under
-ignored `.humanish/lab/oss/<run-id>/`.
-
-## Safety Rules
-
-- GitHub `owner/repo` slugs only.
-- Private repos require an authorized runtime token. Token-backed runs redact
-  repo labels in durable artifacts by default and must not appear in committed
-  fixtures, docs examples, public issue text, or published media.
-- pnpm dependency build scripts may be allowed only inside the disposable E2B
-  lab so target app surfaces can start. Never use this as a host install
-  default.
-- No credential prompts; smoke clone calls set `GIT_TERMINAL_PROMPT=0`.
-- No commits, pushes, branches, tags, GitHub API mutation, deploys, or issue
-  filing.
-- Do not write key values into committed `humanish/` source.
-- Do not emit PII, PHI, raw private transcripts, private screenshots, secrets,
-  keys, or private upstream artifacts.
-
-## What This Proves
-
-The meta-lab proves the operator control surface and artifact contract for
-watching multiple Codex/E2B setup attempts at once. The live path now detects
-E2B desktop fanout, visible bootstrap terminals, local-package upload,
-disposable authorized-repo setup, target app HTTP readiness when a runnable
-script is present, headed browser-window layout, nested live `--app-url`
-desktop/mobile browser proof when the target app runs, nested Observer opening,
-and top-level lane completion from remote evidence. It does not yet prove a
-general provider-backed multi-step target-app persona runtime.
-
-The smoke harness proves first-run Humanish package compatibility against
-arbitrary public JavaScript repositories:
-
-- setup can patch real package.json files without committing;
-- generated `humanish/` source can coexist with external repo layout;
-- ignored `.humanish/` runtime proof can be generated;
-- the Observer can render from those disposable proofs;
-- verification passes before the clone is discarded.
-
-Neither path may claim private product behavior proof without live, redacted,
-public-safe evidence.
-
-Feedback candidates are derived from this evidence when a lane records concrete
-setup-quality gaps, ceremonial/absent study quality, or actor-reported Humanish
-CLI blockers. `humanish feedback` uses those candidates before falling back to
-generic dry-run follow-up drafts.

package/docs/architecture/runtime-broker-core.md

@@ -1,30 +0,0 @@
-# Runtime broker authority core
-
-The source-only [broker core](https://github.com/danielgwilson/humanish/blob/main/runtime/broker/README.md)
-defines bounded requests and lease decisions for a future Linux runtime owner.
-It is not an installed service, resource controller, or public runtime mode.
-
-Requests cannot supply host paths, service names, commands, user identities,
-resource limits or cleanup claims. The future transport obtains the peer UID
-from the operating system and checks it against owner policy. A separate
-generation capability authorizes each study; default diagnostics and serialized
-ledgers omit the raw capability. Repeating an acquisition or allocation attempt
-does not mint a second resource.
-
-The model reserves capacity through uncertain creation and cleanup. Expiry,
-revocation, detected sleep and host-clock discontinuity cannot be undone by a
-late renewal. A saved launch plan is not dispatch authority: the owner must
-revalidate it immediately before starting a resource. Only a separate owner
-attestation that creation has settled and the acquired resources are absent can
-free a reservation. Recovery reads the ledger for reconciliation, never to resume
-a study or adopt resources by their names.
-
-The owner must persist intent before consuming effects, advance trusted time
-independently of requests, and bind every resource to actual acquired OS
-identities. The pure model does none of those OS operations. Its tests establish
-protocol and state transitions; they do not establish peer authentication,
-durable writes, process containment, watchdog timing, sleep detection or cleanup.
-
-Installation, root-owned storage, authenticated sockets, service supervision,
-artifact verification and VM/network control require separate implementation
-and privileged qualification. The existing npm execution routes are unchanged.

package/docs/release/0.100.0-local-setup.md

@@ -1,31 +0,0 @@
-# 0.100.0 — Local study setup and startup reliability
-
-Initialize a local browser study for your app without writing its lab manifest:
-
-```sh
-humanish init --yes --local-browser http://127.0.0.1:3000 \
-  --local-mission "Create a note and find it again"
-humanish doctor --lab local-browser
-humanish run local-browser
-```
-
-The starter uses local Firecracker browsers and the supported Codex account
-login. Model inference remains remote and consumes account quota; E2B and OpenAI
-API keys are unnecessary for this route. Linux needs local Docker/KVM/TUN;
-supported Apple Silicon Macs use Lima. Init and the TUI explain prerequisites
-and reuse the CLI's readiness checks. Existing lab files are preserved; if the
-local starter already exists, edit its URL and mission in
-`humanish/labs/local-browser.yaml` instead of rerunning the init flags.
-The free preview and hosted starter remain available. Local dry runs work without
-starting a desktop or requesting a model turn.
-
-The first local participant capture now follows the target document's
-DOMContentLoaded event and a paint. Previously a fixed 500 ms wait could expose
-the runtime's starter page. Navigation is bounded, supports redirects, and
-reports startup failures; it does not wait for all app data or subsequent loads.
-Updated x64 and ARM64 runtime images carry the initial-navigation change.
-
-Failed account participant requests retain their finite execution stage, such
-as `thread/start`, in the recording and outcome text. Older recordings remain
-readable. This adds diagnostic evidence without changing retry or deadline
-behavior; it does not claim the cause of a past timeout is known.

