Add new beads for MR orchestrator integration, sync run observability,
metrics collection, logging infrastructure, and CLI verbosity controls.
Update last-touched timestamp.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- prd-observability.md: Product requirements document for the sync pipeline
observability system, covering structured logging, metrics collection,
sync run tracking, and robot-mode performance output
- gitlore-sync-explorer.html: Self-contained interactive HTML visualization
for exploring sync pipeline stage timings and data flow
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Targeted fixes across multiple subsystems:
dependent_queue:
- Add project_id parameter to claim_jobs() for project-scoped job claiming,
preventing cross-project job theft during concurrent multi-project ingestion
- Add project_id parameter to count_pending_jobs() with optional scoping
(None returns global counts, Some(pid) returns per-project counts)
gitlab/client:
- Downgrade rate-limit log from warn to info (429s are expected operational
behavior, not warnings) and add structured fields (path, status_code)
for better log filtering and aggregation
gitlab/transformers/discussion:
- Add tracing::warn on invalid timestamp parse instead of silent fallback
to epoch 0, making data quality issues visible in logs
ingestion/merge_requests:
- Remove duplicate doc comment on upsert_label_tx
search/rrf:
- Replace partial_cmp().unwrap_or() with total_cmp() for f64 sorting,
eliminating the NaN edge case entirely (total_cmp treats NaN consistently)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add end-to-end observability to the sync and ingest pipelines:
Sync command:
- Generate UUID-based run_id for each sync invocation, propagated through
all child spans for log correlation across stages
- Accept MetricsLayer reference to extract hierarchical StageTiming data
after pipeline completion for robot-mode performance output
- Record sync runs in DB via SyncRunRecorder (start/succeed/fail lifecycle)
- Wrap entire sync execution in a root tracing span with run_id field
Ingest command:
- Wrap run_ingest in an instrumented root span with run_id and resource_type
- Add project path prefix to discussion progress bars for multi-project clarity
- Reset resource_events_synced_for_updated_at on --full re-sync
Sync status:
- Expand from single last_run to configurable recent runs list (default 10)
- Parse and expose StageTiming metrics from stored metrics_json
- Add run_id, total_items_processed, total_errors to SyncRunInfo
- Add mr_count to DataSummary for complete entity coverage
Orchestrator:
- Add #[instrument] with structured fields to issue and MR ingestion functions
- Record items_processed, items_skipped, errors on span close for MetricsLayer
- Emit granular progress events (IssuesFetchStarted, IssuesFetchComplete)
- Pass project_id through to drain_resource_events for scoped job claiming
Document regenerator and embedding pipeline:
- Add #[instrument] spans with items_processed, items_skipped, errors fields
- Record final counts on span close for metrics extraction
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Overhaul the CLI logging infrastructure for production observability:
CLI flags:
- Add -v/-vv/-vvv (--verbose) for progressive stderr verbosity control:
0=INFO, 1=DEBUG app, 2=DEBUG all, 3+=TRACE
- Add --log-format text|json for structured stderr output in automation
- Existing -q/--quiet overrides verbosity for silent operation
Subscriber architecture (main.rs):
- Replace single-layer subscriber with triple-layer setup:
1. stderr layer: human-readable or JSON, filtered by -v flags
2. file layer: always-on JSON to daily-rotated logs (lore.YYYY-MM-DD.log)
3. MetricsLayer: captures span timing for robot-mode performance payloads
- Parse CLI before subscriber init so verbosity is known at setup time
- Load LoggingConfig early (with graceful fallback for pre-init commands)
- Clean up old log files before subscriber init to avoid holding deleted handles
- Hold WorkerGuard at function scope to ensure flush on exit
Doctor command:
- Add logging health check: validates log directory exists, reports file
count and total size, warns on missing or inaccessible log directory
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Introduce the foundational observability layer for the sync pipeline:
- MetricsLayer: Custom tracing subscriber layer that captures span timing
and structured fields, materializing them into a hierarchical
Vec<StageTiming> tree for robot-mode performance data output
- logging: Dual-layer subscriber infrastructure with configurable stderr
verbosity (-v/-vv/-vvv) and always-on JSON file logging with daily
rotation and configurable retention (default 30 days)
- SyncRunRecorder: Compile-time enforced lifecycle recorder for sync_runs
table (start -> succeed|fail), with correlation IDs and aggregate counts
- LoggingConfig: New config section for log_dir, retention_days, and
file_logging toggle
- get_log_dir(): Path helper for log directory resolution
- is_permanent_api_error(): Distinguish retryable vs permanent API failures
(only 404 is truly permanent; 403/auth errors may be environmental)
Database changes:
- Migration 013: Add resource_events_synced_for_updated_at watermark columns
to issues and merge_requests tables for incremental resource event sync
- Migration 014: Enrich sync_runs with run_id correlation ID, aggregate
counts (total_items_processed, total_errors), and run_id index
- Wrap file-based migrations in savepoints for rollback safety
Dependencies: Add uuid (run_id generation), tracing-appender (file logging)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
11 isomorphic performance fixes from deep audit (no behavior changes):
- Eliminate double serialization: store_payload now accepts pre-serialized
bytes (&[u8]) instead of re-serializing from serde_json::Value. Uses
Cow<[u8]> for zero-copy when compression is disabled.
- Add SQLite cache_size (64MB) and mmap_size (256MB) pragmas
- Replace SELECT-then-INSERT label upserts with INSERT...ON CONFLICT
RETURNING in both issues.rs and merge_requests.rs
- Replace INSERT + SELECT milestone upsert with RETURNING
- Use prepare_cached for 5 hot-path queries in extractor.rs
- Optimize compute_list_hash: index-sort + incremental SHA-256 instead
of clone+sort+join+hash
- Pre-allocate embedding float-to-bytes buffer with Vec::with_capacity
- Replace RandomState::new() in rand_jitter with atomic counter XOR nanos
- Remove redundant per-note payload storage (discussion payload contains
all notes already)
- Change transform_issue to accept &GitLabIssue (avoids full struct clone)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The sync pipeline was bottlenecked at 10 req/s (hardcoded) with
sequential project processing and no retry on rate limiting. These
changes target 3-5x throughput improvement.
Rate limit configuration:
- Add requestsPerSecond to SyncConfig (default 30.0, was hardcoded 10)
- Pass configured rate through to GitLabClient::new from ingest
- Floor rate at 0.1 rps in RateLimiter::new to prevent panic on
Duration::from_secs_f64(1.0 / 0.0) — now reachable via user config
429 auto-retry:
- Both request() and request_with_headers() retry up to 3 times on
HTTP 429, respecting the retry-after header (default 60s)
- Extract parse_retry_after helper, reused by handle_response fallback
- After exhausting retries, the 429 error propagates as before
- Improved JSON decode errors now include a response body preview
Concurrent project ingestion:
- Derive Clone on GitLabClient (cheap: shares Arc<Mutex<RateLimiter>>
and reqwest::Client which is already Arc-backed)
- Restructure project loop to use futures::stream::buffer_unordered
with primary_concurrency (default 4) as the parallelism bound
- Each project gets its own SQLite connection (WAL mode + busy_timeout
handles concurrent writes)
- Add show_spinner field to IngestDisplay to separate the per-project
spinner from the sync-level stage spinner
- Error aggregation defers failures: all successful projects get their
summaries printed and results counted before returning the first error
- Bump dependentConcurrency default from 2 to 8 for discussion prefetch
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Two hardening changes to the dependent queue and orchestrator:
- dependent_queue::fail_job now propagates the rusqlite error via ?
instead of silently falling back to 0 attempts when the job row is
missing. A missing job is a real bug that should surface, not be
masked by unwrap_or(0) which would cause infinite retries at the
base backoff interval.
- orchestrator::enqueue_resource_events_for_entity_type replaces
format!-based SQL ("SELECT {id_col} FROM {table}") with separate
hardcoded queries per entity type. While the original values were
not user-controlled, hardcoded SQL is clearer about intent and
eliminates a class of injection risk entirely.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
tracing-indicatif pulled in vt100, arrayvec, and its own indicatif
integration layer. Replace it with a minimal SuspendingWriter that
coordinates tracing output with progress bars via a global LazyLock
MultiProgress.
