Repository navigation
Releases
Release v0.11.0
What's New in v0.11.0
Security
-
read_filewithholds a file with more than one hard link (#703). A hard link to.envunder another name (notes.txt) matched no name rule and was returned verbatim. On Unix a regular file whosenlinkis above 1 is now refused, and the refusal points toread_file_raw(approval); a single-link file is unaffected, and nothing changes on other platforms. It reads only thenlinkof the metadata it already has: no directory scan, so a legitimately linked file (a hard-linked build artifact) needsread_file_rawtoo. The git diff filter andcora_searchhits stay name-based. -
A known secret is now redacted in more of the forms it takes in a sink (#681). The matcher (
SecretSet::redact) only knew the value alone, its JSON escape, base64 of the value alone and percent-encoding, soAuthorization: Basic base64("user:" + token), a dockerconfig.jsonauth, a base64'd.env,xxd -phex, a JSON\/or\uXXXXescape (Pythonensure_ascii, PHP) and a value split by a zero-width character reached the provider, the log and the other sinks unredacted. A value now also matches as base64 (standard and URL-safe) at each of the three offsets inside a larger payload, keeping only the characters that do not depend on the neighbouring bytes, as lower- and uppercase hex, and with\/and\uXXXXescapes (surrogate pairs included). Text holding a zero-width, bidi or soft-hyphen character is scanned a second time with those characters removed, and the matches are mapped back onto the original. At most 24 needles per value, the 8-character floor still applies (no needle is shorter), and a text without invisible characters is scanned once; redaction throughput on mixed text is about 350 MB/s (was about 660 MB/s). Residual, documented in the threat model: arbitrary transforms (line-wrapped base64, mixed-case hex, gzip, ROT13, a value cut in pieces) are not recognised; the shape detector (#597) cannot close that either. -
A server the plugin lock refuses no longer reaches the model through resources or prompts, and an unapproved or revoked server is not started at all (#680). Before, a refused server's tools stayed out, but tole still connected to it, listed its prompts as
/mcp:commands, offered it inmcp_read(ReadOnly, no approval) and rendered its prompts into the next message. Nowregister_server_toolsrecords the admission decision on the session's own registry (ToolRegistry::mcp_admitted, per session, never process-wide), the roster probes extras and offersmcp_readonly for admitted servers,tole chatlists and renders/mcp:prompts only for them, and the host skips a server with no lock entry, a revoked one, or every server under an unreadable lock before connecting (plugin_gate::Gate::refuse_unseen), with the same refusal note as before. A server whose manifest changed is still connected once, to detect the change. Prompts and resource templates stay outside the manifest hash; they are reachable only from an admitted server. -
Live tool events no longer show secrets the session log hides (#677). The
tool_startedinput, thetool_finishedpreview of an output or error, and the reasoning that serve SSE (GET /sessions/{id}/events), ACPtool_call_updateand the TUI forward were the raw values, while the log of the same turn was redacted (#596). The turn loop now redacts every observer event with the session's own secret set (Storage::egress_secrets), before the 200-character preview cut, so a cut never leaves part of a secret. Streamed answer text is model output and is not redacted live (its committed entry is). -
The advisor's context no longer lets a secret dodge redaction by being split with an invisible character (#587).
advisor::cleanredacted known secret values and only then stripped control characters, so a value split by a zero-width, bidi or control character (tok\u{200b}-...) was not redacted and re-formed once the character was removed. Control and invisible characters (newline and tab kept) are now stripped first, then secrets are redacted, then the fence markers are neutralized, then the text is capped. The multi-round discussion sanitizer (#588) uses the same order. -
A session directory swapped for a symlink no longer redirects the session log (#639).
JsonlStorageresolves its directory once, walks the canonical path from/withO_NOFOLLOW | O_DIRECTORYat every component, holds that directory handle, and opens the session file, its.locksidecar and compaction's temp file relative to it (the rename and the directory sync too). Before,O_NOFOLLOWcovered only the last component, so a job that swapped.toleor.tole/sessionsfor a link could have the.locksidecar created elsewhere. Unix only; no new dependency; other targets keep the path-based open. Regression tests swap the directory aftercreateand afteropenand assert nothing is created through the link. -
Webhook routes on
tole serve, part 1b of #582: the security gate. No job is fired yet; filters and the action come in part 2, andtole hookin part 3.- Each
[hooks.<route>]table in the user config servesPOST /hooks/<route>. A project config cannot define or change a route: itshookstable is ignored with a warning. - A route needs an HMAC-SHA256 key, named by
hmac_key_envorhmac_key_file(a regular file, not a link or FIFO, at most 4 KiB, at least 16 bytes). A route that cannot load stopstole servebefore it listens, naming the route. Onlyinsecure_loopback = trueskips the signature, and it loads on a loopback bind only, for loopback clients. - Deliveries are checked with
tole_core::webhook(#643): nativeX-Tole-TimestampandX-Tole-Signaturewithin ±300 s, or GitHubX-Hub-Signature-256. The bearer token is not a webhook credential, and a failed signature counts toward the source IP's failed-auth limit. - Every delivery id is recorded as
hook_seenin the cron ledger. A repeat gets200 duplicate, after a restart too; an accepted delivery gets202. - Limits per route:
rate_per_minute(default 30, at most 600), counted from the ledger so it survives a restart;max_body_bytes(default 256 KiB, at most 1 MiB);- an unknown route (404) or an oversized body (413) is answered before the body is read.
