Release: Self replies reach her Persona; one conversation at a time; resume; project-name addressing; stream thoughts #8
2026-11-07
stream: the proxy-environment test is an integration target, so the lib carries no environment read
Moves the_stream_agent_ignores_the_proxy_environment to crates/mind-body/tests/stream_proxy.rs, as control.rs carries its env-touching tests. It asserts through the public surface: with a proxy in the child's environment the stream's agent reaches its endpoint directly and follows no redirect, and a plain ureq agent in the same environment takes the proxy. No behaviour change. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: the binary's argument reader carries the reasoned allow its siblings carry
mind-stream reads its own arguments like the other binaries, which each allow the crate's disallowed-methods lint with a reason; one range assertion written as contains. No behaviour change. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
release-reply-path: two fixtures reconciled after the merges
A DM's room is its room id under the reply cut (the steward cut's hand-built DM delivery carried the alias form); a request after a result opens a turn under the task cut, so the resume cut's bound test closes that turn with a result. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Merge hands/eureka-body-stream-thoughts (
e2e0488) into eureka-body/release-reply-path
# Conflicts: # crates/mind-body/src/persona.rs
Merge hands/eureka-body-project-name-addresses-steward (
25502a5) into eureka-body/release-reply-path
# Conflicts: # crates/mind-body/src/persona.rs # crates/mind-body/src/persona/tests.rs
Merge hands/eureka-body-self-session-resume (
46800b8) into eureka-body/release-reply-path
# Conflicts: # crates/mind-body/src/self_run.rs # crates/mind-body/src/self_run/decide.rs # crates/mind-body/src/self_run/decide/tests.rs # crates/mind-body/tests/public_surface.txt
self: silent-stop kinds cover every kind the CLI queues nothing for on our stop; capture fixtures killshell and deadline-race; tests kill hand mutants H2 and H5
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: public_surface snapshot carries Usage's kind fields
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a stop of ours owes no turn only for a shell task (task_type local_bash, Bash or Monitor); every other or unread kind owes its turn when stopped
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: idleness reads the CLI's own background-task level and its notification-to-turn rule; at idle the oldest waiting conversation goes first
Cut: State.tasks, Event::TaskStarted, the tool_use_id keying of task lines and the notification replay, and TaskNotified's tasks.remove gate. running is a copy of the CLI's last background_tasks_changed, replaced whole; a notification owes a turn when the task was backgrounded and the session did not stop it with TaskStop (or KillShell, KillBash: the stream shows the alias as the model named it); Polled and Inbox carry other_first, so a served conversation's newer brief no longer outruns an older brief of another. Fixtures bg-fgto, bg-kill and bg-monitor pin the three falsified cases; the probe's fake API gains a generic tool step and a timed scenario. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
probe: background-task captures with the real claude 2.1.312 (bg-fgto, bg-kill, bg-monitor)
A foreground task moved to the background at its timeout, a task stopped with TaskStop, and a persistent Monitor; scrubbed to /probe/work, the key is the fake one. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a session serves one conversation at a time; another conversation's brief is written only at idle, and every result belongs to the conversation being served
decide owns idleness (no turn open, nothing outstanding, no backgrounded task running, no notification owed its turn) and the switch of conversation; read_usage reads the task lines for liveness only; the CLI's replay of a task notification is no echo of ours. SelfInbox serves one Conversation and attributes a result by the span, not by the order of the stream. Deleted: briefless_open, briefless_answers, turn_conversation, turn_allows, may_tell, one_conversation_told, note_task, tasks, notified, the Persona pointer line and its plumbing. ScheduleWakeup and CronCreate are denied to the Self session. Tests replay the real-order captures with briefs spliced in. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
probe: background-task captures with the real claude 2.1.312 (bg-idle, bg-fold, bg-other, bg-race, bg-agent, bg-late)
The harness runs the CLI against the fake Messages API with a fake key and a throwaway config dir: bg and agent plan steps, a Monitor step, --disallowedTools through DENY. Captured: an Agent call with run_in_background gives task_started (is_backgrounded true, the call's tool_use_id) and a task_notification with the same id; a background Bash ending while no turn is open makes the CLI begin its own turn (init, a request, no echo); ScheduleWakeup and CronCreate are absent from the init tools when denied; a Monitor is a backgrounded task with the same two lines, so it needs no denial. Host paths scrubbed to /probe/work. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: LF again; the stale boundary docs name the label as the gap's origin
Restores LF on stream.rs, stream/tests.rs and bin/mind-stream.rs (5defa18 converted them). Renames the capped-repo test for the rule built; build_seed's since is the newest thought's label tick time; STREAM_MIN_GAP_S counts from the label, so a model run of M seconds leaves 900 - M s of quiet after admission. No behaviour change.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a notification is pending by its task's id alone and a turn that took in its conversation's briefs takes it in; tests for the survivors