- Add src/cli/progress.rs: shared MultiProgress singleton via LazyLock
and a SuspendingWriter that suspends bars before writing log lines,
preventing interleaving/flicker
- Wire all progress bar creation through multi().add() in sync and
ingest commands
- Replace IndicatifLayer in main.rs with SuspendingWriter for
tracing-subscriber's fmt layer
- Remove tracing-indicatif from Cargo.toml (drops vt100 and arrayvec
transitive deps)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
GitLab returns null for the label/milestone fields on resource_label_events
and resource_milestone_events when the referenced label or milestone has
been deleted. This caused deserialization failures during sync.
- Add migration 012 to recreate both event tables with nullable
label_name, milestone_title, and milestone_id columns (SQLite
requires table recreation to alter NOT NULL constraints)
- Change GitLabLabelEvent.label and GitLabMilestoneEvent.milestone
to Option<> in the Rust types
- Update upsert functions to pass through None values correctly
- Add tests for null label and null milestone deserialization
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
client.rs:
- fetch_all_resource_events() now uses tokio::try_join!() to fire all
three API requests (state, label, milestone events) concurrently
instead of awaiting each sequentially. For entities with many events,
this reduces wall-clock time by up to ~3x since the three independent
HTTP round-trips overlap.
main.rs:
- Removed async from handle_issues() and handle_mrs(). These functions
perform only synchronous database queries and formatting; they never
await anything. Removing the async annotation avoids the overhead of
an unnecessary Future state machine and makes the sync nature of
these code paths explicit.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
events_db.rs:
- Removed internal savepoints from upsert_state_events,
upsert_label_events, and upsert_milestone_events. Each function
previously created its own savepoint, making it impossible for
callers to wrap all three in a single atomic transaction.
- Changed signatures from &mut Connection to &Connection, since
savepoints are no longer created internally. This makes the
functions compatible with rusqlite::Transaction (which derefs to
Connection), allowing callers to pass a transaction directly.
orchestrator.rs:
- Deleted the three store_*_events_tx() functions (store_state_events_tx,
store_label_events_tx, store_milestone_events_tx) which were
hand-duplicated copies of the events_db upsert functions, created as
a workaround for the &mut Connection requirement. Now that events_db
accepts &Connection, store_resource_events() calls the canonical
upsert functions directly through the unchecked_transaction.
- Replaced the max-iterations guard in drain_resource_events() with a
HashSet-based deduplication of job IDs. The old guard used an
arbitrary 2x multiplier on total_pending which could either terminate
too early (if many retries were legitimate) or too late. The new
approach precisely prevents reprocessing the same job within a single
drain run, which is the actual invariant we need.
Net effect: ~133 lines of duplicated SQL removed, single source of
truth for event upsert logic, and callers control transaction scope.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Two bug fixes:
1. extractor.rs: The content hash was computed on the pre-truncation
content, meaning the hash stored in the document didn't correspond
to the actual stored (truncated) content. This would cause change
detection to miss updates when content changed only within the
truncated portion. Hash is now computed after truncate_hard_cap()
so it always matches the persisted content.
2. dependent_queue.rs: claim_jobs() had a TOCTOU race between the
SELECT that found available jobs and the UPDATE that locked them.
Under concurrent callers, two drain runs could claim the same job.
Replaced with a single UPDATE ... RETURNING statement that
atomically selects and locks jobs in one operation.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Three bugs fixed:
1. Early return in orchestrator when no discussions needed sync also
skipped resource event enqueue+drain. On incremental syncs (the most
common case), resource events were never fetched. Restructured to use
if/else instead of early return so Step 4 always executes.
2. Ingest command JSON and human-readable output silently dropped
resource_events_fetched/failed counts. Added to IngestJsonData and
print_ingest_summary.
3. Progress bar reuse after finish_and_clear caused indicatif to silently
ignore subsequent set_position/set_length calls. Added reset() call
before reconfiguring the bar for resource events.
Also removed stale comment referencing "unsafe" that didn't reflect
the actual unchecked_transaction approach.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Integrate resource event fetching as Step 4 of both issue and MR
ingestion, gated behind the fetch_resource_events config flag.