- The key never appears in errors, logs, responses,
config checkor the ledger. - Accepted limits, written in the threat model:
- a GitHub delivery replayed after the 7-day retention verifies again, since GitHub signs no timestamp;
- the ledger grows by one line per accepted delivery and is not rotated;
- bad-signature limiting is per IP and in memory.
- Each
-
Agent-created cron jobs (#581 part 2). An agent creates a job only through the new
cron_createtool (Risk::Write, so each call is approved or allowlisted). The tool exists inrun,chatandservesessions at depth 0 outside plan mode; a cron run, a child agent and the MCP server face do not have it. The host saves the job paused in the session workspace, withcreated_by = agentandcreator_allow(the session's allow list). The tool input refuses an id, a script, skills, a profile, a pause state and every provenance field by name. Itsallowmay name only the creator's allow patterns, and the creator's deny patterns are always kept; acommand:delivery is refused.tole cron resumeis the human enable and now recordsenabled_by = cli.- Behavior change: every job not saved through
tole cron create(a hand-written line withoutcreated_byincluded) isblockedunless it was enabled withtole cron resume(not_enabled), stays within its creator's allow list and deny patterns (allow_exceeds_creator), and is not a script (script_untrusted). - Behavior change: an agent run whose workdir holds the cron dir is
blocked(ledger_in_workdir), because its file tools could write job records. Give such a job a workdir that does not contain$TOLE_CRON_DIR(by default<codecora home>/tole/cron). tole cron editkeeps an agent job's creator, grant and enable, and refuses an--allowbeyond the creator's.- The ledger fold keeps a live job's
created_byandcreator_allowfrom the line that created it (tole_core::ledger::PROVENANCE), so a later line cannot change them. - No config defines jobs, and a cron run reads no project config from its workdir. When jobs gain MCP servers, they go through the plugin lock (#636).
Design:
docs/automation-design.md§5. - Behavior change: every job not saved through
-
Sensitive files are read redacted (#598 part b).
read_fileon.env*,*.pem,id_*,.npmrc,.netrcor*.key(file name, case-insensitive, by the requested name or the resolved real path, so a symlink to.envcounts) returnsKEY=«redacted»lines, a key file a single«redacted», and no content hash. The verbatim read is the newread_file_rawtool,Risk::Write: approve its call, or start with--allow read_file_raw. Thegittool's diff hides every hunk line of a sensitive file, andcora_searchhides the text of a hit in one.read_filenow opens a file without blocking or following a final symlink and checks it on the handle, so a FIFO swapped in after the check cannot hang the read.tole doctoradds a secrets report: which secret names are set, which tools may receive each, what reaches the children of model commands and of a child session (never a value). Tool list and approval:docs/secrets-integrations-design.md§10. -
Known-value secret redaction, part 1 of #595 (epic #594). New
tole-coremodulesecretswithSecretSet: it collects the values of every secret-named env var (the ENV-1 name rule, nowsecrets::is_secret_name, shared with the child-env scrub) plus explicitly named values (a provider key, URL userinfo passwords), andredactreplaces each value and its base64 (standard and URL-safe), percent-encoded and JSON-escaped forms with«NAME»; overlapping matches merge, unm...
tole v0.10.1
What's New in v0.10.1
Patch release: TUI fixes from the Cora scan backlog (#516) — and GitHub Releases now ship binaries via release.yml (#515).
Fixed
- Cora-scan minors on the TUI surface (#514): the unknown-command hint no longer byte-slices the typed command (empty or multibyte input could panic), the
@filewalk also skips names containing a newline, and the status snapshot drops a discarded float division and a duplicateimplblock. Waived with reasons: holding the session lock across a turn (exclusive&mutstorage by design), counting compactions viacompaction::history(one-shot command, no storage counter), and un-cleaned temp dirs instatustests (OS temp, no tempfile dependency).
📦 Platforms
| Platform | File |
|---|---|
| Linux (x86_64) | tole-x86_64-unknown-linux-gnu-v0.10.1.tar.gz |
| Linux (ARM64) | tole-aarch64-unknown-linux-gnu-v0.10.1.tar.gz |
| macOS (Apple Silicon) | tole-aarch64-apple-darwin-v0.10.1.tar.gz |
| Windows (x86_64) | tole-x86_64-pc-windows-msvc-v0.10.1.zip |
🚀 Install
# From crates.io (canonical — `tole upgrade` tracks it)
cargo install tole-cliOr download a binary archive below, verify it against checksums-sha256.txt, and put tole on your PATH.
Full changelog: https://github.com/codecoradev/tole/blob/main/CHANGELOG.md
tole v0.10.0
tole v0.10.0
TUI steering, follow-up queue, collapsible output, @file, markdown & diff styling (#462).
Added
- TUI steering, follow-up queue, collapsible output,
@file, markdown and diff styling (#462):Enterwhile a turn runs pushes the draft on the existing soft-interrupt queue (#377), so the turn commits it as an attributed user message at its next safe point and continues (nothing is cancelled; a steer the turn never reached runs next, or returns to the editor after a cancel);Alt+Enterqueues a follow-up shown asqueued: …rows above the editor, the head runs after the turn ends with an answer (one at a time),Alt+Uppops the last back into the editor, andEsc/Ctrl+Ccancel and return the whole queue to the editor.Ctrl+Oexpands/collapses tool output (multi-line results are collapsed to the first line plus[+n lines · Ctrl+O]),Ctrl+Thides/shows reasoning; both are UI state only. Typing@opens a file panel (depth ≤ 3, ≤ 2000 files listed, 50 shown,Tab/Enterinsert the path); answers get minimal markdown styling (headings,**bold**, inline code, fenced code) and+/-/@@rows of tool output are green/red/dim, all sanitized before styling. Not included (deferred): tables, nested lists, italics and links in markdown; retroactive collapse of already-printed scrollback; diffs beyond the 200-character tool preview;@insertion at an in-buffer cursor. - TUI model picker, thinking level and slash-command menu (#461):
/modelandCtrl+Lopen a filter-as-you-type picker over the provider's/models,/model <id>sets one directly,Ctrl+Pcycles the models used in this run;/thinking [level]andShift+Tabcycleminimal → low → medium → high → unsetand the editor border shows the level; typing/opens a command menu (Tabcompletes,Up/Downselect) for/compact,/new,/resume(picker over the global session index),/session(the status lines),/nameand/quit; unknown commands are refused with a prefix hint. Model, effort and name are session registers (fact/model,fact/reasoning_effort,fact/session_name), so the next provider call uses them and they survivetole resume;/nameis whatsessions --allandresume <name>show./compactrequests the forced pass before the next prompt (no mid-turn compaction; its optional instructions are not supported yet and are reported as ignored). Not included (deferred):/resumepaging beyond the newest 50, replaying the old transcript after/new//resume. - Terminal UI MVP (#460): bare
toleon a TTY now talks to the agent. Inline viewport with a streaming transcript (answer text as it arrives, reasoning dimmed, tool calls and results as compact lines), a multi-line editor (Entersends,Shift+Enter/Alt+Enter/Ctrl+Jnewline,Up/Downhistory), a footer (folder, session, model, effort,context used / window (%), session tokens from thestatussnapshot), inliney/napproval for Write and Destructive tools (the normal--allow/--yesrules still apply), andEscto cancel the running turn. Sessions are ordinary tole sessions (tole resume,sessions --all). Hiddentole tuisubcommand (what baretoleruns) goes through the normal config gate. Not included (deferred): in-line cursor movement, slash commands beyond/quit, memory loop, MCP prompt commands, sibling notices, soft interrupts (#377/#383). - Reasoning effort (#457):
--effort <minimal|low|medium|high>/TOLE_REASONING_EFFORT/ configreasoning_effort(flag > env > config > user config > unset, validated; shown bytole config check) is sent asreasoning_efforton OpenAI-compatible chat completions only when set, so an unset request body is byte-identical. A session's durablefact/reasoning_effortregister overrides it and survives resume (an empty or invalid value falls back to the default). A provider that answers HTTP 400 namingreasoning_effortgets ONE degraded retry without it: the rejection is remembered for that provider and recorded under the namespacedtole_degradedkey of the step's usage row, never counted as a provider step. New core API:openai::{parse_reasoning_effort, REASONING_EFFORTS},OpenAiProvider::with_reasoning_effort,Provider::last_degraded(default None). Not included (deferred): a setter for the register (the TUI/thinkingcommand, #464) and the ACP config option. - Multi-client daemon design note (#382):
docs/daemon-design.mdrecords how jcode's daemon works (single server, Unix socket, client registry, 5-minute idle timeout,exec()hot-reload, client auto-reconnect), a clearly speculative mapping onto tole, the user stories that would un-park it, and an explicit anti-goal. PARKED (tier B): docs only, no code, nothing to build without a user story and owner approval. - Claude Code session import (#381):
tole import-claude <transcript.jsonl>imports a Claude Code session (the~/.claude/projectsJSONL orclaude -p --output-format stream-jsonoutput) as a NEW tole sessioncc-<claude-session-id>; continue it withtole resume cc-<id> "<prompt>". Prompts and assistant text become messages,tool_use/tool_resultblocks become intent / result entries (Claude Code tool names are kept as history only, never executed), model, version, branch and timestamps are recorded in thefact/cc_importregister, and the header takes the transcript's cwd. Additive: existing sessions are untouched and importing the same session twice is refused. Malformed lines, thinking blocks and sidechain (subagent) lines are skipped (malformed ones with a warning); a dangling prompt or open tool call at the end is dropped and reported so the session rests atFinal. New core API:ccimport::{import_file, import_str, CcImport},JsonlStorage::cwd. Not included (deferred): attaching to a runningclaudeprocess (and its double-attach guard), image blocks, a live provider round trip (unverified). - Detach a running turn (#217): stop waiting without aborting anything.
POST /sessions/{id}/detach(serve;202,409when no turn runs) and the MCP tooltole_session_detachset a newCancelToken::detach()flag. The running tool settles and its result is committed; before the next provider call the turn writes a durabledetachedentry and settlespc=Finalas the newTurnOutcome::Detached(stop reasondetached;tole runexits 9). The session stays prompt-resumable, and a cancel always outranks a detach. A background job started withjob_startin a serve/MCP/ACP session now nudges that session when it exits: a notice ([tole] background job <id> finished …) is queued on the soft-interrupt queue and delivered at the next safe point, which is the next turn's first when the session is idle. New core API:CancelToken::{detach, is_detached},TurnOutcome::Detached,JobStartTool::with_completion_notice. Not included (deferred): the Ctrl-B key and atole detachcommand forchat/run(they have no concurrent input while a turn runs; this lands with the terminal UI, #464), moving an in-flightrun_commandinto a job (tools are not interruptible mid-call), nudges that survive a host restart (the queue is in memory;job_pollstill has the result), and the ACP face. - MCP prompts and resources (#213): the MCP client now probes each server's
prompts/resourcescapabilities (probe-first: a server that advertises neither is never asked and adds nothing). Intole chat, a server's prompts are listed at startup and invoked as/mcp:<server>:<prompt> [key=value ...]; the server renders the prompt, tole prints it and sends it as that turn's user message (text parts only, capped at 256k chars). Servers that advertiseresourcesadd one ReadOnlymcp_read {server, uri}tool (resources/read; text contents only). Server prompt text is server-controlled content (threat-model note added); tool calls it provokes still pass the approver. New core API:mcp::{probe_server_extras, get_server_prompt, parse_prompt_command, McpReadTool, McpExtras, McpPromptInfo}. Not included (deferred):tole prompt runoutside chat, quoted/spaced argument values, resource templates,resources/listsurfacing, and subscriptions. - Session forks (#210):
tole branch <id> [--at <entry-id>]starts a new session holding the source's history up to a settled entry (a final assistant answer; default: the last one), so approach B can be tried without losing approach A; continue it withtole resume <new-id> "<prompt>".tole sessions --treenests forks under the session they were cut from. A fork is a separate append-only file whose header recordsforkedFrom: {session, entry}(additive; older files replay unchanged), so replay and todo hydration are correct per branch and the source is never touched. Usage rows and registers are not copied. New core API:JsonlStorage::fork/forked_from,storage::ForkOrigin.--atand the default cut only accept a final assistant message with no tool calls hanging off it. Not included (deferred): several leaves inside one file with lineage-following readers, resume by branch head, and a--forkflag onresume. - Image input (#212):
tole run|chat|mission --image <path|data:…;base64,…>(repeatable) attaches images to the first message, andtole_core::turn::run_turn_with_imagesis the embedder entry point. Paths must resolve inside the workspace jail (..and symlink escapes refused); inline images are spilled to<codecora home>/tole/images/; png/jpeg/gif/webp sniffed from bytes, 10 MiB cap. The OpenAI wire getstext+image_urlcontent parts; the write-once log stores only a reference (payload.images:path,mime,bytes,fnv1a64), and each request re-loads the file re-checking jail and hash, so replay/resume keeps working and a changed or missing file reaches the model as an[image unavailable: …]note. A non-vision model is a loudmodel-capabilityerror:TOLE_VISION=0refuses before any request, ...
tole v0.9.0
Context compaction, global session index, command allow/deny, failover chain, child-agent handoff.
Includes the 0.8.2 hardening patch (see CHANGELOG.md [0.8.2]), which was never tagged or published on its own.
Added
- Context compaction, wired (#440, 2/2):
chat,resumeandmissioncompact between turns once a prompt reaches--compact-at(default 0.8; configcompact_at) of the model's context window. The window isTOLE_CONTEXT_WINDOW/ configcontext_window, else what the provider's/modelsreports for the session's model (context_length,context_window,max_context_length,max_model_len,max_input_tokens,top_provider.context_length) — looked up lazily, only once a prompt passes 8k tokens, at most once per process; with neither, compaction stays off and says so once. A failed summary call is reported and the session continues uncompacted.--no-compaction/ configno_compactionturns it off;tole statuslists each compaction;tole config checkshows the effective values and their sources. - Context compaction, core (#440, 1/2):
tole_core::compaction.compact()summarizes the settled turns before the kept tail with the session's provider and commits ONEcompactionentry (summary, first kept entry, compacted count, prompt size before) together with the summary call's usage row — atomic, append-only, the transcript is never rewritten.view()is what the provider sees: the transcript as is, or the latest summary (inside the untrusted-content fence used for tool results, introduced as a record rather than instructions) followed by the kept tail; the turn loop sendsview().due()compares the last call made AFTER the last compaction againstPolicy { window, at, keep_turns }, so the summary call never re-triggers it. Inactive until the CLI wiring (window resolution,--compact-at,--no-compaction, mission/chat/resume hooks,tole status) lands. - Global session index (#439): every face that starts a session (
run,chat,mission, serve/ACP/MCP) appends one line to$CODECORA_HOME/tole/sessions-index.jsonl(default~/.codecora/tole/, owner-only, append-only, a bad line is skipped): id, absolute session file, cwd, face, start time — never prompt text.tole sessions --alllists sessions from every project newest first, with their cwd and first-prompt preview read from the session file (gone files are hidden);--grep <text>filters on prompt, name or cwd.tole resume <id>for a session of another project refuses with the exact command to run there (cd <project> && tole resume <id>): resuming from here would apply this directory's workspace jail, config and trust to that session.TOLE_NO_SESSION_INDEX=1turns recording off. - Command-level allow/deny patterns (#438):
run_command:<argv pattern>in--allow/ configallowmatchesrun_commandcalls by the argv the tool executes (same splitter, no shell, so;never chains a second command): one glob per argument, a final lone*= any further arguments, the program compared literally for an allow (run_command:*allows nothing). A leading!makes any pattern a deny that wins over every allow and over--yes; a deny compares the program by file name, case-insensitively, without.exe/.cmd/.bat/.com(/bin/RM,rm.exevsrm *). Fail closed: while arun_commanddeny is configured, a command runs unattended only when an explicitrun_command:allow names it (a barerun_commandallow or--yesis not enough) and never when its program is a known wrapper (sh,bash,env,sudo,xargs, ...); everything else asks the human (refused where nobody can be asked). Denies travel with allow lists, a child agent always inherits its parent's denies, and a child's command pattern must be covered argument by argument by a parent pattern (list_covers: under a parent command deny only a parentrun_command:allow covers it; a child wildcard only by the same glob or a lone*). InAllowlistApprover::deny_only(no allows), arun_command:deny refuses everyrun_commandcall.tole_core::approval::pattern_verdictis the one evaluation both approvers use. --trust workspace(#437):read_onlypluswrite_file/edit_filethrough a path guard. The tools are jailed to the workspace root (canonicalized, so../and symlinks cannot escape), and the guard keeps files another tool may later execute prompting: any dot-component (.git/hooks,.envrc,.vscode/,.tole/,.github/), build and task files (build.rs,Cargo.toml,Makefile,package.json,pyproject.toml,SKILL.md,AGENTS.md, ...) and scripts (.sh,.ps1, ...).delete_file(Destructive) andrun_commandkeep prompting; the preset needs an explicit--workspace(or configworkspace) and refuses a home or filesystem root. Not a sandbox: ordinary source files stay writable. The guard checks the RESOLVED target the tool will write (relative to the canonical jail root, symlinks followed; Windows trailing dots/spaces normalized, 8.3 short names refused), never the raw path text. It is a pattern suffix (write_file@workspace) matched bytole_core::approval::pattern_allows, so--allowand child agents use it too. API:Tool::approval_target(defaultNone; decorators must forward it) andToolRequest::targetare new.web_searchbackends beyond the fleet contract (#436):TOLE_WEB_SEARCH_METHOD(GETdefault, orPOSTwith the JSON body{"query", "limit"}),TOLE_WEB_SEARCH_HEADERS(Name=Valuepairs,${VAR}expanded from the environment, sent only to the backend; transport headers and line breaks refused; errors never echo a value) andTOLE_WEB_SEARCH_FIELDS(wire=backend, e.g.snippet=description,results=items). A malformed value fails startup instead of dropping the tool. API:tole_core::web::WebSearchTool::from_envnow returnsResult<Option<Self>, String>;WebSearchTool::new(url)andfrom_lookupare new.- User config (#433):
$CODECORA_HOME/tole/config.toml(default~/.codecora/tole/config.toml) holds global defaults with the project config's keys and validation. Precedence is flag > env > project config > user config > default, key by key; it needs no trust record,--no-configignores it too, andtole config checkshows both files and the winning layer per key ((user config)). TOLE_SKILL(#434): preload skill files from the environment (SKILL.mdpaths, separated likePATH) — the env layer of theskillkey, so--skillstill replaces it and it replaces the config'sskilllist. Server faces refuse it like--skill, and the refusal names the variable.- Code review with cora (#422, 3/3): when
