fix(context): recover from provider overflow

[omarluq]

No description provided.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 05e41707-d2f3-4b10-a0a2-805f296f3866

📥 Commits

Reviewing files that changed from the base of the PR and between 65bbfd4 and 22c7b3a.

📒 Files selected for processing (2)

• internal/assistant/context_overflow_compaction_test.go
• internal/assistant/runtime_model.go

🚧 Files skipped from review as they are similar to previous changes (1)

• internal/assistant/runtime_model.go

---

📝 Walkthrough

Summary by CodeRabbit

• New Features

  • Automatic provider-overflow recovery: compacts conversation context, emits compaction events, rebuilds and retries the completion once, records compaction metadata, and preserves the original provider error when no compaction occurs; optional rebuilt-budget validation.

• Improved Detection

  • Enhanced detection of provider "context window" errors to reliably trigger recovery.

• Refactor

  • Test harness and runtime setup reorganized for clearer test construction.

• Tests

  • Expanded coverage and helpers for overflow recovery, event emission, error paths, input guards, and detection logic.

Walkthrough

Routes completions through a recovery flow that detects provider context-window errors, compacts session context, rebuilds the completion request, optionally validates rebuilt budget, emits compaction/usage events, and retries once; adds IsContextWindowError and comprehensive tests.

Changes

Provider Context Window Overflow Recovery

Layer / File(s)	Summary
Context window error detection helpers internal/assistant/retry.go	Adds IsContextWindowError, normalizedErrorMessage, and contextWindowProviderMessage and refactors error normalization used by retry logic.
Overflow recovery data structures and functions internal/assistant/context_overflow_compaction.go	Adds providerOverflowRecoveryInput, Runtime.completeWithProviderOverflowRecovery, and Runtime.recoverProviderContextOverflow to detect context-window failures, emit compaction events, compact session context (CompactSessionFrom), rebuild completion requests (buildCompletionRequest), emit rebuilt usage, optionally validate rebuilt budget, and retry completion.
Runtime model completion routing internal/assistant/runtime_model.go	modelResponse now calls completeWithProviderOverflowRecovery instead of completeWithRetry, enabling overflow recovery for model completions.
Tests, harness updates, and test helpers internal/assistant/context_overflow_compaction_test.go, internal/assistant/context_auto_compaction_*.go, internal/assistant/export_test.go, internal/assistant/retry_test.go	Adds integration and unit tests for overflow recovery, IsContextWindowError cases, recovery error paths and input guards, test clients/helpers (including recordingCompletionClient), small test-harness refactors, and three exported test helpers on Runtime to exercise the recovery entry points.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

• omarluq/librecode#46: Related retry/error-classification refactors in internal/assistant/retry.go.
• omarluq/librecode#83: Related work on context compaction and request rebuilding; overlaps at compaction + rebuild integration.

Poem

🐰 When tokens crowd and providers say "too wide",
I nibble branches, keep the meaning inside.
Rebuild the prompt from a compacted root,
Retry once gently — hop back to the route.
Tests hum in burrows, proving recovery's stride.

🚥 Pre-merge checks | ✅ 3 | ❌ 1

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Description check	❓ Inconclusive	No pull request description was provided by the author, making it impossible to assess relevance to the changeset.	Add a pull request description explaining the purpose, behavior, and testing of the provider overflow recovery implementation.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'fix(context): recover from provider overflow' clearly and concisely summarizes the main change: adding provider overflow recovery functionality to handle context-window exhaustion.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/context-overflow-compaction

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

❌ Patch coverage is 93.54839% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 69.24%. Comparing base (e1a0488) to head (22c7b3a).

Files with missing lines	Patch %	Lines
internal/assistant/context_overflow_compaction.go	91.30%	4 Missing and 2 partials ⚠️
internal/assistant/retry.go	94.59%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #87      +/-   ##
==========================================
+ Coverage   69.11%   69.24%   +0.12%     
==========================================
  Files         214      215       +1     
  Lines       18998    19104     +106     
==========================================
+ Hits        13131    13229      +98     
- Misses       4717     4722       +5     
- Partials     1150     1153       +3     

Flag	Coverage Δ
unittests	69.24% <93.54%> (+0.12%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[coderabbitai]

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/assistant/context_overflow_compaction.go`:
- Around line 62-85: Change the compaction event flow so the pre-compaction
event is clearly an attempt and the post-compaction event is emitted from the
rebuilt context state: update the first runtime.emitContextCompaction call
(before runtime.CompactSessionFrom) to use wording like "attempting compaction
before retry" so it doesn't promise success, call runtime.CompactSessionFrom as
now, then on success emit the post-compaction event using compactionMessage with
the recovered session/build budget from the rebuilt state (use the budget value
on recoveredEntry rather than input.build.Budget) and only emit that
post-success event after err == nil; leave the
isCompactNothingToDoError(providerErr) early-return behavior unchanged but
ensure it does not cause a misleading pre-compaction success message.

In `@internal/assistant/retry.go`:
- Around line 258-266: The current contextWindowProviderMessage function is too
permissive because the "token limit" substring matches quota/billing or
plan-limit errors; narrow this by replacing the generic "token limit" pattern
with more specific context-window phrases (e.g., "token limit exceeded for
request", "token limit for this request", or "context token limit") and add
explicit negative/regression checks to ignore billing/quota messages such as
"daily token limit exceeded", "quota exceeded", "billing", or "plan limit" so
IsContextWindowError/nonRetryableProviderMessage won't misclassify those errors;
update the patterns slice in contextWindowProviderMessage accordingly and add
unit tests covering a regression case like "daily token limit exceeded".

🪄 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: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 265f782a-a101-424f-a442-7ddb0b4938d4

📥 Commits

Reviewing files that changed from the base of the PR and between e1a0488 and ba463ab.

📒 Files selected for processing (4)

• internal/assistant/context_overflow_compaction.go
• internal/assistant/context_overflow_compaction_test.go
• internal/assistant/retry.go
• internal/assistant/runtime_model.go

[coderabbitai]

Actionable comments posted: 1

🤖 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/assistant/context_overflow_compaction_test.go`:
- Around line 129-139: The test expectations assert the inner error codes
instead of the runtime wrapper codes returned by recoverProviderContextOverflow;
update the failing cases that use providerOverflowFailingSummaryClient and
providerOverflowLargeSummaryClient to expect the wrapped codes: replace wantCode
"compact_summarize" with "context_overflow_compact" for the
providerOverflowFailingSummaryClient case, and replace the inner provider code
(testContextWindowExceededOopsCode) with
"context_budget_after_provider_overflow_compact" for the
providerOverflowLargeSummaryClient case — also update the identical expectation
in the other occurrence noted (the second test block around the same clients) so
all tests reflect the wrapper error codes from recoverProviderContextOverflow.

🪄 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: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 54d15641-70d6-4a50-baa1-c43d5e2b5cfc

📥 Commits

Reviewing files that changed from the base of the PR and between ba463ab and 67b15e2.

📒 Files selected for processing (7)

• internal/assistant/context_auto_compaction_internal_extra_test.go
• internal/assistant/context_auto_compaction_test.go
• internal/assistant/context_overflow_compaction.go
• internal/assistant/context_overflow_compaction_test.go
• internal/assistant/export_test.go
• internal/assistant/retry.go
• internal/assistant/retry_test.go

🚧 Files skipped from review as they are similar to previous changes (2)

• internal/assistant/retry.go
• internal/assistant/context_overflow_compaction.go

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud