The local text read path opened the same file for every consumer. A ranged
read of a file within the snapshot cap cost four opens and three decodes:
an 8KiB binary sniff, a streaming scan for the rendered window, a whole-file
read for bracket context, and another whole-file read to hash the snapshot.
Whole-file reads under the structural summarizer paid a fifth. Two of those
readers also ran normalizeToLF over the same bytes.
Read the bytes once at or below SNAPSHOT_MAX_BYTES and derive every view
from them: sniff the leading 8KiB of the buffer, slice the rendered window
out of it under the identical line and byte budgets, index bracket context
into its addressable lines, and hand the normalized text to the snapshot
store and the summarizer. Past the cap nothing wants the whole file, so the
streaming reader stays.
Line byte lengths are walked out of the buffer rather than measured on the
decoded strings, so reported byte counts and the truncation boundary stay
exact for content that is not valid UTF-8.
The buffered text is BOM-stripped for hashing, matching the decoder the
patcher's live read uses. A whole-file read of a BOM file previously hashed
its tag from BOM-bearing text, so the following edit only applied through
stale-hash recovery and told the model the file had changed externally when
it had not.
Also stop rejoining lines into a fresh whole-file string on the way to
tree-sitter when the caller still holds that text, and drop the unread
selectedBytesTotal accounting from the streaming reader.
Measured on 966 differential cases across CRLF, BOM, lone-CR, invalid-UTF-8,
oversized-line, empty, no-trailing-newline, multi-range and raw shapes: the
spurious recovery warning is the only behavioral difference. Raw reads, which
skip the tree-sitter parse that dominates everything else, get 30-45% faster
(2.7MB: 10.3ms -> 5.7ms); non-raw reads 1-3%.
Retry: fixed changelog bundle probe asserting latest release equals VERSION (fails on releases with no coding-agent changelog content); widened issue-4593 watchdog test budgets from 5ms to 50ms against CI runner scheduling noise.
collect_boundaries walked every node in the file even though the answer is
bounded by the visible window, which cost roughly twice the parse: on an
81KB source the walk was 8.95ms against a 4.32ms parse, and on 1MB it was
138.8ms against 87.6ms.
A node contributes a boundary only when one of its own endpoint lines is
visible, and both of those lines lie inside its raw row span. Every
descendant's span is contained in its ancestor's, so a span holding no
visible line rules out that node and everything beneath it. Skip such
subtrees with a binary search over the merged visible ranges.
The test is the raw span, not endpoint visibility: a node whose span merely
straddles the window has both endpoints outside it yet can contain a child
that opens exactly on a visible line. The raw span is also conservative
relative to node_content_end_line, so the prune needs no reasoning about
that newline adjustment.
Equivalence is proven differentially rather than argued: the pre-prune walk
is retained under cfg(test) and compared for exact Option<Vec<u32>> equality
across 4827 .ts/.py/.rs files and 38,616 comparisons over eight window
shapes, including whole-file-visible, past-EOF, disjoint ranges, empty range
lists and files that fail to parse. Zero mismatches. root.has_error() is
still evaluated on the whole tree before the walk, so pruning cannot change
a None verdict.
Measured on the built addon with a mid-file 40-line window, medians of 20,
against the parse cache alone: 81KB 13.4ms -> 4.45ms cold and 9.04ms ->
0.149ms warm; 1.06MB 188.1ms -> 55.8ms cold and 131.7ms -> 0.440ms warm.
`enclosing_block_boundaries`, `block_range_at` and `summarize_code` each
re-parsed the whole file on every call. The results are not cacheable —
boundaries depend on the caller's visible ranges, which differ per call —
but the `tree_sitter::Tree` is, so cache that instead and hand out
`ts_tree_copy` clones.
Keyed on (xxh64 of the source, source length, language). The hash is a
bucket selector only: a hit re-verifies the stored source against the
request byte-for-byte before returning the tree, so a collision costs a
re-parse and can never yield a tree built from other content. Language is
in the key because the same bytes parsed as TypeScript and as Python are
different trees.
Bounded at 12 slots and 4 MiB of retained source with LRU eviction;
sources above 4 MiB are parsed but never retained. `Tree` is `Send` but
not `Sync`, so entries sit behind a `Mutex` that is held only for a map
probe, a byte compare and a refcount bump, never across a parse or walk.
Error trees are cached like any other: `has_error()` is a property of the
tree, so the callers' own checks reach an identical verdict from a cached
tree, and repeated "does this parse" probes get the speedup too.
Measured (M4 Max, bazel-built .node, median of 20, 1-40 visible):
read.ts 81 KB 13.34 ms -> 8.86 ms on repeat; 1 MB synthetic 225.7 ms ->
138.3 ms. Parser::new + set_language measured at 0.30 us against a
3.91 ms parse, so no parser pooling.
The previous CI reds were a truncated rust-overlay fetch in Evaluate
flake and a Julia kernel shutdown in julia-prelude.test.ts. Neither
path is in this change.
`lsp regressions > detects pyright and pylsp in Windows virtualenv Scripts for
Python-only roots` fails on any machine that has ~/.omp/agent/lsp.json:
expect(config.servers[server]?.resolvedCommand).toBe(localBin)
Expected: ".../.venv/Scripts/pyright-langserver.exe"
Received: undefined
The test was not asserting against the packaged defaults at all. loadConfig
walks the user config dirs (~/.omp/agent, ~/.pi/agent, ~/.claude) via
getConfigDirPaths, which resolves from os.homedir(). Any user lsp.json with a
`servers` block sets hasOverrides, which takes loadConfig off its auto-detect
branch and onto the override branch, where the user's rootMarkers replace the
packaged ones.
On this machine that file overrides pyright with
rootMarkers: ["pyproject.toml", "uv.lock", "requirements.txt", "setup.py"]
which contains neither `pyrightconfig.json` nor `setup.cfg` — precisely the
two markers the test creates. loadConfig therefore returned zero servers.
`ruff` is absent from that file, keeps the packaged rootMarkers, and its
sibling tests pass, which is why only this one failed.
Confirmed by probing inside the test: hasRootMarkers() true and
resolveCommand() returning the .exe, while loadConfig().servers was {} — the
detection helpers were fine, the branch was not.
Point os.homedir() at an empty directory for every test in the file so they
see a pristine environment. Bun's os.homedir() reads the passwd entry rather
than $HOME, so the env var alone does not redirect the walk; both are set.
Verification: 76 pass / 0 fail (was 75 / 1) on a machine with a user lsp.json.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A selector like anthropic/claude-opus-5 is both the anthropic provider's
canonical selector and, verbatim, an OpenRouter model id. When the named
provider is out of the candidate set (disabled, missing creds at boot), the
exact provider-scoped reference misses and the raw-id flat match re-binds the
request onto OpenRouter's same-named model. Requests that should fail closed
instead bill the aggregator at per-token Claude prices.
Raw-id fallback is only legitimate when the named provider does not carry the
id (openai/gpt-4o:extended). Lock the reference when it does by checking the
bundled catalog, then fail rather than shadow. openrouter/anthropic/claude-opus-5
still selects OpenRouter explicitly.
Cursor grok-4.6-xhigh stalled after a short "I'll fetch the page"
preamble because interaction_query frames (including proto field 9)
were dropped and the server waited until the 300s idle watchdog fired.
isMultiplexerSession() treats TMUX, STY, ZELLIJ, HERDR_ENV=1, the CMUX_*
markers and a tmux/screen TERM as authoritative, and routes rendering down the
path that cannot rebuild scrollback. Nine tests across four files assert the
destructive full-paint behavior and fail for anyone running the suite inside
tmux, screen, Zellij or CMUX.
component-render.test.ts already cleared HERDR_ENV for exactly this reason but
covered only one of the nine signals. Replace it with a shared helper that
clears the whole family and restores it afterwards, and call it from the other
three files.
Three tests fail on a clean checkout depending on where the repo lives and
which terminal runs the suite:
- status-line-path builds its fake home inside the checkout, so a clone under
/tmp lands in a SCRATCH_ROOTS prefix and renders the scratch icon.
- status-line/component renders a fixed 120 columns, so a long checkout path
or branch name pushes the cost segment out of the assertion.
- bash-executor runs an interactive login zsh, which loads the system
/etc/zshrc; under Apple Terminal that appends session-save lines to the
captured output. HOME does not isolate a system-level file.
- Updated GPT-5.6 context window floor to 1,000,000 tokens across discovery, policies, and tests.
- Updated model configurations and pricing parameters in catalog models JSON.
A developer shell with ANTHROPIC_BASE_URL set reroutes the effective endpoint
away from official, switching off eager tool-input streaming, long cache
retention, the Cowork TLS profile, the Claude Code session header and priority
service tier. 14 tests across 6 files assert those behaviors and fail on a
clean checkout.
