TUI scrollback re-architecture, Lime chat surface, and deep-audit fixes#158
Merged
Conversation
One design remains in the tree ahead of the Lime rebuild: the zeroline package, its TUI view/skin plumbing (Options.Skin/ThemeVariant/ThemeDark, model skin branches, theme picker), and the zeroline CLI subcommand are gone. /theme keeps its shell-only message.
theme.go is rebuilt around the Lime design tokens (docs/design/ zero_tui_lime.html): near-black panels, one lime accent, solid tint stand-ins for the prototype's rgba overlays. Existing render sites map onto the new role styles (ink/line/userPrompt/toolName/toolArg) without layout changes; no hex exists outside theme.go.
Title bar with brand badge + context window, empty state replacing the deleted splash (block-art 0, tagline, 1-3 starter suggestions), stream blocks (user prompt gutter, muted interim with cursor, final-answer accent rail, done line, bordered notes), tool cards with diff/read/ bash/grep bodies and MiniDot run spinner, borderless composer with run-state hints, and the grouped status line. Final answers are marked at append time; resolved tool calls collapse into their result cards.
Adversarial review of the chat-surface commit confirmed 15 issues, all fixed here: - wrapPlainText sliced wide runes by rune count: CJK/emoji prose could panic the TUI or emit lines ~2x the measure; split at width instead - a cancelled run's streamingText leaked into the next turn's interim block; cancelRun now clears it - orphaned tool-call cards (cancelled/errored turns) re-animated with the spinner on any later run; the glyph is now scoped to the call's own runID - starter-suggestion digit capture was active while the chips were off screen (/clear mid-run); the guard now mirrors the render condition - prompt submits while a deferred Ctrl+C quit is draining could start a run the quit would orphan; submission is blocked while exiting - the running placeholder advertised ctrl+c (which quits) as the interrupt key; it now says esc - read-card parser expected 'N: text' but read_file emits 'N | text', so the gutter and L-range never rendered on real output - toolCard borders were one column short of the body rows - diff tint bands now span the full row; NEW FILE renders green on addBg; preamble/no-newline lines are unnumbered meta and no longer shift hunk numbering - card bodies and system notes paint the panel surface; the head gains the separate faintest arg column (grep target + pattern) - title-bar overflow now truncates the minimal candidate instead of rendering the deleted design's 'ZERO … READY' line - untracked the stray snake-game.html (left on disk, ignored)
Permission prompts render as the amber permission card (PERMISSION badge, risk readout, key chips); resolved prompts collapse to one line (allowed once/always/denied · tool) and unprompted allows fold into the tool card's [auto] tag. Ask-user and spec-review prompts share the card language with line borders. Autocomplete and picker overlays move onto the panel/selection tints with the accent ❯ marker; model-picker rows gain the provider dot and ctx · KEY_ENV metadata from the registry and provider catalog. /resume renders its list as stacked session cards (id + age, title, meta). Key handling is untouched everywhere.
Width tiers re-evaluate from the live width: ≥100 renders the full
spec; 80–99 drops the tool-arg column, header ctx, and the status
'interactive' group; 58–79 drops the diff/read gutters, bares the
badge, and keeps provider+tokens+mode; <58 renders a single-segment
header, rail-less cards, and a mode-only status line. Table-driven
tests cover {58,70,80,100,120} plus a frame-wide invariant that no
emitted line exceeds the terminal at any tier. Card bodies expand tabs
so width math holds. Dead splash-chip machinery (startupCommandNames,
startupOrder) is gone; README's TUI section now describes Lime.
A second adversarial review confirmed 14 issues, all addressed: - permission collapse and tool-card pairing are now run-scoped (rcKey): providers with repeating synthetic ToolCallIDs (Gemini gemini_tool_N) could attribute a decision or result to a different run's call - rehydrated always-grants no longer mislabel as 'allowed once' (GrantMatched fallback) - the width invariant now actually holds everywhere: empty-state tagline/hint, ask-user rows, permission detail blobs, pending image chips, and the sessions trailing hint are wrapped or fitted; the invariant test renders the empty state and all of those rows at widths 24-120 - every card (permission, ask-user, spec-review, sessions, notes) drops side borders on tiny terminals, not just tool cards - /resume card fields are sanitized against separator bytes from user-controlled titles - the picker's selected row paints its gap so the selBg band is solid - dead splash helpers (normalizedStartupWidth, countLines, nonEmpty, unreachable rowWelcome branch) and stale /theme comments are gone - README no longer claims transcript scrolling keys or Tab-accept behavior that never existed; chatWidth documents its 24-col floor Tag tui-lime-s4 moves to this commit.