Orchestrator changes:
- Add ProgressEvent variants: ResourceEventsFetchStarted,
ResourceEventFetched, ResourceEventsFetchComplete
- Add resource_events_fetched/failed fields to IngestProjectResult
and IngestMrProjectResult
- New enqueue_resource_events_for_entity_type() queries all
issues/MRs for a project and enqueues resource_events jobs via
the dependent queue (INSERT OR IGNORE for idempotency)
- New drain_resource_events() claims jobs in batches, fetches
state/label/milestone events from GitLab API, stores them
atomically via unchecked_transaction, and handles failures
with exponential backoff via fail_job()
- Max-iterations guard prevents infinite retry loops within a
single drain run
- New store_resource_events() + per-type _tx helpers write events
using prepared statements inside a single transaction
- DrainResult struct tracks fetched/failed counts
CLI ingest changes:
- IngestResult gains resource_events_fetched/failed fields
- Progress bar repurposed for resource event fetch phase
(reuses discussion bar with updated template)
- Accumulates event counts from both issue and MR ingestion
CLI sync changes:
- SyncResult gains resource_events_fetched/failed fields
- Accumulates counts from both ingest stages
- print_sync() conditionally displays event counts
- Structured logging includes event counts
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Automated formatting and lint corrections from parallel agent work:
- cargo fmt: import reordering (alphabetical), line wrapping to respect
max width, trailing comma normalization, destructuring alignment,
function signature reformatting, match arm formatting
- clippy (pedantic): Range::contains() instead of manual comparisons,
i64::from() instead of `as i64` casts, .clamp() instead of
.max().min() chains, let-chain refactors (if-let with &&),
#[allow(clippy::too_many_arguments)] and
#[allow(clippy::field_reassign_with_default)] where warranted
- Removed trailing blank lines and extra whitespace
No behavioral changes. All existing tests pass unmodified.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds two new categories of integrity checks to 'lore stats --check':
Event FK integrity (3 queries):
- Detects orphaned resource_state_events where issue_id or
merge_request_id points to a non-existent parent entity
- Same check for resource_label_events and resource_milestone_events
- Under normal CASCADE operation these should always be zero; non-zero
indicates manual DB edits, bugs, or partial migration state
Queue health diagnostics:
- pending_dependent_fetches counts: pending, failed, and stuck (locked)
- queue_stuck_locks: Jobs with locked_at set (potential worker crashes)
- queue_max_attempts: Highest retry count across all jobs (signals
permanently failing jobs when > 3)
New IntegrityResult fields: orphan_state_events, orphan_label_events,
orphan_milestone_events, queue_stuck_locks, queue_max_attempts.
New QueueStats fields: pending_dependent_fetches,
pending_dependent_fetches_failed, pending_dependent_fetches_stuck.
Human output shows colored PASS/WARN/FAIL indicators:
- Red "!" for orphaned events (integrity failure)
- Yellow "!" for stuck locks and high retry counts (warnings)
- Dependent fetch queue line only shown when non-zero
All new queries are guarded by table_exists() checks for graceful
degradation on databases without migration 011 applied.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Extends the count command to support "events" as an entity type,
displaying resource event counts broken down by event type (state,
label, milestone) and entity type (issue, merge request).
New functions in count.rs:
- run_count_events: Creates DB connection and delegates to
events_db::count_events for the actual queries
- print_event_count: Human-readable table with aligned columns
showing per-type breakdowns and row/column totals
- print_event_count_json: Structured JSON matching the robot mode
contract with ok/data envelope and per-type issue/mr/total counts
JSON output structure:
{"ok":true,"data":{"state_events":{"issue":N,"merge_request":N,
"total":N},"label_events":{...},"milestone_events":{...},"total":N}}
Updated exports in commands/mod.rs to expose the three new public
functions (run_count_events, print_event_count, print_event_count_json).
The "events" branch in handle_count (main.rs, committed earlier)
routes to these functions before the existing entity type dispatcher.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
New module src/core/dependent_queue.rs provides job queue operations
against the pending_dependent_fetches table. Designed for second-pass
fetches that depend on primary entity ingestion (resource events,
MR close references, MR file diffs).