A notification with no tool_use_id registers nothing (the open-turn gate still holds it by its request). Tests: the notified conversation's brief joins its turn, a notification of another conversation stays pending through a brief turn, two conversations' notifications open a turn no brief may join, an idle non-request line opens no turn, a resolved claim is read by its own key, a claimed cut spec stays on the start line. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: the public surface names STREAM_POLL_S
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a briefless turn is open from a request with nothing echoed until its result; the start message states the routing rules
A request with no line of ours echoed is a turn that took in no brief, notification or not; until its result only the conversation it continues, or the notification's task's, is written. Nothing told yet means no other conversation exists. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self_run tests: the fake claude scripts that loop end when their stdin closes
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona: a reply that cannot be placed is a text-free pointer in the public session
A reply that answers no brief, or briefs of two conversations, was read by nothing and re-paged on every Persona read. It is now a route in the public conversation marked as a pointer: the line carries the reply id and a fixed note, never its text, and the Persona reads the reply by id. Heard, it is resolved like any reply. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a notification's result is its task's turn's; a briefless turn holds other conversations; a resolved claim's conversation is read by key
The stream's task_started and task_notification events carry the tool_use_id of the call that started the task. The actuator records, per task, the briefs the starting turn was answering; a briefless result after a notification answers those (any notification it cannot place answers none). While a notification's turn is open, only the conversation it is attributed to may be written. A restart reads a resolved claim's conversation by key and opens the gate for it alone. The flush result is the note to the successor, in the journal: it is no reply. The start message states the rules. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: soul pass s2 fixes (thinking window, capped per repo, no proxy env, identical lines, forge read at gap marks)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self_run resume: a request with no echo before it is bounded by the result before it (kills the result-line mutant)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: soul pass s1 fixes (stale repo news, userinfo bypass, reads, prompt version, poll cost)
- Repo activity is news only for events after her newest thought, and the notice is stamped with the event time; a capped repo counts only beside such an event. - OllamaStream::open parses the URL as the HTTP client does and refuses userinfo, a path or a query. - A thought's reads cite what the capped context sections kept. - Every thought records prompt_version: Ghostlight revision, word budget, guidance and the continuer frame (thought.v2, not yet deployed). - tick decides due-ness from responsibilities and news before building the seed or reading the persona. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self tests: the gate follows the conversation of a line written and unechoed when a result arrives
Kills the conversation_of mutants (== to !=, body to None) that the previous pass left. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: her own row is in the addressing once, whatever the table holds
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
address, persona tests: no hang on a self-overlapping name; the hash order, the fallback's own row and the poll's table refresh are pinned
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self_run resume: the kept conversation lives in the unit second state directory, outside her working directory; a resumed start consumes its record before spawning; the warm window counts from the line before a request
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: a reply to no brief is told nowhere; the rotation margin ends a start; an idle session tells its own retell, revision or notice; a reply that fails to resolve twice is retried once
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: a result that took in no brief answers nothing once two conversations were told; the flush answers no brief; the start line holds one conversation
A reply to no brief is told in no session and stays in force (ruling briefless-result-routing). A restart drops a claim that is not of the start set conversation. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: the first-line golden follows the
to sentence
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
mind-body: the committed surface follows HeartbeatView and Addressing; the responsibilities list says which chat names are exact; a test's direct message does not name a Mind
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
thought: four candidates are valid, pinned
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stream: the stimulus bound is one cut and the reads follow it; the newest thought is read newest-first
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
heartbeat: the swarm's addressing table in heartbeat.cc, written by the poll; Personas rank the stagger and list
to from it
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
address: a one-word repo name matches as the repo spells it, a split phrase in any case; a self-overlapping name is found; stagger test helpers search a bounded range
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
heartbeat tests: a member naming nothing of hers scores nobody
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
mind-body: the committed surface names address's impl; the responsibilities test sees the chat names
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: the name-ranked event is named, not searched for