…odule fixes TUI: settled-row flush frontier prints history to native terminal scrollback (tea.Println, ordered + ack-serialized); View() renders only the live tail. Full diffs (400-line flush cap) and OSC 8 file:// links land in history; streamed interim text, cancellation markers, ask_user exchanges, and composer input history are preserved; run-scoped dedup keys fix Gemini's repeating tool-call ids; /exit, Ctrl+C-after-Esc, and cross-session late flushes no longer orphan or contaminate checkpoints. Cross-module: zero search panic + offset mapping, usage report robustness, 529 retry, SSE keep-alive watchdog, registry corrections, update --check on dev builds, MCP ping + mcp list --json secret redaction, sandbox.network config knob, doctor apiKey presence, atomic config writes, C1-safe notifications, rune-safe truncation everywhere, gofmt + vet CI gates, dead glamour dependency dropped. Full audit report: docs/audit/2026-06-10-deep-audit.md (175 findings). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (4)
🚧 Files skipped from review as they are similar to previous changes (3)
WalkthroughAdds a Lime single-column chat design and interactive prototype; removes the legacy zeroline renderer; refactors TUI model, rendering, flush, and session resume flows; enforces UTF‑8-safe truncation across modules; tightens tool schemas and agent max-turn behavior; adds gofmt/go vet CI checks and extensive Lime rendering/width tests. ChangesLime TUI rebuild
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45–90 minutes Possibly related PRs
Suggested reviewers
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
internal/zerogit/zerogit.go (1)
201-211:⚠️ Potential issue | 🟠 Major | ⚡ Quick winInconsistent byte vs. rune limits between ValidateMessage and truncateSubject.
ValidateMessagerejects subjects whenlen(firstLine) > 72(bytes), buttruncateSubjectnow counts and truncates on rune boundaries (line 467:runes := []rune(value)). This means:
- A 72-byte ASCII subject passes validation and
truncateSubjectleaves it intact.- A 60-byte subject with multi-byte Unicode (say, 50 runes) passes validation but
truncateSubjectstill leaves it intact.- A generated 72-rune subject with multi-byte chars (say, 100 bytes) would fail validation even though
truncateSubjectproduced it.The comment at line 465 says the old 72-byte limit "rejected valid non-ASCII subjects," which is correct. But now validation still uses the byte limit while generation uses runes. Consider updating
ValidateMessageto count runes (or at least document the intentional divergence if byte-length matters for git tooling).Also applies to: 464-472
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@internal/zerogit/zerogit.go` around lines 201 - 211, ValidateMessage currently enforces a 72-byte limit using len(firstLine) but truncateSubject operates on rune boundaries, causing inconsistent behavior for non-ASCII input; update ValidateMessage to count runes (e.g., use utf8.RuneCountInString or convert to []rune) and enforce a 72-rune limit on firstLine so it matches truncateSubject (or alternatively add a clear comment in ValidateMessage explaining a deliberate byte-based limit), ensuring the change references the ValidateMessage function and the existing truncateSubject behavior.
🧹 Nitpick comments (2)
internal/imageinput/imageinput.go (1)
62-62: 💤 Low valueUnchecked
file.Closereturn value (linter hint).For a read-only file, ignoring the Close error is defensible since the data was already read and there's no write buffer to flush. If you want to silence the linter, you can assign to a named return or use
defer func() { _ = file.Close() }().🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@internal/imageinput/imageinput.go` at line 62, The defer currently calls defer file.Close() with its return value ignored; update the defer in the function that opens the file (where file.Close() is used) to explicitly handle or discard the error to satisfy the linter — e.g., replace defer file.Close() with defer func() { _ = file.Close() }() or capture and log the error from file.Close() inside that deferred wrapper so the linter is satisfied while preserving current behavior.Source: Linters/SAST tools
internal/tui/command_center.go (1)
106-122: ⚡ Quick winConsider sanitizing additional control characters in card fields.
