Repository navigation
Releases
oars 0.5.5
- Wire format 7 and field list 14 are unchanged from 0.5.4 (
CAPTURE_VERSION35); the pool reads formats 3 to 7. oars pool deposit --send-allsends, in order and one batch at a time, every batch thatoars pool outboxprinted and that is unchanged (the covenant rule of--send; a batch never printed or changed is skipped and counted; not with--send). A failed batch stops the run with the sent, skipped, failed, and left counts and a non-zero exit, a 429 or a 5xx is tried up to three times, and running it again sends only whatsent.logdoes not hold.oars pool depositbuilds the outbox a hundred sessions at a time, andoars pool outboxand--send-allread one batch at a time, so a history of thousands of sessions builds and sends in a small heap (a 7,000-session build under a 160 MB setting; thesessions depositcapture still reads the whole store once at its end, andoars pool outbox --jsonstill builds one document, so it is not memory-bounded).oars pool joinis now the whole first deposit in one command: sign-in, the region question, an exclusion term file made for you to edit, the batch, what it holds, the covenant screen with the phrase you type yourself, and the first send. A second run continues at the first step not done,--harness-versionis optional (given, it fetches after the send), and--plan --jsonlists each step, its state, andneeds_personfor the covenant phrase and the raw prompt, and for the region question and the term file while unanswered, changing nothing. Each step leaves ajoin-steptelemetry event of closed words (docs/telemetry.md).oars card,oars club show,oars board, and the Friday line now point the ashore tooars pool join.oars pool invitemakes one vouch code (a founding or standing account, three open at most) and prints it once with the join line;oars pool join --invite <code>redeems it after sign-in;oars boardandoars club showprint a founding member's crew number beside the nickname, and the share card's depositor line addsfounding crew #Nonceinviteorjoinhas learned it.oars pool depositreads each chunk of sessions by the byte offsets its one scan of the log recorded, instead of streaming every log line once per chunk: the 7,000-session build ran in 219 s against 261 s, with the same batches. A row appended to the log during the run is never read through the scan, and if the log is replaced during the run (a compaction), the build reads it by streaming, as before, so nothing is skipped.- Internal, no behavior change:
oars sessions refresh(and the refresh a deposit ends with) now scans the log once, then reads and folds the store a hundred sessions at a time, so it no longer holds every tool call and prompt of the history: over 7,000 sessions it finishes in 28 s under a 256 MB heap setting (it ran out of memory at 160 MB), and every derived file is byte-identical at any run size. - Internal, no behavior change: the
refreshderivations (cells, increments, fit reads, consequences, and reversions) now fold run by run over a log scan that keeps each session's parents and capture versions, with arefreshpintest that pins every derived file's bytes;oars sessions refreshoutput is byte-identical. - The weekly
harness-readersworkflow now fails when a harness's latest npm release is past the version its fixture was modeled on (versions.json, default a different minor), and a localshape.mjstool lists the field names a real session file has that a fixture lacks, with no values. The Claude Code and Codex fixtures are refreshed to 2.1.322 and 0.167.0, including a Codex code-mode cell.
oars 0.5.4
- Wire format 7 and field list 14 are unchanged from 0.5.3 (
CAPTURE_VERSION35); the pool reads formats 3 to 7. oars pool jointakes you from nothing to a first deposit and your first fetch in one command: it runslogin,deposit,outbox --explain,deposit --sendfor each batch not yet sent, andfetchin order, shows what each step does, and when a step stops, running it again continues from there (a held sign-in, an unsent batch, and a sent batch are not done twice). It needs--harness-version, asfetchdoes, and with--jsonit explains on stderr before the first send.oars pool outboxandoars pool outbox --explainno longer fail with "Invalid string length" on a large outbox (about 6,500 sessions): they print one batch at a time, and--jsonwrites the batches one at a time when one string would be too long.oars pool deposit --sendended by Ctrl-C now exits 130, the shell convention for an interrupt; the text is unchanged, and any other way of stopping keeps its exit code.oars pool remove <hash>removes one batch and its queued raw copies from this machine's outbox and keeps that session out of every later deposit, the background sender's included, until--restore; nothing is sent, and it refuses while a send holds the lock.oars pool deposit --sendnow says the pool may already have received the batch when the connection drops after the request went out, and marks nothing sent.
oars 0.5.3
- Wire format 7 and field list 14 are unchanged from 0.5.2 (
CAPTURE_VERSION35); the pool reads formats 3 to 7. oars pool deposit --sendnow takes over a send lock whose holder is no longer running at once (it used to wait ten minutes), says within a second who holds a live lock and how long it will wait, and ends with a refusal that says when a retry can succeed (as soon as the holder ends, for a live holder on this machine). It printsSending batch <hash> to <server>...when a send starts, ends a pool that does not answer within 30 seconds with a plain sentence that nothing was marked sent (and, once the request carrying the batch had gone out, that the pool may already have received it), states a dropped connection in plain words with the raw reason in parentheses, and stops waiting or sending when you press Ctrl-C, close the terminal, or the program that started it is killed, then says the batch stays queued (and, once the request had gone out, that the pool may already have received it), that the background sender sends it once the covenant is accepted, and how to keep a folder out or stop this machine sending. The background sender gives up on a pool that does not answer within 30 seconds too, so it no longer holds the send lock while the pool stalls.oars pool depositnow says that nothing was sent yet and that oars sends what is queued automatically once the covenant is accepted. Exit codes are unchanged.oars pool deposit --sendprints three short lines on stderr before the covenant screen (what a deposit gives, the weekly fetch limit a counting deposit lifts for 90 days, and a pointer to the full text) and two after a successful acceptance (sending starts, andoars pool fetch --helpfor a first download). They sit outside the screen: the covenant text, its id2.1-ec0ca3f6ff73, and the hash recorded at acceptance are unchanged, and nothing prints on a wrong phrase, a refusal, or a failed post.- The copyright holder in the LICENSE files is now Many Oars Co., which received the project from Switchyard Software Company by written assignment effective 2026-11-11. The license terms are unchanged.
oars 0.5.2
- Wire format 7 and field list 14 are unchanged from 0.5.0 (
CAPTURE_VERSION35); the pool reads formats 3 to 7. - Client deposits now redact at version 4 (
REDACTION_VERSION; a queued version 3 raw copy is stale and is not sent): system-role request blobs and memory-file text get the injected-block marks, synthetic provider token prefixes are masked, and token counts, scoped npm names (@scope/name), and1password:are no longer masked.docs/sessions-store.mdlists the history. oars --helpnamesoars sessions tryon thesessionsline, the empty-store message saysoars sessions depositreads your own session history, and@manyoars/oars-clientnow declares the CLI's Node floor (22.23.4), so installing the CLI on Node 22 no longer prints engine warnings (with @manyoars/oars-client 0.1.2).- Each release now lists
io.github.manyoars/oarson the MCP registry from GitHub Actions over OIDC (release-mcp.yml), with no stored token; a dispatch can list an existing tag. oars pool loginsurvives a failed poll: a network error or a 5xx waits twice as long (up to 60 seconds) and polls again until the device code expires,slow_downstill adds 5 seconds, andaccess_deniedandexpired_tokenstill end it at once; exit 8 comes only once the code has expired with the last poll still failing.oars pool depositjudges a session ended by its transcript's own last line, not by when it was captured, so a store captured a minute ago from days-old transcripts builds its outbox at once; a session whose last line is under 6 hours old still waits, and the nightly sender sends a session only an hour after it first captured it.oars pool loginandoars pool depositname the exclusion-terms file's path andOARS_EXCLUSION_TERMSwhen there is no terms file; the deposit stays fail-closed without one.
oars 0.5.1
-
0.5.0 was tagged but never reached npm because the release job's pack step failed, so 0.5.1 is the first npm release of the 0.5 line.