Queue operations:
- enqueue_job: Idempotent INSERT OR IGNORE keyed on the UNIQUE
(project_id, entity_type, entity_iid, job_type) constraint.
Returns bool indicating whether the row was actually inserted.
- claim_jobs: Two-phase claim — SELECT available jobs (unlocked,
past retry window) then UPDATE locked_at in batch. Orders by
enqueued_at ASC for FIFO processing within a job type.
- complete_job: DELETE the row on successful processing.
- fail_job: Increments attempts, calculates exponential backoff
(30s * 2^(attempts-1), capped at 480s), sets next_retry_at,
clears locked_at, and records the error message. Reads current
attempts via query with unwrap_or(0) fallback for robustness.
- reclaim_stale_locks: Clears locked_at on jobs locked longer than
a configurable threshold, recovering from worker crashes.
- count_pending_jobs: GROUP BY job_type aggregation for progress
reporting and stats display.
Registers both events_db and dependent_queue in src/core/mod.rs.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
New module src/core/events_db.rs provides database operations for
resource events:
- upsert_state_events: Batch INSERT OR REPLACE for state change events,
keyed on UNIQUE(gitlab_id, project_id). Wraps in a savepoint for
atomicity per entity batch. Maps GitLabStateEvent fields including
optional user, source_commit, and source_merge_request_iid.
- upsert_label_events: Same pattern for label add/remove events,
extracting label.name for denormalized storage.
- upsert_milestone_events: Same pattern for milestone assignment events,
storing both milestone.title and milestone.id.
All three upsert functions:
- Take &mut Connection (required for savepoint creation)
- Use prepare_cached for statement reuse across batch iterations
- Convert ISO timestamps via iso_to_ms_strict for ms-epoch storage
- Propagate rusqlite errors via the #[from] LoreError::Database path
- Return the count of events processed
Supporting functions:
- resolve_entity_ids: Maps entity_type string to (issue_id, MR_id) pair
with exactly-one-non-NULL invariant matching the CHECK constraints
- count_events: Queries all three event tables with conditional COUNT
aggregations, returning EventCounts struct. Uses unwrap_or((0, 0))
for graceful degradation when tables don't exist (pre-migration 011).
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Extends GitLabClient with methods for fetching resource events from
GitLab's per-entity API endpoints. Adds a new impl block containing:
- fetch_all_pages<T>: Generic paginated collector that handles
x-next-page header parsing with fallback to page-size heuristics.
Uses per_page=100 and respects the existing rate limiter via
request_with_headers. Terminates when: (a) x-next-page header is
absent/stale, (b) response is empty, or (c) page is not full.
- Six typed endpoint methods:
- fetch_issue_state_events / fetch_mr_state_events
- fetch_issue_label_events / fetch_mr_label_events
- fetch_issue_milestone_events / fetch_mr_milestone_events
- fetch_all_resource_events: Convenience method that fetches all three
event types for an entity (issue or merge_request) in sequence,
returning a tuple of (state, label, milestone) event vectors.
Routes to issue or MR endpoints based on entity_type string.
All methods follow the existing client patterns: path formatting with
gitlab_project_id and iid, error propagation via Result, and rate
limiter integration through the shared request_with_headers path.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds a new boolean field to SyncConfig that controls whether resource
event fetching is performed during sync:
- SyncConfig.fetch_resource_events: defaults to true via serde
default_true helper, serialized as "fetchResourceEvents" in JSON
- SyncArgs.no_events: --no-events CLI flag that overrides the config
value to false when present
- SyncOptions.no_events: propagates the flag through the sync pipeline
- handle_sync_cmd: mutates loaded config when --no-events is set,
ensuring the flag takes effect regardless of config file contents
This follows the existing pattern established by --no-embed and
--no-docs flags, where CLI flags override config file defaults.
The config is loaded as mutable specifically to support this override.
Also adds "events" to the count command's entity type value_parser,
enabling `lore count events` (implementation in a separate commit).