sanitizeCardFieldcurrently strips only\x1f(field separator),\n, and\x00(prefix marker). User-controlled fields likesession.Titlecan legally contain tabs, carriage returns, or other whitespace/control characters that might disrupt card rendering. Consider normalizing all control characters (e.g.,r < 0x20 || r == 0x7f) to spaces, similar to the approach ininternal/notify/notify.golines 134-139.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@internal/tui/command_center.go` around lines 106 - 122, sanitizeCardField currently only strips sessionsCardFieldSep, newline, and NUL but should normalize all ASCII control characters to avoid rendering issues; update sanitizeCardField to iterate runes in value and replace any r where r < 0x20 || r == 0x7f with a space (preserving other characters), and then also remove the sessionsCardFieldSep and NUL as before—use the same control-character normalization approach as in internal/notify/notify.go (lines ~134-139) and keep the function name sanitizeCardField and constants sessionsCardFieldSep/sessionsCardsPrefix unchanged.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@internal/modelregistry/catalog.go`:
- Line 36: The catalog entry for the deprecated Anthropic model in
anthropicModel("claude-haiku-3.5", "Claude Haiku 3.5",
"claude-3-5-haiku-20241022", ...) incorrectly omits ModelCapabilityVision even
though claude-3-5-haiku-20241022 is multimodal; update the capabilities slice
for that entry to include ModelCapabilityVision (i.e., add ModelCapabilityVision
alongside ModelCapabilityPromptCache) or alternatively add an explicit
comment/documentation in the same entry explaining why this model should be
treated as text-only if that is intentional.
In `@internal/update/update.go`:
- Around line 137-142: The currentVersion fallback in internal/update/update.go
indiscriminately coerces any parse error to "0.0.0"; instead, in the error
handling where currentVersion is set after a failed parse (the block that
currently sets currentVersion = "0.0.0"), restrict the fallback to known dev
markers only (e.g., if the original version string is "dev", "devel", or empty)
and for all other parse errors return or propagate an error (do not silently
coerce). Update the code around the currentVersion assignment and the parsing
error path so that non-dev malformed versions cause the function to return an
error, while dev markers still map to "0.0.0" to preserve --check behavior.
---
Outside diff comments:
In `@internal/zerogit/zerogit.go`:
- Around line 201-211: ValidateMessage currently enforces a 72-byte limit using
len(firstLine) but truncateSubject operates on rune boundaries, causing
inconsistent behavior for non-ASCII input; update ValidateMessage to count runes
(e.g., use utf8.RuneCountInString or convert to []rune) and enforce a 72-rune
limit on firstLine so it matches truncateSubject (or alternatively add a clear
comment in ValidateMessage explaining a deliberate byte-based limit), ensuring
the change references the ValidateMessage function and the existing
truncateSubject behavior.
---
Nitpick comments:
In `@internal/imageinput/imageinput.go`:
- Line 62: The defer currently calls defer file.Close() with its return value
ignored; update the defer in the function that opens the file (where
file.Close() is used) to explicitly handle or discard the error to satisfy the
linter — e.g., replace defer file.Close() with defer func() { _ = file.Close()
}() or capture and log the error from file.Close() inside that deferred wrapper
so the linter is satisfied while preserving current behavior.
In `@internal/tui/command_center.go`:
- Around line 106-122: sanitizeCardField currently only strips
sessionsCardFieldSep, newline, and NUL but should normalize all ASCII control
characters to avoid rendering issues; update sanitizeCardField to iterate runes
in value and replace any r where r < 0x20 || r == 0x7f with a space (preserving
other characters), and then also remove the sessionsCardFieldSep and NUL as
before—use the same control-character normalization approach as in
internal/notify/notify.go (lines ~134-139) and keep the function name
sanitizeCardField and constants sessionsCardFieldSep/sessionsCardsPrefix
unchanged.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: c11a0439-31f6-41e7-bd0a-2983509154bf
⛔ Files ignored due to path filters (1)
go.sumis excluded by!**/*.sum
📒 Files selected for processing (76)