-
oars pool depositnow builds a Codex session and queues its redacted rollout as the raw copy (with a yes to the raw prompt): the exclusion check reaches the rollout where Codex keeps it, through the Codex source, so a Codex session is scanned and cleared or excluded for a term instead of being left out as having no raw. Its<user_instructions>,<environment_context>, and AGENTS.md text are replaced as for any harness.oars sessions exclusionsandoars sessions gold samplecount a Codex session by what its rollout holds, not as having no raw. -
oars club create|join|leave|showandoars board(each with--json): make a club, join one by its code or the open board, leave, see the members, and read the open board without signing in. Create and join show the pool's consent text and need you to typeyesat a terminal. Errors lead with the cause (not_signed_in, also when the pool no longer accepts your sign-in;club_full,nickname_taken,nickname_invalid,rate_limitedwith when to retry). See the README, "club". -
oars cardnow counts the last N whole UTC dates ending today, so a 30-day card can no longer read more than 30 active days. Before, the window was N times 24 hours and could touch one more date.
oars 0.4.1
Text only, no change in behavior. The telemetry notice, oars telemetry status, the oars login error, and the package README said or implied the Many Oars pool at pool.manyoars.com was not deployed. It is live, and the notice and README now say so. The notice also now names two things 0.4.0 always sent without naming: the harness name and version and the model name on hook events, and the lists of harness versions and model names in the session store snapshot. Which fields are sent, the telemetry default, and the consent flow are unchanged; oars telemetry off, OARS_TELEMETRY=0, and DO_NOT_TRACK=1 still turn it off.
oars 0.4.0
Changes since 0.3.0. The next-node read (hook install adds UserPromptSubmit, and recall reads the kept pool bundle for the session's own harness), the raw-log consent prompt, the nightly send, Pi coding agent sessions (--allow pi), and quoting of a Codex argv word that holds a space.
oars sessions deposit --allow pireads Pi coding agent session files (~/.pi/agent/sessions, or$PI_CODING_AGENT_DIR) into the same store as the other harnesses:harness: pi, the active branch of its entry tree, the real exit code of a failed shell call, and usage that Pi already reports net of cache reads. The reader is written from Pi's public docs and source and has never run against a live store (docs/sessions-store.md, "The Pi source", names the hand run that would settle it);pool fetch --harness piand the closed telemetry list accept the new harness name.- A Codex shell call sent as an argv keeps a word that holds a space or quote as one quoted word (
bash -lc 'git revert HEAD'), so a git undo run through the wrapper reads asrevertthe way it does on the other harnesses (CAPTURE_VERSION25, so the next deposit re-reads every transcript). - Public text states what the code does at this commit (t897): the READMEs,
docs/, the CLI help, the telemetry notice, the raw prompt, and the package description no longer say the pool is reachable, that a batch is always printed before the nightly task sends it, thatraw/content never leaves (a redacted copy does, after a yes to the raw prompt), or thatoars sessionsuploads nothing; figures from one machine are removed fromdocs/gate.md. Every claim checked, with its evidence, is indocs/copy-audit.md. - The next-node read (t926).
hook installaddsUserPromptSubmitand widensPostToolUseFailureto every tool the action table maps. Recall reads the kept pool bundle for the session's own harness and version beside the local store (pool-replica.ts, shared withpool fetch; none or a corrupt one is local only). A failed call is answered with what followed that failure (same action and error class), silent on a same-step retry or a local edge under 5 followed steps;SessionStart(andUserPromptSubmit, when it changed) adds a table of the positions where earlier sessions split on the next step (at most 600 bytes). Every such read is arecalls.jsonlrow under the 10 percent holdback, and the old post-tool signature line moves under it too, someasure'shookHitsPerSessionfalls: a post-tool hit now needs a read that was served or held back. The pre-tool push keeps its gate role and no longer pushes a base rate: it needs a shape that fails half its calls or more (PUSH_FLOORS). Telemetry's hook name list gainsprompt. - The send lock treats EPERM and EACCES on Windows as contention, like EEXIST: an open that meets a delete still pending on the lock file is retried within the wait instead of failing the send.
- The raw prompt names every class the redactor cannot make safe,
localNamesfalls back toUSERorUSERNAMEwhen the account has no passwd entry (a container run with--user <uid>), and a stale send lock is taken over under a guard file that re-reads the lock first, so a live lock is never moved or removed, and a waiter gives up after its wait even while a guard is held (t884). - The raw copy masks a local name before the path rules run, and also in its encoded-folder form, so a hyphenated or spaced name (
mary-jane,John Smith) is masked whole inC:\Users\...andC--Users-...instead of leaving its second word (t884). A name that is a common word (dev,node,app) is masked wherever it stands as a word, which mangles the copy but never leaks a name.docs/sessions-store.mdlists both under what the redactor cannot catch. - The send lock is kept alive while its holder runs (touched every quarter of the stale window), a stale lock is taken over by one waiter only (under a guard file), and a run removes the lock only if it still names that run, so a long first night of raw parts cannot be taken over by a second sender (t884).
- The redacted raw copy masks the local username (the account's and the home folder's name, as a whole word, as
<user>), the harness's encoded project folders (-Users-name-,C--Users-name-,-home-name-), shell prompts (name@box:~$), and IPv6 addresses.REDACTION_VERSIONis 1, the first shipped rule set. After an explicit no to the raw prompt,depositsays so and namesoars pool raw --consent yes;verifysaysrawMaskedcounts placeholders (t884). - The nightly task sends the pool outbox after covenant acceptance (t884): new skeleton batches, and the redacted raw copies (
PUT /v1/raw/:sha256/parts/:n, then/complete) only after a yes to the raw prompt and for a folder not at skeleton only or off. It never accepts a covenant; with an unusable server, no sign-in, or no acceptance it sends nothing. Sent hashes are kept insent.log, a failed night is picked up the next night, andpool-nightly.jsonrecords counts only.oars pool outboxprints what is queued when you run it; the nightly task builds new batches and sends them in the same run without printing them first. The README anddocs/pool.mdsentences that said nothing is sent are corrected. - The nightly send and
deposit --sendtake one send lock per outbox (asend.lockfile besidesent.log), so two runs wait for each other instead of posting one batch twice (one case is left: two waiters that clear a dead guard file at the same moment can both take it over);--sendrecords the hash it sent insent.logtoo and skips a hash that file already holds (one line, "already sent at ", exit 0, no request), the rule the nightly follows, andoars pool outboxmarks each sent batch and raw copysent: <time>(text and document).oars sessions verifyreportssentBatchesandsentRawbesidestaleBatches, andrawCopiesandrawMasked, the values masked in the queued raw copies. The deposit summary names the real command,oars pool raw --consent yes. oars sessions deposit --allow otelreads agent traces written in the OpenTelemetry GenAI conventions (OTLP JSON files in the folderOARS_OTEL_TRACESnames, else<config>/oars/otel-traces) into the same store as the coding harnesses, with the same field classes, redaction, and verify checks (t927). It is convention-derived and has not run on a real trace; seedocs/sessions-store.md.- Capture keeps a skill-run name, a web host, and a pull's package name as they are after your redaction (t903,
CAPTURE_VERSION24, so the next deposit re-reads every transcript from the harness orraw/): before, each was<literal>unless a public list held it, sopool deposithad no name to send for a real session.<literal>stays for a value your redaction changes, that is not a name, or that is over 128 characters, and a pull's version is still masked beside a package no public list holds. They leave the machine only in pool batches built throughpool deposit's one gate (your exclusion terms apply there), sent bydeposit --sendor the nightly task, andpool outboxprints what is queued when you run it. A transcript neither the harness norraw/still holds keeps its<literal>values. - Three interventions read what the transcript shows (