package/docs/release/0.100.1-participant-continuity.md

@@ -1,20 +0,0 @@
-# 0.100.1 — Continuing participant conversations
-
-Local Codex participants now keep one conversation for their entire study,
-including closing feedback. Previously, each screenshot started a new thread
-with only eight turns of summarized history. Earlier observations, actions and
-participant context now remain in the Codex conversation, with Codex managing
-context compaction.
-
-Each participant has its own process and thread. The existing browser executor,
-recordings and separate post-run analyst are preserved. Request deadlines still
-apply per turn; thread-cumulative token usage is converted to per-turn receipts.
-Codex omits compaction requests from its thread totals, so turns that compact
-retain known counts and report incomplete usage.
-An interrupted or failed participant closes its session rather than silently
-restarting with empty memory. Dollar usage remains unknown for account-backed
-inference.
-
-Existing E2B/API and hosted Codex participants already continued their
-conversations and retain those paths. Older Observer recordings remain readable.
-No setup changes or Firecracker image downloads are required for this fix.

package/docs/release/0.101.0-codex-ui-tools.md

@@ -1,28 +0,0 @@
-# 0.101.0 — Shared Codex UI tools
-
-Codex participants now interact through a native Humanish UI tool. One continuing
-Codex conversation can call the tool repeatedly, inspect each new screenshot,
-and give its final feedback. Humanish executes and records the inputs, including
-rejected or skipped actions, before returning their actual status to Codex.
-
-Local Firecracker and hosted E2B Codex participants use the same implementation.
-The duplicate hosted adapter is removed. Hosted studies retain their operator
-login, configured model selection and explicit reasoning effort; local account
-studies retain their qualified isolated login profile. Codex runs on the host,
-with desktop inputs sent to the selected sandbox. No credentials are moved into
-the desktop and no new Firecracker image is required.
-
-The native integration uses Codex CLI **0.154.0** and its experimental dynamic
-tool protocol. Other CLI versions fail with an unsupported-version diagnostic.
-Update to the supported version when moving a hosted study from an older CLI.
-
-The existing loop still owns recordings, action limits, stop conditions and
-cancellation. Pending tool calls are not counted as finished model requests.
-Known pending usage informs spend estimates; final usage is recorded once when
-the native turn settles. Per-inference token counts preserve API pricing tiers
-inside a continuing conversation. Interrupted or compacted requests retain explicit usage
-uncertainty. Account inference has unknown dollar cost.
-
-Automatic analysis remains a separate, restricted analyst. The OpenAI Responses
-computer-use provider and Claude Code conversation adapter remain supported.
-Older recordings remain readable in Observer.