.github/workflows/ci.yml.gitignoreREADME.mddocs/audit/2026-06-10-deep-audit.mddocs/design/ZERO_LIME_TUI_REBUILD_PROMPT.mddocs/design/zero_tui_lime.htmlgo.modinternal/agent/system_prompt.gointernal/cli/app.gointernal/cli/app_test.gointernal/cli/cron.gointernal/cli/exec.gointernal/cli/exec_sessions.gointernal/cli/exec_writer.gointernal/cli/extensions.gointernal/cli/sandbox.gointernal/cli/usage.gointernal/cli/zeroline.gointernal/config/resolver.gointernal/config/types.gointernal/config/writer.gointernal/doctor/doctor.gointernal/imageinput/imageinput.gointernal/mcp/server.gointernal/modelregistry/catalog.gointernal/notify/notify.gointernal/providers/providerio/providerio.gointernal/providers/providerio/retry.gointernal/sandbox/safe_command_test.gointernal/search/search.gointernal/sessions/replay.gointernal/tui/ask_user_test.gointernal/tui/autocomplete_test.gointernal/tui/command_center.gointernal/tui/commands.gointernal/tui/commands_test.gointernal/tui/flush.gointernal/tui/flush_test.gointernal/tui/image_attach.gointernal/tui/image_attach_test.gointernal/tui/model.gointernal/tui/model_test.gointernal/tui/options.gointernal/tui/picker.gointernal/tui/picker_test.gointernal/tui/rendering.gointernal/tui/rendering_lime_test.gointernal/tui/run.gointernal/tui/session.gointernal/tui/session_controls.gointernal/tui/session_controls_test.gointernal/tui/session_test.gointernal/tui/spec_mode.gointernal/tui/spec_mode_test.gointernal/tui/startup.gointernal/tui/startup_test.gointernal/tui/theme.gointernal/tui/transcript.gointernal/tui/tui_fixes_test.gointernal/tui/view.gointernal/tui/width_tiers_test.gointernal/tui/zeroline_view.gointernal/tui/zeroline_view_test.gointernal/update/update.gointernal/zerocommands/backend_snapshots.gointernal/zerocommands/backend_snapshots_test.gointernal/zerocommands/sandbox_snapshots.gointernal/zerogit/zerogit.gointernal/zerogit/zerogit_test.gointernal/zeroline/markdown.gointernal/zeroline/markdown_test.gointernal/zeroline/render.gointernal/zeroline/render_overlay_test.gointernal/zeroline/render_test.gointernal/zeroline/theme.gointernal/zeroline/tokens_test.go
💤 Files with no reviewable changes (14)
- internal/zeroline/tokens_test.go
- internal/zeroline/theme.go
- internal/tui/zeroline_view_test.go
- internal/zeroline/render_test.go
- internal/tui/commands_test.go
- internal/zeroline/markdown_test.go
- internal/tui/options.go
- internal/zeroline/render_overlay_test.go
- internal/cli/zeroline.go
- internal/tui/zeroline_view.go
- internal/zeroline/markdown.go
- internal/cli/app_test.go
- internal/zeroline/render.go
- internal/tui/tui_fixes_test.go
There was a problem hiding this comment.
🧹 Nitpick comments (1)
internal/tools/ask_user_test.go (1)
61-72: ⚡ Quick winRefactor test to check actual field values instead of JSON string matching.
The string-based verification doesn't confirm that
minItemsandminLengthare on the correct fields—it only checks that these strings appear somewhere in the marshaled JSON. A more robust test would directly inspect the schema structure.♻️ Proposed refactor to check schema fields directly
func TestAskUserToolAdvertisesNonEmptyQuestionSchema(t *testing.T) { - data, err := json.Marshal(NewAskUserTool().Parameters()) - if err != nil { - t.Fatalf("marshal schema: %v", err) - } - schema := string(data) - for _, want := range []string{`"minItems":1`, `"minLength":1`} { - if !strings.Contains(schema, want) { - t.Fatalf("ask_user schema missing %s: %s", want, schema) - } + schema := NewAskUserTool().Parameters() + + // Verify questions array has minItems constraint + questions := schema.Properties["questions"] + if questions.MinItems == nil || *questions.MinItems != 1 { + t.Fatal("questions array should have minItems:1") + } + + // Verify question field has minLength constraint + questionField := questions.Items.Properties["question"] + if questionField.MinLength == nil || *questionField.MinLength != 1 { + t.Fatal("question field should have minLength:1") } }🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@internal/tools/ask_user_test.go` around lines 61 - 72, The test TestAskUserToolAdvertisesNonEmptyQuestionSchema currently checks for substrings in the marshaled JSON; instead unmarshal NewAskUserTool().Parameters() into a map[string]interface{} (or typed struct), navigate to the schema fields under "properties" (e.g., locate properties["question"]["minLength"] and properties["choices"]["minItems"] or similar names the tool uses) and assert their numeric values equal 1; update assertions to fail with clear messages if those specific fields are missing or have the wrong value. Ensure you use NewAskUserTool().Parameters() as the source and reference the exact property keys when checking minLength and minItems.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In `@internal/tools/ask_user_test.go`:
- Around line 61-72: The test TestAskUserToolAdvertisesNonEmptyQuestionSchema