cora mcp(cora ≥ 0.19) is attached, the system prompt oftole run/tole chatnames the agent-as-reviewer flow —mcp_cora_cora_review_preparewith the diff, the model's own judgement, thenmcp_cora_cora_review_finalizewith the same diff and its findings — so "review this change" uses cora's deterministic checks and verdict without an LLM key in cora. Absent in plan mode and without the review tools. - Supported sibling versions (#422): cora ≥ 0.19.0 and uteke ≥ 0.21.1, kept in one place (
tole_cli::ecosystem).tole doctorprints anecosystem:section reporting each binary as supported, too old (with its upgrade command) or not installed; it never blocks, since both are optional. - Provider failover chain (#209):
failover::FailoverProviderretries a step on the next provider of an ordered chain when the active one is unavailable (connection failure, timeout, dropped stream, HTTP 5xx or 429); request errors (401/403/404/400, malformed responses) fail loud and never fail over. The answering provider stays active and the primary is retried after a cooldown. Each step records the route undertole_routein its usage row (durable, auditable); rows holding only tole's own keys are not counted as provider steps. Wired into every session face throughTOLE_FALLBACK_BASE_URL/TOLE_FALLBACK_API_KEY/TOLE_FALLBACK_MODEL(a reserve without its own key is refused; the primary's key is never sent to another host). A stream that breaks after part of the answer was already shown live does not fail over (no duplicated output); a reserve that fails before the cooldown ends wraps back to the primary; and after a failover only the last provider's error decides the turn loop's transient retry, so an earlier timeout no longer makes a reserve's 401 look retryable. - Child-agent handoff schema and verify gate (#383):
tole_core::handoffdefines the typed handoff document (roleexplore/implement/verify/fix, artifact references by path and rev, verification evidence), strict validation of the model-authored input (workspace-relative paths, option-safe refs, bounded fields), the close gate (an implement/fix child without passing verification isUnverified/VerifyFailed, never closed silently) and the prompt for the follow-upfixchild.agent_starttakesroleandconsumes(artifact references the child reads itself), a role child gets ahandoff_submittool, andagent_pollreports the gate and a ready-madefix_prompt(the parent decides whether to start the fix child; nothing is spawned automatically). Design and anti-goals (no DAG engine) indocs/child-agents-handoff.md. - Sibling change notices, core (#380):
sibling_watchtracks which files each child agent reads and writes in one shared append-only log (owner-only, bounded) and tells a child, once per stale path, when a sibling changed a file it had read or written; theTrackedtool decorator recordsread_file/write_file/edit_file/delete_fileon success. Interrupts gain asource: a notice is delivered at the next safe point as an attributed entry (source: notice) and never withdraws output or supersedes a final answer. Wiring intotole runchildren follows separately. - Sibling change notices (#380):
sibling_watchtracks wh...
tole v0.8.1
Usage reporting, chunked read_file, scan-driven fixes.
Fixed
- The usage report no longer renders an absent per-step
messagescount as a measured0; it isnullin the status JSON when unknown (found by the pre-tag scan). read_filerejects an emptypathwith a clear message instead of a confusing "is a directory" error.
Added
tole statusand the serve status JSON now report token usage, cache-hit rate and a per-step request-size split (system+tools vs history, stored per step undertole_wirein the usage record);fact/missiongainscached_tokens. Instrumentation only: the wire, budgets and durable entry shapes are unchanged (#211, part 1 of 3).
Changed
read_filenow returns at most 20,000 characters per call and acceptsoffset/limit(characters); the result carriestruncated,total_charsandnext_offsetso large files are read in chunks instead of being re-sent whole on every later step (#211, part 3 of 3).
tole v0.8.0
Project config (.tole/config.toml) with a trust model, CodeCora-home resolver, gh remote host fix.
Added
- Project config
.tole/config.toml(#208): per-directory defaults formodel,base_url,
system_prompt,memory,sessions_dir,workspace,trust,allow,mcp_server, the
hooks (on_pretool/on_posttool/on_turnend),skill,plan_mode,no_auto_mcp,
no_skillsand the[mission]table (max_steps,max_minutes,max_tokens,verify,
verify_timeout). Precedence is flag > env > config > default; list keys are replaced
wholesale by the higher layer and booleans can only be turned on by flags. The file is
untrusted input from the repo: it applies only after you trust it (tole config trust;
content-bound, stored outside the repo in$CODECORA_HOME/tole/trusted-configs.json).
Interactive commands ask[y/N]showing the full content and a diff;serve/acp/mcpand
non-interactive runs fail closed.--config <path>names a file explicitly (trusted) and
--no-configignores it. Secrets (api_key, tokens, ...) are rejected,Destructivetools
can never be allowlisted through the config, and faces that refuse hooks, skills or MCP
servers refuse them from the config too. tole config check|trust|untrust: validate the file and show every effective value with its
source (flag/env/config/default); trust or untrust the current content.
Changed
- CI: the bare
--no-default-featuresprofile oftole-cliis now tested (not only checked), and the ACP/serve e2e tests are gated onshell-tools(#349). --skill/--no-skillsontole serve/acp/mcp/missionare now refused in builds without themcpfeature too (they were silently ignored there), and the refusal also covers values coming from the project config (#346).