CAPTURE_VERSION23,EXTRACTOR_VERSION15, so the next deposit re-derives the store from its log):git checkout feature/xandgit checkout "fix-1"no longer readrevert, and a checkout with an unlisted flag such as--oursreadsrevertwhen it restores a file; capture keeps whether the target was a path or a word, never its text, and a step captured before keeps its old reading. A Codex decline (exec command rejected by user,patch rejected by user) and interrupted turn, and Gemini CLI's[Operation Cancelled] User cancelled ..., readuser-stopby their fixed text or record type. A Codexturn_abortedwith reasoninterruptedis now written as the interrupt marker prompt (CAPTURE 23), so it setsstoppedinterruptas Claude Code's does and that session's prompt count,tsLast, and skeleton move; a decline leavesstoppednull. That changes theuser-stopbullet below (EXTRACTOR_VERSION14) where it says Codex records no interrupt marker as a prompt record: a Codex interrupt now does, and Codex and Gemini declines still do not. The same bullet's "so their prompts read as before" is also no longer true for Codex and Gemini declines: their declines no longer read as before, they readuser-stop.docs/sessions-store.mdhas the table by harness. oars sessions deposit --allow opencodereads OpenCode (sst/opencode) sessions into the same store as Claude Code, Cursor, Codex, Copilot, Gemini CLI, and Grok Build (t924): skeletons withharness: opencode, the same field classes and redaction, and the samerecall,measure,books, andverify. It readsopencode.dbread-only, and was written from OpenCode's public source (MIT) and tested on synthetic fixtures; it has never run against a live store (seedocs/sessions-store.md). A new source takes noCAPTURE_VERSION.- Served cells carry what they cover, and a step with no recorded cost is absent (t921). A pool cell holds
tokenSessionsandtokenPrompts(migration0013_token_coverage.sql, optional on the bundle's cells), so the cost read'stokensPerPromptdivides by the prompts that recorded input and isnullwhen none did; a row from before the migration is coverage unknown and the read saystokenCoverage: "unknown"instead of guessing. A tool-step cell (and an error cell) leaves a step whose cost was not recorded out oftokens,tokenSessions, andtokenPrompts, and readsnullwhen none was, where it used to read 0. The cost method version stays 2. - The pool seals each staged part of a batch sent in parts (AES-GCM under a key derived from
INGEST_HASH_KEY), so no host is readable in `deposit_part...
oars 0.3.0
Changes since 0.2.0. Four command groups are new (keys, sessions, gate, pool); login and logout keep their flags.
- The hosted-agent commands of 0.2.0 (
deploy,status, andtemplates) are not in this release; a script that calls one gets the unknown-command exit (2). oars keys create|list|revokemanages your API keys.keys create --workload <id>mints a key limited to one workload (authoritiesAGENT_OPERATEandAGENT_READonly), so a scheduled agent can change its own schedule without being able to reach another workload.oars sessionskeeps a local, deterministic memory of your own Claude Code sessions:deposit,recall(with--mcpto serve the same query to a running agent),lineage,read,books,measure,verify,exclusions, andhook, plusgold sample,gold arms, andgold score. It makes no model call and no network call, exceptgold arms, which sends redacted text through your own Claude subscription.deposit --allowalso reads Cursor, Codex, Copilot CLI, Gemini CLI, Grok Build, and VS Code Copilot Chat sessions (Stable, Insiders, and OSS builds), each opt-in by name; the VS Code source is written from VS Code's public source and synthetic fixtures and has never run against a live store, and its sessions carry no per-model input figure, soinput0 there means not known.- Every deposited transcript is also hard-linked (copied across volumes) into
<config>/oars/sessions/raw/, on by default, so it outlives Claude Code's own cleanup. These are raw transcripts on your machine, and the files themselves are not uploaded. oars sessions hook installwrites four fail-open hooks (SessionStart,PreToolUse,PostToolUseFailure,SessionEnd) into a Claude Code settings file;PreToolUsehas no matcher, so it starts node on every tool call, and for a call shape the store has seen fail it adds counts as context, with 10 percent of those held back (they are recorded and nothing is injected);hook statuslists them andhook uninstallremoves only the entries install wrote. Install is idempotent, and raisescleanupPeriodDaysin that file to at least 365 (uninstall leaves it raised).--schedule nightlyregisters a daily deposit task on Windows. A session-end deposit that finds the store locked records its session inpending/in the store and exits; the next deposit of any kind deposits the pending sessions first. The step log shows that as adepositrow with statusdeferred, whichmeasureprints asdeposit@<ev> deferred (store locked).- What a deposit captures (capture version 17): per session, the setup it ran with (permission mode, plugins, hook count, instruction bytes), the tool funnel (tools offered, loaded, removed), the trust shape (interrupts, rejections, policy denials, minutes the agent ran between human turns), the
errorCodeclass of each failed call, the packages an install or runner call resolves and the dependency names a manifest shows, and stack strata (language, repo-file bucket, framework, countedexplore). A call the user refused is a failed call witherrorCodeuser-rejectedfor Claude Code, Cursor, Codex, and Gemini. A Gemini session that ran on Windows stores its shell calls asPowerShell, and Gemini sessions under~/.cache/.geminiare read. A rotated Copilot log (events.N.jsonlbesideevents.jsonl) is read oldest first as one session. The MCP tool name aftermcp__is redacted wherever it is stored. Recall floors a step's token count to its bucket, so no local cell holds an exact number, and cells are weekly (keyed on the Monday of the UTC week), with day cells for the reads of days. A language share on the wire may nameother, the name a touched file with an extension off the pinned list gets, so a session that touched a.mdfile is not refused; the frame'slanguageslist stays closed to the pinned names. The store is at extractor version 12 (a derived prompt carries its strata), and an older store re-derives from its log with no transcript read. A Copilot session's reasoning tokens are shared over the responses of its model in proportion to their output, no line above its own output, so a healthy session passes the thinking invariant ofoars sessions verify --strict. - On a large store (a test store whose log holds superseded records), a deposit that skips unchanged files needs a 104 MB heap and
verify248 MB; a store too large for node's default heap still needsNODE_OPTIONS=--max-old-space-size=<MB>(docs/sessions-store.md). A deposit whose read of its ownstore.lockis refused at the end (an antivirus share violation on Windows) removes the lock instead of leaving it for 10 minutes, a store lock whose holder hangs is refreshed for at most an hour and, once it has aged 10 more minutes, the next deposit breaks it and prints one line naming the holder's pid, and a Codex thread that a crash tore stays readable. A deposit that loses its store lock (held past the hour cap, then broken by a waiter that a second deposit followed) checks that the lock file still carries its own token before it writesstate.json, replaces the log, or compacts, and stops withThis run lost the store lock to another deposit(exit 1) when it does not, before any further capture, raw archive, pending, derived, or hook index write. It never overwrites the second deposit's--allowand--redact-domainentries, so no domain stops being masked. oars sessions verifyaudits each session against its raw file (raw-file match, Merkle root, count bindings), reports every session on its own, and scores each against your own other weeks on its harness (selfBaseline).verifyandbooksname the project folders that hold transcripts but are not on the saved allowlist, which are never read (scopein--json); that list never failsverify --strict.oars sessions exclusionsscans your sessions against a local term file (OARS_EXCLUSION_TERMS) and writes the excluded ones to a list, printing counts only.gold samplerefuses a draw with no term file, and--exclude-termonly adds to that file's terms.oars sessionsships a pinned public vocabulary (VOCAB_VERSION1, snapshot of 2026-11-04,packages/cli/vocab/, about 6 MB raw in the package) and the functions that project the fields a pool would carry; they compute from the store and write nothing to disk. A shell call projects toprogram,subcommand, andflagsfrom that program's list, sogrep AcmeWidgetMergerisgrepalone; a program, package, tool, or skill outside the vocabulary isprivate,custom, or only a count. A client more than 30 days past its snapshot builds no pool batch (vocabulary_expired).oars gate installandoars gate uninstallwire one fail-openPreToolUsehook into Claude Code's user settings (rerunning changes nothing, andinstallsets the file's mode after writing and follows a symlink that points at nothing yet), andoars gate hookrecords each package, MCP server, and skill pick as it happens intocrossings.jsonl, in under a second.oars gate replayfinds