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds six new types for deserializing responses from GitLab's three
Resource Events API endpoints (state, label, milestone):
- GitLabStateEvent: State transitions with optional user, source_commit,
and source_merge_request reference
- GitLabLabelEvent: Label add/remove events with nested GitLabLabelRef
- GitLabMilestoneEvent: Milestone assignment changes with nested
GitLabMilestoneRef
- GitLabMergeRequestRef: Lightweight MR reference (iid, title, web_url)
- GitLabLabelRef: Label metadata (id, name, color, description)
- GitLabMilestoneRef: Milestone metadata (id, iid, title)
All types derive Deserialize + Serialize and use Option<T> for nullable
fields (user, source_commit, color, description) to match GitLab's API
contract where these fields may be null.
Includes 8 new test cases covering:
- State events with/without user, with/without source_merge_request
- Label events for add and remove actions, including null color handling
- Milestone event deserialization
- Standalone ref type deserialization (MR, label, milestone)
Uses r##"..."## raw string delimiters where JSON contains hex color
codes (#FF0000) that would conflict with r#"..."# delimiters.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Introduces five new tables that power temporal queries (timeline,
file-history, trace) via GitLab Resource Events APIs:
- resource_state_events: State transitions (opened/closed/reopened/merged/locked)
with actor tracking, source commit, and source MR references
- resource_label_events: Label add/remove history per entity
- resource_milestone_events: Milestone assignment changes per entity
- entity_references: Cross-reference table (Gate 2 prep) linking
source/target entity pairs with reference type and discovery method
- pending_dependent_fetches: Generic job queue for resource_events,
mr_closes_issues, and mr_diffs with exponential backoff retry
All event tables enforce entity exclusivity via CHECK constraints
(exactly one of issue_id or merge_request_id must be non-NULL).
Deduplication handled via UNIQUE indexes on (gitlab_id, project_id).
FK cascades ensure cleanup when parent entities are removed.
The dependent fetch queue uses a UNIQUE constraint on
(project_id, entity_type, entity_iid, job_type) for idempotent
enqueue, with partial indexes optimizing claim and retry queries.
Registered as migration 011 in the embedded MIGRATIONS array in db.rs.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- .claude/agents/test-runner.md: New Claude Code agent definition for
running cargo test suites and analyzing results, configured with
haiku model for fast execution.
- skills/agent-swarm-launcher/: New skill for bootstrapping coordinated
multi-agent workflows with AGENTS.md reconnaissance, Agent Mail
coordination, and beads task tracking.
- api-review.html, phase-a-review.html: Self-contained HTML review
artifacts for API audit and Phase A search pipeline review.
- .beads/issues.jsonl, .beads/last-touched: Updated issue tracker
state reflecting current project work items.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
AGENTS.md: Comprehensive rewrite adding file deletion safeguards,
destructive git command protocol, Rust toolchain conventions, code
editing discipline rules, compiler check requirements, TDD mandate,
MCP Agent Mail coordination protocol, beads/bv/ubs/ast-grep/cass
tool documentation, and session completion workflow.
README.md: Document NO_COLOR/CLICOLOR env vars, --since 1m duration,
project resolution cascading match logic, lore health and robot-docs
commands, exit codes 17 (not found) and 18 (ambiguous match),
--color/--quiet global flags, dirty_sources and
pending_discussion_fetches tables, and version command git hash output.
docs/embedding-pipeline-hardening.md: Detailed spec covering the three
problems from the chunk size reduction (broken --full wiring, mixed
chunk sizes in vector space, static dedup multiplier) with decision
records, implementation plan, and acceptance criteria.
docs/phase-b-temporal-intelligence.md: Draft planning document for
transforming gitlore from a search engine into a temporal code
intelligence system by ingesting structured event data from GitLab.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Three targeted regression tests covering bugs fixed in the embedding
pipeline hardening:
- overflow_doc_with_error_sentinel_not_re_detected_as_pending: verifies
that documents skipped for producing too many chunks have their
sentinel error recorded in embedding_metadata and are NOT returned by
find_pending_documents or count_pending_documents on subsequent runs
(prevents infinite re-processing loop).
- count_and_find_pending_agree: exercises four states (empty DB, new
document, fully-embedded document, config-drifted document) and
asserts that count_pending_documents and find_pending_documents
produce consistent results across all of them.