currently checks for substrings in the marshaled JSON; instead unmarshal
NewAskUserTool().Parameters() into a map[string]interface{} (or typed struct),
navigate to the schema fields under "properties" (e.g., locate
properties["question"]["minLength"] and properties["choices"]["minItems"] or
similar names the tool uses) and assert their numeric values equal 1; update
assertions to fail with clear messages if those specific fields are missing or
have the wrong value. Ensure you use NewAskUserTool().Parameters() as the source
and reference the exact property keys when checking minLength and minItems.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: f2611a7f-e6ce-4bc2-bbdc-76a349ca0bd0
📒 Files selected for processing (5)
internal/tools/ask_user.gointernal/tools/ask_user_test.gointernal/tools/types.gointernal/tui/rendering.gointernal/tui/rendering_lime_test.go
🚧 Files skipped from review as they are similar to previous changes (2)
- internal/tui/rendering_lime_test.go
- internal/tui/rendering.go
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
internal/config/resolver.go (1)
70-76:⚠️ Potential issue | 🟠 MajorAdd resolver tests for
sandbox.networkallow/deny validation
internal/config/resolver_test.gohas nosandbox.network/NetworkAllow/NetworkDenycases and no test asserts the resolver error"invalid sandbox.network %q: expected allow or deny".- Existing
network: denyassertions are for CLI effective-policy output/snapshots and don’t exercise the resolver’s invalid-value branch.Add resolver tests for both the success paths (
allow/deny) and the fail-loud invalid path (matching theTestResolveSandboxMaxAutonomyInvalidFailsLoudpattern).🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@internal/config/resolver.go` around lines 70 - 76, Add unit tests in internal/config/resolver_test.go that exercise the sandbox.network validation: create resolver test cases that set cfg.Sandbox.Network to "allow" and "deny" and assert the resolver returns success (no error) and produced ResolvedConfig, and add a failing case (following the TestResolveSandboxMaxAutonomyInvalidFailsLoud pattern) that sets cfg.Sandbox.Network to an invalid value and asserts the resolver returns an error matching "invalid sandbox.network %q: expected allow or deny". Reference the cfg.Sandbox.Network field, the sandbox.NetworkAllow/sandbox.NetworkDeny enum values, and the resolver code path that returns ResolvedConfig{} on error to locate where to call into the resolver and assert the expected outcomes.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In `@internal/config/resolver.go`:
- Around line 70-76: Add unit tests in internal/config/resolver_test.go that
exercise the sandbox.network validation: create resolver test cases that set
cfg.Sandbox.Network to "allow" and "deny" and assert the resolver returns
success (no error) and produced ResolvedConfig, and add a failing case
(following the TestResolveSandboxMaxAutonomyInvalidFailsLoud pattern) that sets
cfg.Sandbox.Network to an invalid value and asserts the resolver returns an
error matching "invalid sandbox.network %q: expected allow or deny". Reference
the cfg.Sandbox.Network field, the sandbox.NetworkAllow/sandbox.NetworkDeny enum
values, and the resolver code path that returns ResolvedConfig{} on error to
locate where to call into the resolver and assert the expected outcomes.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: 8acc413e-a509-4a5b-b503-fe2aec3c57ff
📒 Files selected for processing (4)
internal/agent/loop.gointernal/agent/loop_test.gointernal/config/resolver.gointernal/config/resolver_test.go
🚧 Files skipped from review as they are similar to previous changes (1)
- internal/agent/loop_test.go
Contributor
Zero automated PR reviewVerdict: No blockers found Blockers
Validation
ScopeHead: This deterministic review checks validation status and basic diff hygiene. A human reviewer still owns product judgment and design quality. |
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@internal/agent/guardrails.go`:
- Around line 197-202: The code only resets
state.toolOnlyTurns/state.toolOnlyReminderSent when hasVisibleText is true, so
empty non-tool turns keep the tool-only streak; change the control flow around
the hasToolCalls and hasVisibleText checks so any non-tool-only turn resets the
streak: treat "not (hasToolCalls && !hasVisibleText)" as the reset case and in
that branch set state.toolOnlyTurns = 0 and state.toolOnlyReminderSent = false;
update the block around hasToolCalls/hasVisibleText in guardrails.go (the code
that references state.toolOnlyTurns and state.toolOnlyReminderSent) accordingly.