Fixed
- The
ghtool is only targeted at a remote whose host is exactlygithub.com: the string no longer matches inside a longer hostname (github.com.evil.io), the userinfo or another forge's path, andhost:portURLs parse correctly (#352). tole approvalsnow honorsTOLE_SERVE_URL(precedence: --url > env > default http://127.0.0.1:7801) as its help text always claimed (#339).- CodeCora home is resolved in one place: skills and the update-check cache no longer fall back to the current directory when HOME is unset, and CODECORA_HOME now means the same thing everywhere (
$CODECORA_HOME/tole/...; the update-check cache moves from$CODECORA_HOME/.codecora/toleto$CODECORA_HOME/tole) (#344).
tole v0.7.2
Scan backlog, systemone error bodies, bare-profile build, dependency updates.
Fixed
- systemone_decide surfaces the provider's 4xx/5xx error body (bounded) instead of "no error body" (#328).
- tole-cli: the bare
--no-default-featuresprofile now compiles (#330); CI now
checks that profile andtole-core --no-default-features --testsso neither can regress.
Changed
- Dependency updates:
rmcp3.4.1 → 3.5.0,tokio1.53.1 → 1.53.2,
libc0.2.189 → 0.2.190,uuid1.26.1 → 1.27.0 (lockfile only). - Docs: the
write_fileparent-component symlink race is now documented as a
residual limitation andgit status/diffare documented as repo-wide by
design; the previous "TOCTOU-safe" wording overclaimed (#329). - CI: Dependabot/Renovate PRs are exempt from the Branch Naming and PR
Description checks (#322).
tole v0.7.1
Scan-gate hardening, tool-call authorization gate, SSRF and approval-prompt fixes.
Changed
- Approval prompts now show control characters, bidi overrides and line separators in tool descriptions as visible escapes (shared sanitizer in ToolRegistry::decide) (#327).
- Unknown-tool aborts (fresh and guarded replay) now settle the turn to
Finallike every other abort (#316). A baretole resume <id>after one reports "nothing to resume"; send a new prompt instead. The Tier 1 abort-path contract moved from "parks at Planning, resume continues" to "settles to Final, next prompt continues". serveapproval-resume now honors/cancel, andBudgetExhaustedmaps tomax_tokensthere too (#318).- Tool specs sent to the provider now use a static
Tool::summary()(defaulted method) instead ofdescribe(Null); git and systemone were being advertised with validation error strings (#320). - The MCP server path (
RegistryServer::execute_checked) now authorizes tool
calls through the same crate-internal gate asdrive/resume_turn.
Behavior change forRegistryServer::newembedders: pre-hooks configured on
an embedder-supplied registry are now enforced there (a configured deny-hook
must not be bypassable on one path). No in-repo face is affected, since
none can attach pre-hooks. Destructive refusal and error strings are
unchanged (#303, part 3 of 3). - Internal refactor, no behavior change:
driveandresume_turnnow share
one crate-internal tool-call authorization gate (gate.rs) with a typed
denial and aPermitrequired to execute non-ReadOnly tools; tool risk is
read once per call. Durable entry shapes and error strings are unchanged
(#303, part 2 of 3). - Behavior change:
agent_pollis nowRisk::Write(a successful poll
consumes the mailbox: it writesmeta.jsonand forgets mailbox memories),
matching the rule that a ReadOnly poll never mutates.--trust internal
allowlists the exact nameagent_poll, so poll loops stay prompt-free and
agent_startstill prompts; users on theread_onlypreset who approved
agent_startare now prompted on everyagent_poll; plan mode drops it
together withagent_start(#300).
Fixed
- Security: server-supplied text is sanitized before it reaches MCP approval
prompts (#286); the subprocess env scrubber matches SECRET/TOKEN/
PASSWORD/API_KEY-shaped names by substring (#284,TOKENIZERS_PARALLELISM
kept); memoryrecallpasses the query after--(#280); a failed or
timed-out request-body read on the MCP HTTP face is a 400, not an empty
body (#277); MCP registration sync waits longer than the serial reactor's
worst-case queueing (#285). - Redirects:
web_fetchresolves relativeLocations per RFC 3986 against
the current hop and refuses non-http(s) hops (#281). gitrejects NUL bytes in argv with a clear error (#282).- Behavior change:
AllowlistApprover::new(patterns, Deny|Ask)now
ignorespatterns— the default verdict is final (fail closed). Embedders
that usednew(p, Deny)as an allowlist must useallow_only(p)(#283). - Behavior change: the
web_fetchSSRF guard now also refuses 100.64.0.0/10
(CGNAT / Tailscale-style overlay networks), 0.0.0.0/8, 192.0.0.0/24,
198.18.0.0/15, multicast (224.0.0.0/4, ff00::/8), reserved 240.0.0.0/4,
site-local fec0::/10, and 6to4/Teredo addresses that embed a private IPv4.