the same crossings in session logs already on disk, andoars gate reportcounts that one file once per crossing, by who recorded it. Names are read exactly: annpx <name>whose folder holdsnode_modules/.bin/<name>islocal-bin-run, notfetch-and-run; a name is not sent to a registry whenHOME,sudo, a sourced file, or a package-manager config setting could change its configuration; a version with a character such as#,&,?,;, or%counts asunreadable-name; and a resumed transcript counts a call once.oars gate hookadds a note about a name first published within 30 days or not found, with no policy file needed.gate-policy.json(in the sessions store, written by you) sets each class (young-name,not-found) towarn,block, oroff, with a day count and a holdback share (default 10 percent, which withholds notes at random and never a block). A note states a registry fact; a block is a labeled deny that exists only fornot-found(an npm name; a Python name never readsnot-found) and only when you writeblock. Each row records itsarm, thelineShown, thelineHeld, and thepolicyit ran under, andoars gate policyprints the policy in force. A registry ask the hook has to cut is finished by a detached helper that fills the cache; a call starts at most three helpers. The hook makes the lock for a helper in the same step that starts it, and takes the lock back before the process can end when the helper cannot start (the start throws, or the child reports an error), so no lock outlives its helper and keeps a name unasked for five minutes. A hook the deadline ends stops its own ask and writes one not-read row, written before the stop so a slow stop (ataskkillon Windows) cannot cost it; the deadline writes nothing when the hook's own append has already put the record on disk and then stalled, and another hook's row in the same file never makes it skip its own. One name twice in a call is one note and one counted row, and a block label prints once (the call is still denied). A kept registry answer is read only in the folder that asked. A crossing goes nowhere unless you opt in, and no opt-in exists yet.- The pool's ledger has no expiry: migration
0009_ledger_no_delete.sqlmakes the database abort anyDELETEof adepositsrow or acontributorsrecord. A counting batch turns an account frompendingtostandingafter 90 days and stays on the ledger. The pool counts two strata of their own beside the single ones, (language, share) and (repo size, explore bucket), each held to five accounts and served only in the bundle'sstrata; provider errors inliveare not built yet. The pool's bundle serves tool-day reads from day cells (a tool's failure rate and its share of the day's steps, with the floor counted within its day) and a version-free (name,platform, `shel...
oars 0.2.0
oars templateslists the starter templates, andoars deploy --template <id>deploys one without an image of your own.oars loginworks with no flags once Many Oars has launched: the production issuer is built in, and
https://manyoars.com/.well-known/oars.json({"launched": true}) switches it on. The flag never supplies the URL.
Before launch,loginsays "Many Oars has not launched yet: pass --issuer-url or set OARS_ISSUER_URL." and exits 2.
--issuer-urlandOARS_ISSUER_URLstill win.- An exit 6 ends with a "Needs you:" line that names the link for the step only a person can do, such as finishing account setup.
deployrefuses a mix of flags with a message that uses the serial comma; a--templatethe catalog lacks exits 2 and an
unreachable catalog exits 8, before anything is registered.
per-task-agent 0.2.2
Image: ghcr.io/manyoars/per-task-agent@sha256:3dfe58ccacfcf8bf136e487e0dd6152092d26e6349f9671a2c46f78a097c07b3
Pin this reference in a catalog entry, not the tag.
Required environment: OARS_PUBLIC_URL (the agent's own public https origin; the platform injects it beside AOR_SDK_KEY). Checkout success/cancel URLs are built from it, and the server refuses to start without it (unless OARS_DEV_FAKE_BILLING=1).
Changed since 0.2.1:
- A billing-unavailable problem that used to surface as a bare 503 is now logged with its code and status, so an operator can see why.
- The template README now documents
OARS_PUBLIC_URLto match the config, and says the platform injects it. - The template release workflow builds the image once and pushes that same image, and the release checkout keeps no credentials.
- The template test headers and workflow tests were tightened (a read-only check job, a tag-only release job, no docker build step in check).