In `@internal/tools/shell_runtime.go`:
- Around line 20-21: The windowsBashStyleCDPattern is too permissive and matches
cmd.exe's valid "cd /d ..." usage, and the " ls -" substring check in
internal/tools/bash.go is too broad; tighten the regex in
windowsBashStyleCDPattern to reject the special "/d" switch (e.g. use a negative
lookahead: cd\s+/(?!d\b)[a-z0-9_./~-]+ with (?i) for case-insensitivity) and
replace the naive substring check for " ls -" in the pre-execution blocker with
a word-boundary or whitespace-aware regex (e.g. (?i)(^|\s)ls\b\s+-) so only
actual "ls" commands with flags are matched; update references in
internal/tools/shell_runtime.go (windowsBashStyleCDPattern) and
internal/tools/bash.go (the pre-execution check) accordingly.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: 38d6661b-8f49-48a5-b649-829831714890
📒 Files selected for processing (8)
internal/agent/guardrails.gointernal/agent/guardrails_test.gointernal/agent/loop.gointernal/tools/bash.gointernal/tools/bash_tool_test.gointernal/tools/shell_runtime.gointernal/tui/rendering.gointernal/tui/rendering_lime_test.go
🚧 Files skipped from review as they are similar to previous changes (3)
- internal/tools/bash_tool_test.go
- internal/agent/loop.go
- internal/tui/rendering.go
Vasanthdev2004
approved these changes
Jun 10, 2026
Collaborator
|
@coderabbitai approve |
✅ Action performedComments resolved and changes approved. |
This was referenced Jun 10, 2026
Merged
This was referenced Jun 15, 2026
This was referenced Jul 1, 2026
This was referenced Jul 10, 2026
This was referenced Jul 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR lands the Lime TUI work rebased onto current
main, plus a full deep audit of the codebase with fixes for the highest-impact findings.TUI: real terminal scrollback (settled-row flush frontier)
The chat surface previously rendered the whole transcript inside Bubble Tea's managed View — the inline renderer clips anything taller than the terminal, so history above one screen was unrecoverable (scrolling up showed pre-launch screen contents). Now:
tea.Println(strictly ordered, ack-serialized);View()renders only the live tail (running tool cards, undecided prompts, streaming text, composer, status). Scroll-up, selection, and copy of full history work natively.file://hyperlinks — cmd/ctrl+click opens the edited file (URLs percent-encoded; ANSI truncation is OSC-aware and re-closes open links)./resumeand/rewindflush rehydrated history to scrollback in one batch; ask_user questions and answers persist and rehydrate.TUI correctness fixes
gemini_tool_N) no longer silently drop tool cards, live or on/resume./exitmid-run and Ctrl+C-after-Esc no longer orphan rewind checkpoint blobs (quit deferred until the cancelled run's session events flush);/specgated while exiting./resume-ing another session can't be contaminated;/rewindis blocked while a flush is outstanding; cancelled-run usage is recorded.looksLikeDiffno longer hijacks output containing---separators; rune-safe truncation of titles and tool output; dead footer/actions/theme code removed.Cross-module audit fixes
zero search: panic + invalid-UTF-8 context fixed (lowered-text offsets mapped back to the original); unknown--session-iderrors instead of silently returning 0 hits; corrupt sessions skipped, not fatal.zero usage report: corrupt sessions skipped; works outside a git repository.zero update --checkworks on dev/source builds.pingper spec;zero mcp list --jsonstructurally redacts env/header values.sandbox.network: allow|denyconfig knob — the engine'sNetworkDenydefault was previously unreachable from any config surface, making "Sandbox violation [network]" impossible to opt out of.zero doctorreports apiKey presence instead of an unconditional[REDACTED]; atomic config writes; C1 control bytes stripped from OSC-9 notifications; imageinput surfaces real open/read errors; rune-safe truncation across exec/stream-json/cron/sessions/zerogit/agent system prompt.glamourdependency dropped.Audit report
docs/audit/2026-06-10-deep-audit.md— 175 findings across all 44 packages with evidence and suggested fixes; the "Fixed in this pass" section maps to this PR, the rest are open for follow-up (top items: MCP stdio framing, hooks never dispatched, reasoning effort never serialized, specialist child permissions, bash timeout kill semantics, macOS sandbox-exec profile).Test plan
go test ./...green (new regression tests for the flush frontier, dedup keys, input history, UTF-8 cuts, OSC truncation, wrap indentation, diff detection)go vet ./...andgofmt -lclean; both now enforced in CI🤖 Generated with Claude Code
Summary by CodeRabbit
New Features
Bug Fixes
Documentation