Fetching a host on a tailnet now requiresTOLE_WEB_ALLOW_PRIVATE=1(#317). tole acp: a hanging/modelsendpoint no longer re-stalls every
session/new(probe wait 60 s → 5 s, failures cached for 30 s, probe runs
before the sessions lock is taken) (#319).chatregisters and hydratestodo_read/todo_write(#276); an agent
mailbox is marked consumed only after a successful recall (#287).- Tests: the perf
resume_replaygate actually replays 25 tool calls and
asserts it (#278);turn_looptemp dirs are unique per call (#279).
Architecture-review follow-ups (#293–#299):
- The
ghtool is registered only when a GitHub origin is detected; the
hardcodedcodecoradev/tolefallback on the run/chat/resume/mission and
tole mcpfaces is gone (#293). tole serve --transport mcphonors--plan-modein its server-level
registry (only ReadOnly tools remain) (#294).- One session-id rule everywhere:
[A-Za-z0-9_-], at most 64 bytes
(tole_core::storage::is_valid_session_id); an id accepted by serve/ACP
is no longer rejected bytole chat --resume/tole status(#295). tole acpcaps live sessions at 256 with non-busy eviction (shared helper
with serve and the MCP session tools) and prunes the approval state of
evicted sessions (#299).
tole v0.7.0
Mission mode, durable task list, update machinery.
Highlights
Mission mode (issues #198 / #199 / #201)
Long autonomous runs with a model-maintained durable task list — todo_write / todo_read persist in the session transcript, survive crashes (hydrate-on-open), and enforce one in_progress task at a time. Budget tiers (--max-steps/--max-tokens/--max-minutes, --trust internal headroom), verify gates (--verify with retry loop), and nonzero exit on any non-complete status for CI-style gating.
Update machinery (issue #220)
Startup banner when a newer crates.io release exists (24 h cache, TOLE_NO_UPDATE_CHECK=1 opt-out) and tole upgrade [--check] [--yes] with post-install binary verification.
Web tools (issue #215)
web_fetch / web_search — read-only internet, SSRF-guarded (private-range refusal, DNS rebinding checks), 512 KB cap, content-type allowlist.
Also in this train
- ACP streaming with plan visibility (#196), remote approvals (#200), run ergonomics:
--prompt-file,--name,--timeout(#216), mobile-control REST guide (#202), MCP-over-HTTP default (#168) with hardened framing (#190/#191/#193).
Quality gate
Pre-tag full-codebase cora scan: every MAJOR triaged and fixed across PRs #236–#245 and #261–#275 — per-session todo state, MCP session-registry-first routing (bogus id fails closed), argv depth-guard hardening (quoting / env -i / exec -c / substitution escapes), eval judges requiring exit 0 (v0.5.0 false-PASS eliminated), git stdout cap + pathspec-magic jail, poll/step dual budgets, storage id backstop, and more. One real regression (mission exit sentinel) was caught by the e2e suite before merge and fixed.
Full log: CHANGELOG.md
tole v0.6.0
Four faces on one durable core, hardened end-to-end.
Added
systemone_decide(#172/#173) — typed decisions (choice / score / noul + confidence) from any System One backend; active withSYSTEMONE_API_KEY,SYSTEMONE_BASE_URLselects the backend (hosted Jev default, self-hosted compatible).- Depth-1 child agents (#171/#174):
agent_start/agent_pollspawn and harvest durable child sessions — structurally no grandchildren, ephemeral uteke mailboxes, optional per-child git worktrees (--agents-worktree). - ACP native model picker + per-session approval controls (#176/#177/#179):
TOLE_MODELSadvertises a durable model switch; an approval selector (ask/auto) plusallow_alwaysrelax the gate from the editor without weakening the Destructive tier. - Cooperative cancellation on all three server faces (#178/#184/#185): ACP
session/cancel, RESTPOST /sessions/{id}/cancel, MCPtole_session_cancel— the turn settlesstopReason: "cancelled"as a normal durable end; single in-flight tool calls are not interrupted (spec-conformant). giteatool (#169) — the Gitea counterpart ofghover the instance REST API.- SKILL.md support (#161/#162),
--trustpresets (#160) with theTOLE_TRUSTenv actually read (#165/#166).
Changed
mcp-httpis a default feature oftole-cli(#168): a plaincargo install tole-cliexposes every documented face, includingtole serve --transport mcp.
Fixed / Hardened
- MCP-over-HTTP accept-loop hardening (#190/#191): 32-connection cap, 30s header-read timeout, 30s bounds on auth/body phases — slowloris parity with the REST face (#136).
- The HTTP/1 timer wiring (#193):
header_read_timeoutrequires a wired timer (hyper panics per connection otherwise) — caught by the pre-tag live smoke test, invisible to CI. H1-only preserved (no h2c sniffing). - Build/CI: the no-
mcpprofile compiles again (#175/#180); project-sync board-sync fixes with loud failures (#181–#183).
Validation
- Full-codebase cora scan (release gate): the single MAJOR fixed pre-tag; 351/351 workspace tests; QA-sandbox live smoke green (auth, SSE, cap enforcement, h2c refusal, recovery, zero panics); CI green on all trains.