package/docs/release/0.102.0-participant-media.md

@@ -1,43 +0,0 @@
-# 0.102.0 — Optional participant camera and speech
-
-Codex participants can listen and speak while using a conferencing app. The same
-continuing conversation handles browser actions and spoken replies on local
-Firecracker and hosted E2B desktops. No separate conversational agent or speech
-API key is required. Codex inference still uses the selected remote account.
-
-Declare `execution.desktop.media.microphone.source: speech` for speech, and
-`camera.source: synthetic` for a changing test-pattern camera. Local desktops
-support both through native browser devices and ordinary permissions. Humanish
-downloads the optional media runtime when needed; `humanish runtime setup --media`
-can prepare it ahead of a study. Browser-only studies retain the smaller runtime
-and start no media processes.
-
-Hosted speech automatically selects a public E2B template and uses the same
-worker. E2B's existing fake-camera flags also replace its microphone, so hosted
-camera and speech cannot be combined. Explicit compatible template overrides
-remain supported. Hosted speech requires `@e2b/desktop` 2.3.2 or newer for the
-background command handle's streaming stdin methods.
-
-Recordings and analysis retain spoken text and transcriptions from actual speaker
-audio under the existing redaction rules. Observer links findings to those
-entries and nearby screenshots. Raw audio/video is not recorded. The initial
-voice is robotic, recognition is English, and clarification may be necessary;
-this is turn-taking rather than interruptible realtime voice conversation.
-
-Linux acceptance includes two concurrent participants entering a real conference
-through different application roles, receiving counterpart video, and exchanging
-previously undisclosed information by voice. Automatic analysis and owned-resource
-cleanup completed. The Whisper `base.en` runtime also passed a Mac
-conversation, including reciprocal video, a fresh fact and question, and spoken
-correction and confirmation of a misheard answer. Recognition still makes
-meaningful errors; this pass depended on clarification. Analysis of the retained
-Mac recording validated both participant reviews and a shared finding about the
-corrected misunderstanding, with all selected captures and earlier attempts
-preserved.
-
-Analysis now retains specific, safe validation failure codes. Its citation
-instructions distinguish each participant's own evidence from cross-participant
-observations in findings. The evidence validator remains unchanged.
-
-See [participant media](../architecture/participant-media.md) for configuration and
-the [hosted template recipe](../../runtime/browser-media/E2B.md) for maintenance.

package/docs/release/0.103.0-desktop-recording.md

@@ -1,29 +0,0 @@
-# 0.103.0 — Optional desktop video and audio
-
-Computer-use studies can retain an MP4 alongside screenshots, actions,
-participant feedback and analysis:
-
-```yaml
-execution:
-  desktop:
-    recording: { audio: true }
-```
-
-Set `audio: false` for screen-only video. Omit `recording` for the existing
-lightweight captures. Recording does not enable camera or speech features.
-
-Observer plays the selected participant's video on the existing study timeline.
-Opening a participant, scrubbing and returning to the grid preserve the cursor.
-Audio starts muted; the selected participant has an explicit audio control.
-Unavailable video falls back to screenshots where they exist.
-
-Local Firecracker and hosted E2B use the same capture settings and evidence
-format. Local capture uses the optional media runtime archive. Recorder failure
-preserves the study; a playable interrupted or size-limited recording is marked
-partial. The per-participant file limit is 128 MiB.
-
-Continuous media stays local-only. Screenshot redaction does not redact video
-or audio. Portable HTML exports explicitly omit continuous media; analysis
-continues to inspect screenshots and text. See the
-[recording guide](../architecture/desktop-recording.md) for supported routes,
-capture semantics and export behavior.

package/docs/release/0.103.1-analysis-capture-allocation.md

@@ -1,18 +0,0 @@
-# 0.103.1 — Fairer analysis capture allocation
-
-Analysis now gives reclaimed screenshot slots to eligible participants with
-fewer admitted images. Previously, a participant whose images exceeded its
-initial byte reservation could receive only its ending capture while other
-participants received extra captures, even with enough image bytes remaining.
-
-The fix preserves the 40-capture cap, byte and read limits, source identities,
-and within-session selection order. Existing recordings and analyses remain
-unchanged; a new analysis uses the corrected allocation. Completion still
-requires affirmative evidence for every essential assignment requirement.
-
-A synthetic 48-capture regression reproduces the missed beginning frame, and
-selection on a retained study confirms more balanced participant coverage.
-These checks establish allocation behavior, not improved model interpretation
-or guaranteed selection of every decisive moment. The original outcome-evidence
-incident in [#851](https://github.com/danielgwilson/humanish/issues/851) remains
-under investigation; this release does not claim to resolve it.

package/docs/release/0.103.2-analysis-ending-context.md

@@ -1,21 +0,0 @@
-# 0.103.2 — Adjacent ending evidence for analysis
-
-A participant can inspect a result and then scroll or navigate before ending.
-Analysis now prioritizes the preceding screenshot alongside the final view,
-after beginning and explicit failure context. Previously, session sampling
-could omit that nearby result while retaining a final heading or navigation page.
-
-The 40-capture cap, byte and read limits, and completion requirements are
-unchanged. Selection uses source order, not success declarations. Existing
-recordings and analysis attempts remain unchanged; reanalysis creates a new
-attempt using the revised selection.
-
-In a retained two-participant case, the omitted views showed the required final
-rows. Paired replays with the same current analyst prompt and transport changed
-both outcomes from unknown to completed with supporting visual citations.
-An authored 90-capture control retained a supported local-save success while
-leaving contradicted and unverified submission claims unknown.
-This addresses the observed omission in
-[#851](https://github.com/danielgwilson/humanish/issues/851); it does not guarantee
-selection of every decisive moment. Reports still disclose partial coverage,
-and an unverified essential requirement still warrants an unknown outcome.

package/docs/release/0.104.0-local-captured-inboxes.md

@@ -1,29 +0,0 @@
-# 0.104.0 — Local captured inboxes
-
-Local browser participants can now create an account, open their assigned inbox
-and follow a verification link without mailbox-provider credentials. Start
-`humanish comms catch`, configure the app to send email to that catch, and declare
-`comms.email.external.catchBaseUrl` in the local lab. Doctor checks the catch's
-recipient routes and explains setup failures before a participant starts.
-
-Each desktop forwards only its assigned recipient's read-only inbox. Operator
-mail, other recipient routes and the private delivery drain are unavailable
-through that forward. The existing renderer, mission instructions, shared study
-runner and digest-only comms evidence are reused. Supported inline raster images
-render; unsupported images retain an explicit placeholder.
-
-This captures the target app's outgoing mail. It does not receive arbitrary
-internet email or add real-provider receiving to local-agent studies. Existing
-hosted receiving and its management-credential boundary remain unchanged.
-
-Verification includes two concurrent Linux local Codex participants with distinct
-messages and independently verified app accounts, automatic analysis, supported
-and unsupported image rendering, a browser-runtime isolation/cleanup check, and
-real OpenSSH forwarding with controller-death cleanup. Native ARM runtime builds
-exercise the same recipe; this release does not claim a new Mac end-to-end run.
-Raw study screenshots remain local-only under the normal verification rules.
-
-The capability table and older media/account/timeout issues were also reconciled
-with shipped behavior. Token counts do not establish plan quota or dollar cost;
-the original intermittent pre-profile timeout remains unresolved. Recording
-startup gaps remain tracked separately in #854.

package/docs/release/0.104.1-recording-startup.md

@@ -1,24 +0,0 @@
-# 0.104.1 — Recording startup
-
-Mixed desktop recording no longer waits behind the recording mix's default
-two-second PulseAudio buffer. Its null sink now uses the supported no-rewind mode,
-which bounds that buffer to 50 ms. FFmpeg settings, timestamps, startup ordering
-and resource ownership are unchanged; screen-only recording uses the same path
-as before.
-
-The reproduced Linux VM baseline had a 1.67-second gap after its first frame.
-The corrected recording began video at 0.187 seconds and audio at 0.244 seconds,
-with continuous frame cadence. Six browser flashes and tones were retained and
-aligned within one 15 fps video frame. Actual E2B mixed and screen-only captures
-also had continuous video cadence. These are retained checks, not universal
-startup-latency guarantees or a new Mac acceptance study. Longer installed checks
-also retained variable guest capture gaps, including a 1.33-second mid-recording
-stall. Those gaps remain tracked in #854; this correction addresses the cold mix
-buffer.
-
-Device and transport startup still precede capture. Cold audio sources may deliver
-sound later than the playback request; early silent audio is not proof that every
-requested sound was captured. Original recordings are preserved, with no fabricated
-frames or timestamp repair. Interrupted recording still produces partial metadata
-and releases its recorder and audio processes. Screenshots remain the default
-and the fallback if optional recording fails.

package/docs/release/0.104.2-recording-timestamps.md

@@ -1,32 +0,0 @@
-# 0.104.2 — Recording timestamp precision
-
-Desktop video encoding now uses the microsecond time base supplied by capture.
-The previous encoder default rounded timestamps to the nominal 15 fps grid and
-dropped some distinct captures taken close together after an input stall. The
-shared local/E2B recipe specifies the encoder time base and variable-frame-rate
-output; the capture rate, audio
-settings, resource limits and metadata clock remain unchanged.
-
-In a controlled one-second capture interruption, the previous recipe received
-96 captures and retained 82 output frames. The correction retained all 96 with
-timestamps within 1 ms of their input times, while preserving the missing interval.
-The H.264 profile and level remained unchanged. Explicit variable-frame-rate
-output also prevents older FFmpeg defaults from synthesizing duplicate frames
-to fill a gap. No original recording is rewritten.
-
-This is a timestamp-fidelity correction, not a fix for every guest execution stall.
-The #854 follow-up reproduced browser pauses and guest CPU contention, without
-establishing one offending process. A recorded frame can correctly show an
-unchanged screen during an application pause. Separate follow-up recordings
-retained every paired event but measured audio 64–147 ms before visible flashes;
-source delivery and recording precision must be distinguished before attributing
-that offset. The earlier one-frame timing result was a single-run measurement,
-not a universal guarantee. #854 remains open for those limits.
-
-Final local interruption controls retained 94/94 screen-only and 96/96 mixed
-input frames, with no output frames inside the interruption. E2B’s FFmpeg 4.4
-retained 106/106 screen-only and 105/105 mixed input frames, each within 1 ms
-of input timestamps. Browser decoding and seeking passed, and externally stopped
-capture still returned partial metadata with recorder/audio cleanup. A local VM
-retained ten complete paired events through 60 seconds, within one video frame
-in that run; this does not close the separate intermittent-stall investigation.