Add withOfficialAnthropicEndpoint(), a beforeEach/afterEach pair that removes
the variable and restores it, and call it from the six affected files.
The ask tool stores the user-attached note in result details (note on the
single-question shape, per-item note inside results[]), but the shared
tool renderer dropped it: AskAnswer carried no note field and the card
never rendered one. Extract and render it as a muted row so HTML exports
and the collab guest view keep the user's context.
omp update re-threw Bun's raw fetch() UnsupportedProxyProtocol error, telling CLI users to pass verbose:true to fetch() — an instruction unavailable through the CLI. The update fetch catches now detect this failure and report which proxy env var uses an unsupported scheme plus the http/https requirement.
Fixes#8784
Force process termination via `postmortem.quit(0)` in
`Completions.run()` after writing completion scripts to stdout. Loading
all command modules during completion generation leaves open event loop
handles (sockets, timers) that prevent natural process exit, causing
tools like chezmoi to hang.
Per review on #8717: the customSystemPrompt assignment ran after the hook,
so when both options.customSystemPrompt and an extension payload
replacement were set, the option silently won. Now the option is applied
before the hook (the extension can inspect or drop it in its replacement),
matching anthropic, where the hook runs right before serialization and is
the last word on the wire body.
Adds regression tests: replacement drops customSystemPrompt, replacement
overrides it, and the option still applies when the hook returns undefined.
OpenCode substitutes {env:VAR} and {file:path} in config text at load
time, but OMP's OpenCode discovery ran the generic ${VAR}-only
expandEnvVarsDeep, leaving those tokens literal. An MCP header like
`Bearer {env:MCP_KEY}` reached the server verbatim and returned 401.
The OpenCode loader now applies OpenCode's own substitution to raw
config text before parsing: {env:VAR} -> env value or empty string,
{file:path} -> trimmed, JSON-escaped file contents resolved relative to
the config dir / ~ / absolute, skipping tokens on // comment lines.
Fixes#8778
The Home Manager module symlinked ~/.omp/agent/config.yml to a read-only
/nix/store path. OMP acquires an advisory lock on and atomically rewrites
its config when persisting runtime changes; on macOS the lock backend
creates an flock sidecar next to the target, so the first startup save
failed with "Failed to acquire native file lock ... Permission denied
(os error 13)", breaking every launch once programs.omp.settings was set.
Copy the generated config into place as a writable regular file via
home.activation instead. OMP can now lock and rewrite it, and the next
home-manager switch reapplies the declared settings. The DAG entry is
written literally so the home-manager-free module evaluation in flake.nix
keeps working; that check now asserts the activation entry.
Fixes#8775
usageReservePct scoped limits through scopeClaudeLimitsForModelHardBlock,
which drops a Fable/Mythos weekly tier row until confirmed exhaustion
(>=100% or server exhausted). That guard is correct for credential-wide
hard blocks but wrong for the opt-in, non-destructive reserve fallback:
a tier row at 96% was removed before reserve health, so the model stayed
healthy and kept serving past the configured margin.
Added a scopeLimitsForReserve strategy hook (falls back to scopeLimits)
and pointed the Claude strategy at scopeClaudeLimitsForModel, so reserve
health honors the mapped tier row while credential hard blocks and all
other providers are unchanged.
Fixes#8773
Point xai and xai-oauth at grok-4.6, already in the bundled catalog.
Tests pin the default id in models.json and load picker fixtures from
the catalog so the next bump does not rot hardcoded name or cost.
Current Finder Cmd+C pasteboards advertise both a public.file-url and a
generated 1024x1024 file-icon bitmap. arboard::get_image() succeeded with
the icon, so handleImagePaste attached it before readMacFileUrlsFromClipboard
was reached, sending a generic PNG document icon to vision models instead of
the copied screenshot.
Probe the Darwin file URLs before the bitmap representation so an image-file
URL wins over the co-advertised icon. Pure bitmap pasteboards (screenshots,
browser copies) and non-image file URLs still fall through to the image/text
paths. readMacFileUrls is a no-op off Darwin.
Fixes#8769
Strict Anthropic-compatible endpoints (Z.AI GLM at api.z.ai/api/anthropic)
reject a whole request when a tool_result block carries content: [] ,
returning 400 code 1213 "The prompt parameter was not received normally".
The official API accepts both shapes, so the empty array only surfaced on
compatible endpoints once a tool returned empty output on a vision-capable
model (text-only models already encode the joined empty string).
Normalize the empty array to "" at encode time, alongside the existing
error-placeholder normalization from #2250.
A ThinkingLoop abort is the loop guard asking for a same-model resample
(it injects a thinking-loop-redirect notice that only makes sense on the
model that looped), not a provider failure. #handleRetryableError routed
it through the generic retryable-error branch, so on attempt 1 it called
noteRetryFallbackCooldown + #tryRetryModelFallback and could switch to
another family from retry.fallbackChains while parking the original
selector on a 5-minute cooldown. A healthy Grok 4.6 planning turn got
replaced by whatever the chain listed next.
Carve ThinkingLoop out of the model-fallback branch and out of the
Fireworks Fast->base degrade so the loop guard always re-samples the same
model; the retry budget still bounds a genuinely stuck stream.
Fixes#8760
Review noted that the option whitelist still let generic short flags
escape: `rm -rf -v /` and `rm -rf -i /` were not classified critical,
which is the same separator class the change set out to close.
Pin only the recursive/force flag and skip any other options on either
side of it, which also removes the need to enumerate long options. An
absolute target is still required, so `rm -rf -- ./build`,
`rm --recursive --force ./dist` and `rm -v /tmp/scratch` remain benign
and are asserted.
`CRITICAL_BASH_PATTERNS` required the target to follow one short flag
cluster directly, so anything in between escaped the check:
rm -rf / matched
rm -rf -- / missed
rm --recursive --force / missed
rm -rf --no-preserve-root / missed
rm --no-preserve-root -rf / missed
The last two matter most. GNU coreutils already refuses `rm -rf /` with
"it is dangerous to operate recursively on '/'" and names
`--no-preserve-root` as the override, so the pattern matched the form
that fails safe and missed the form that does not.
Repeat the option separator instead of assuming the path follows one
cluster, and treat `--no-preserve-root` as critical wherever it appears.
Absolute targets are still required for the first pattern, so
`rm -rf -- ./build` and `rm --recursive --force ./dist` stay benign;
both are asserted in the test.
PR #5270 reported the `--` and long-option forms in July and was closed
unmerged by the contributor-vouch bot rather than on merit. This keeps
its cases, credits them in the tests, and adds the `--no-preserve-root`
forms that patch did not cover.
Add grok-4.6 to the SuperGrok Responses effort allowlist so /model
can select low/medium/high/xhigh. Stale omitReasoningEffort cache
rows no longer hide the dial. max is omitted because api.x.ai 400s.
Biome 2.x fully supports CSS parsing, formatting, linting, and assist
actions via its CLI and LSP proxy. The built-in Biome client config
did not include .css in its fileTypes array, causing OMP to skip Biome
for CSS files even when the project's biome.json enables CSS rules.
This change adds .css to the default fileTypes so that projects with
CSS files automatically route through Biome without needing a per-project
override of the full fileTypes array.
Closes#8735
A plugin installed as a bun git dependency contains its full workspace
tree but no node_modules links: bun materializes a git dep's regular npm
dependencies into the host tree and skips workspace:*/file: edges. Bare
imports between workspace siblings inside such a plugin therefore failed
with 'Cannot find module' - the extension host's package walk only
looked at node_modules dirs.
Teach findNodePackageRoot to also recognize workspace roots on the
walk-up: when a directory's package.json declares workspaces (array or
yarn-style { packages }), scan member manifests for the requested name.
node_modules candidates at the same level still win, so explicitly
installed copies shadow workspace members. Exports conditions (bun ->
source) apply to the member manifest as usual, so unbuilt source-only
members resolve.
Contract tests: a fake installed monorepo tree resolves a bare member
import to the member's bun-condition source entry; an installed
node_modules copy shadows the member.
opencode-go and opencode-zen share loginOpenCode, which hardcoded "Paste your OpenCode Zen API key" and generic instructions. Selecting OpenCode Go therefore prompted for an OpenCode Zen key. loginOpenCode now takes the provider display name and each provider passes its own, so Go asks for a Go key while still opening the shared opencode.ai/auth console where Go keys are minted.
Fixes#8738
resolveUpdateTargetFromPath gated the symlink->realpath resolution on
allowPackageManagers, so binary-only releases (a major bump or an explicit
omp.dist: "binary") wrote to the raw PATH symlink instead of the binary it
resolves to. On an admin shared install where /usr/local/bin/omp is a
root-owned symlink into a group-writable dir this either failed with EACCES
(writing <link>.new into the root-owned dir) or replaced the symlink with a
full copy, stranding the shared install behind a split-brain second binary.