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
stewardship names: chat names a steward lists, and the Persona and heartbeat address her by them
address.rs is the one rule for whether a delivery is for a Mind (tag, or a name as a whole word, a listed name exactly as written and a derived or her own in any case); heartbeat's addresses and names_whole_word move into it, the Persona's to_you and own stagger rank call it. eureka-pipeline PipelineStewardship gains names (absent when empty); facts, schemas and golden follow. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-06
stream tests: kill the surviving mutants of cap, bonds, news, unbidden, flood and cadence
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-07
mind-body: Brief.conversation is crate-visible like its type; start_self_run's inputs stay flat
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self: one Self turn answers one conversation
The start set is one conversation's; poll and tell gate briefs by the conversation of those unanswered; a result that took in no brief continues the turn before; the start message says so. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: a reply is told only in its brief's conversation, end to end
DM invariant across three sessions, a reply alone opening its DM once per start, the oldest waiting thing deciding, pre-deploy briefs going public, a two-conversation reply told nowhere, the yield, one origin per turn, the open-brief mark. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona: a reply is told only in its brief's conversation; a start opens a DM's session for a reply alone
reply_routes derives each in-force reply's conversation from the origins of the briefs it answers (nothing stores it); next_session lets the oldest waiting thing decide, carry's batch no longer owns rooms, conversation or origin; an empty batch takes the session's origin; the brief records the turn's origin; a session with nothing of its own yields to another conversation's reply. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
recall: origin_phrase is the one render of an origin
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
pipeline: a follow-up records the origin of the Persona turn that admitted it
PipelineFollowUp gains origin (absent is not written, so the golden sample is unchanged); MemoryOrigin's doc says it is a brief's too; the header facts carry it; schemas regenerated. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-06
persona: once she answered a line, its heard replies never return to waiting; resolution is retried silently
A heard reply whose resolution did not take (admit failed, or the keyed read errs) went back through untell and was told again. The answered replies now wait in answered_replies, retried at each result and at close (settle_replies), one by one so no refusal holds another hostage; untell serves only the lines she did not answer. Tests pin the retry, the close retry, the reply resolved elsewhere on a refused line, and the unreadable in-force read on that path (told again, kept as retell). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
mind-stream: her stream of thought, offline (cut stream-thoughts steps 2-5)
The stream organ projects her own state through Ghostlight's Projector (a pinned git dependency), lets a base model continue it raw over loopback Ollama, keeps the continuation least near her earlier thoughts by the daemon's scores, and admits it as one thought with its whole trace. mind-stream --instance <slug> [--once] [--dry-run]. Cadence is derived from her newest thought; nothing is delivered to any Persona. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
mind-stream: her stream of thought, offline (cut stream-thoughts steps 2-5)
The stream organ projects her own state through Ghostlights Projector
thought kind: her stream of thought is a document (cut stream-thoughts step 1)
PipelineThought epiphany.pipeline.thought.v2, instance-rooted, validated as its chosen candidate; ForeignInstance refusal, Answered-by-FollowUp or Recorded resolutions, facts, a reads citation role, text index. Schemas, golden sample, huginn schema list regenerated. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona: the mind says whether a heard reply took; tests kill the keyless read and the lost resolution
hear_replies attempts each resolution and untell asks the mind which replies are still in force, so an Ok/Err reading of the admission is no second opinion. The out-of-force test holds replies past the first line so a read by kind alone would retell the resolved one. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona: a reply out of force leaves for good; replies resolve one by one; idle and waiting-duplicate paths pinned
untell checks a reply is still in force before it waits again, and hear_replies resolves each reply on its own, so one refused resolution neither retells a reply nobody can answer nor holds the rest of its batch. Tests: out-of-force retell, a reply reaching an idle session with no chat pending, and a reply waiting across a read while a line is not echoed (a stalled child). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona tests: each reply line carries its own words
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
self_run: REPLY_TEXT_LIMIT is the actuator's, not the public surface
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
eureka-state: schema list names reply before selection
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
persona-hears-her-self: her Self admits each result as a reply; her Persona hears, tells and resolves it; the run-end notice, slot read and closure text die
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
eureka-state: schema list counts the reply kind
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
reply kind: one result of her Self session, as its actuator heard it (eureka-pipeline, huginn-mind, schemas, golden)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>