package/docs/release/0.104.3-local-smtp-catch.md

@@ -1,15 +0,0 @@
-# 0.104.3 — SMTP capture for local app studies
-
-`humanish comms catch --smtp-port 1025` now opens the documented loopback
-SMTP listener. Previously, the host CLI rejected the option even though the
-shared catch already supported SMTP. Apps that send verification mail over SMTP
-can use the same recipient inbox and evidence pipeline as HTTP-based apps.
-
-The optional SMTP port must be valid and distinct from the HTTP capture and
-read-only inbox ports. A busy SMTP port now stops startup before HTTP health
-can report success; previously its background thread failed while HTTP stayed
-healthy. SMTP remains bound to loopback, and omitting the option preserves
-HTTP-only capture. No mailbox provider or runtime image change is needed.
-
-See [the inbox setup guide](../architecture/comms-inbox.md) for app and lab
-configuration. Fixes #878.

package/docs/release/0.20.0-external-public-shared-world.md

@@ -1,43 +0,0 @@
-# 0.20.0 — External-public shared-world plane + CDP lobby-code handoff
-
-Additive minor release. No breaking change to the provisioned-getHost path, its schema, or its verify
-asserts (byte-stable; a snapshot regression pins it).
-
-## Highlights
-
-- **External-public shared-world plane class** (`sharedWorld.planeClass: "external-public"`). A real,
-  operator-OWNED public deployment can now be the shared plane DIRECTLY: `subject.source: app-url` +
-  `subject.topology: shared-world` + `policies.allowPublicTargets: true` +
-  `execution.concurrency > 1`. NO getHost, NO clone, NO subject sandbox, NO seed. The N>1 +
-  `allowPublicTargets` refusal now ROUTES here when `topology: shared-world` is declared, instead of
-  failing.
-- **Operator-ownership attestation** `subject.publicTarget: { owner, authorized: true }` — the honest
-  analog of `exposure: synthetic` for a plane the harness does not own (author-trust, unverifiable).
-- **CDP lobby-code handoff barrier.** A new runtime-only `CuaLoopOptions.onObservedUrl(url)` callback
-  (fired after every `executor.observe()`, never persisted) feeds a host-first barrier: the host seat
-  creates the shared lobby, the orchestrator extracts `/lobby/CODE` from its CDP-observed URL and
-  threads the code into the follower JOIN missions; followers fail closed WITHOUT opening on
-  `HUMANISH_CONCURRENT_SHARED_WORLD_LAB_HANDOFF_TIMEOUT`.
-- **3-mobile lobby-trivia example** (`humanish/labs/lobby-trivia-3player.yaml` + an init template): 3
-  mobile-LAYOUT personas play the same multiplayer lobby on a real public app at once. Dry-run $0.
-- **Honest evidence class.** Attribution stays `shared-world`; every strength claim degrades honestly
-  and is asserted-ABSENT by verify: `subject.state.provenance == external-public` (not seeded);
-  `plane.exposure` ABSENT (no synthetic claim on a real site); `plane.publicOriginDigest` is the
-  OBSERVED origin the seats converged on (redirect-tolerant — a cross-origin apex→www/http→https
-  redirect is normal; the declared origin is recorded separately as `plane.declaredOriginDigest` and
-  is never asserted equal) with every observed `routeHostDigest` converging on it (operator-attested,
-  not harness-controlled); NO
-  authoritative `stateSeries` (Option A); relaxed concurrency-on-pass (temporal co-occupancy only);
-  optional `lobbyConvergenceDigest`. The getHost synthetic gate stays UNCHANGED and is deliberately
-  unreachable from the app-url branch.
-
-## Watch-from-phone
-
-`humanish serve --expose` covers the evolving Observer/artifacts from a phone TODAY. Native
-live-desktop `--expose` streaming on the concurrent path is a clean 0.20.1 fast-follow reusing
-`startExposedObserver` on the concurrent observer server.
-
-## Notes
-
-- Cost tracking (0.19.0) and the CommandExitError resilience fix (0.19.1) remain in place.
-- `execution.timeoutMs` caps the host-first handoff deadline (default 120s).