Target selection no longer depends on the release's distribution channel: a
foreign symlink (a non-manager alias, an admin symlink into a shared install)
resolves to its real binary and the launcher is left intact in every channel.
A package-manager launcher (bun/npm) keeps its deliberate in-place takeover on
a binary-only release, detected by re-classifying the launcher as if managers
were allowed rather than by the channel flag. The manager bin dirs are probed
in the binary channel only when the launcher is a symlink, so plain-file
installs stay probe-free.
Fixes#8732
Every live tool block (streaming args, a running partial tool, or a task
subagent) armed its own 80ms setInterval driving requestComponentRender,
so N concurrent live blocks created N unsynchronized repaint timers that
kept the 30fps render scheduler awake near-continuously during
multi-subagent / workflowz / orchestrate work.
Replaced the per-component intervals with a single shared spinner ticker
that advances the already-phase-locked glyph once per step and repaints
every registered live block in one coalesced frame, independent of block
count. Verified 3 live blocks now share 1 timer instead of 3.
Fixes#8731
smokeTestDaemonBroker mkdtemp'd its runtime dir directly under
os.tmpdir(), and the broker's startup sweep pruneDeadDaemonRuntimeDirs
reclaims path.dirname(runtimeDir). On a default session that is /tmp, so
--smoke-test recursively deleted every aged sibling with no live
broker/clients (tmux/ssh sockets, editor state, build trees) while still
exiting 0.
- client.ts: keep the smoke broker's project and runtime dirs under a
single private mkdtemp parent the process owns.
- presence.ts: refuse any prune root that is not the daemons container
and only prune entries named like a 16-hex daemon scope key, so a
relocated runtime dir can never rm -rf unrelated neighbours.
Fixes#8721
Gemini thought summaries occasionally emit a bare ```thinking / ``````thinking
opener line as a between-summary delimiter. consumeGoogleStream appended
thought-part text verbatim to ThinkingContent, and structured thought parts
bypass the visible-channel leaked-reasoning healers, so the delimiter reached
both live display and persisted transcripts as fence spam.
Route thought-part text through a streaming ThinkingFenceStripper that drops
only a standalone reasoning-fence opener line (>=3 backticks + thinking/
reasoning). Language-tagged code fences, bare closers, and inline mentions are
preserved.
Fixes#8719
prepareCompaction walked the branch from the last compaction and ignored reset_boundary markers, so /compact (and auto-compaction) resurrected pre-/clear turns into the summary even though buildSessionContext already starts the model context after the boundary.
Model reset_boundary as a first-class agent-core session entry and start the summarization window after the latest boundary, dropping the superseded pre-reset compaction summary. A boundary before the last compaction stays superseded by it.
Fixes#8718
The onPayload hook contract (README, docs/extensions.md) is that a non-undefined
return replaces the provider request payload, and every provider except these
three implements it (anthropic, openai-responses family, google, ollama — see
the earlier fix for the responses providers). openai-completions, amazon-bedrock
and cursor invoked the hook fire-and-forget and sent the original payload, so
extensions hooking before_provider_request could never transform the wire body
on these providers.
- openai-completions: await the hook and apply a non-undefined replacement to
the params used for the request body, raw request dump and error-path
fallback state
- amazon-bedrock: same for the ConverseStream command input
- cursor: await the hook for the AgentRunRequest; buildGrpcRequest becomes
async and is exported for direct testing (transport is HTTP/2)
- devin-agent intentionally unchanged: it does not fire the hook at all (its
payload is a protobuf object), which is a feature gap rather than a dropped
replacement; documented in README/docs instead
- regression tests: captured wire body reflects async/sync replacement, and an
undefined return keeps the original payload (completions + bedrock over a
mocked fetch; cursor by decoding the serialized run request)
The default snapcompact frame fonts (X.org 8x13 for every provider, plus the selectable 6x12 and legacy 5x8) drew digit zero as a bare oval visually indistinguishable from letter O. Image-based compaction OCRs the frames back, so 0 and O were mixed up and compacted identifiers (e.g. Slack IDs) got corrupted.
Zero now carries a disambiguating interior mark the O lacks: an ascending slash in 8x13 and a center bar in 6x12/5x8. unscii-8 (8x8/6x6u shapes) already shipped a slashed zero and is unchanged.
Fixes#8713
Sunil Srivatsa