- full_embed_delete_is_atomic: confirms the --full flag's two DELETE
statements (embedding_metadata + embeddings) execute atomically
within a transaction.
Also updates test DB creation to apply migration 010.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add --full / --no-full flag pair to EmbedArgs with overrides_with
semantics matching the existing flag pattern. When active, atomically
DELETEs all embedding_metadata and embeddings before re-embedding.
- Thread the full flag through run_embed -> run_sync so that
'lore sync --full' triggers a complete re-embed alongside the full
re-ingest it already performed.
- Add indicatif spinners to sync stages with dynamic stage numbering
that adjusts when --no-docs or --no-embed skip stages. Spinners are
hidden in robot mode.
- Update robot-docs manifest to advertise the new --full flag on the
embed command.
- Replace hardcoded schema version 9 in health check with the
LATEST_SCHEMA_VERSION constant from db.rs.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Reduces CHUNK_MAX_BYTES from 32KB to 6KB and CHUNK_OVERLAP_CHARS from
500 to 200 to stay within nomic-embed-text's 8,192-token context
window. This commit addresses all downstream consequences of that
reduction:
- Config drift detection: find_pending_documents and
count_pending_documents now take model_name and compare
chunk_max_bytes, model, and dims against stored metadata. Documents
embedded with stale config are automatically re-queued.
- Overflow guard: documents producing >= CHUNK_ROWID_MULTIPLIER chunks
are skipped with a sentinel error recorded in embedding_metadata,
preventing both rowid collision and infinite re-processing loops.
- Deferred clearing: old embeddings are no longer cleared before
attempting new ones. clear_document_embeddings is deferred until the
first successful chunk embedding, so if all chunks fail the document
retains its previous embeddings rather than losing all data.
- Savepoints: each page of DB writes is wrapped in a SQLite savepoint
so a crash mid-page rolls back atomically instead of leaving partial
state (cleared embeddings with no replacements).
- Per-chunk retry on context overflow: when a batch fails with a
context-length error, each chunk is retried individually so one
oversized chunk doesn't poison the entire batch.
- Adaptive dedup in vector search: replaces the static 3x over-fetch
multiplier with a dynamic one based on actual max chunks per document
(using the new chunk_count column with a fallback COUNT query for
pre-migration data). Also replaces partial_cmp with total_cmp for
f64 distance sorting.
- Stores chunk_max_bytes and chunk_count (on sentinel rows) in
embedding_metadata to support config drift detection and adaptive
dedup without runtime queries.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add chunk_max_bytes and chunk_count columns to embedding_metadata to
support config drift detection and adaptive dedup sizing. Includes a
partial index on sentinel rows (chunk_index=0) to accelerate the drift
detection and max-chunk queries.
Also exports LATEST_SCHEMA_VERSION as a public constant derived from
the MIGRATIONS array length, replacing the previously hardcoded magic
number in the health check.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Extend resolve_project() with a 4th cascade step: case-insensitive
substring match when exact, case-insensitive, and suffix matches all
fail. This allows shorthand like "typescript" to match
"vs/typescript-code" when unambiguous. Multi-match still returns an
error with all candidates listed.
Also change ambiguity errors from LoreError::Other to LoreError::Ambiguous
so they get the proper AMBIGUOUS error code (exit 18) instead of
INTERNAL_ERROR.
Includes tests for unambiguous substring, case-insensitive substring,
ambiguous substring, and suffix-preferred-over-substring ordering.
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
Runtime setup:
- Reset SIGPIPE to SIG_DFL on Unix at the very start of main() so
piping to head/grep doesn't cause a panic.
- Apply --color flag to console::set_colors_enabled() after CLI parse.
- Extract quiet flag and thread it to handle_ingest.
Command dispatch:
- Add Completions match arm using clap_complete::generate().
- Resolve all --no-X negation flags in handlers: asc, has_due, open
(issues/mrs), force/full (ingest/sync), check (stats), explain
(search), retry_failed (embed).
- Auto-enable --check when --repair is used in handle_stats.
- Suppress deprecation warnings in robot mode for List, Show, AuthTest,
and SyncStatus deprecated aliases.
Stubs:
- Change handle_backup/handle_reset from ok:true to structured error
JSON on stderr with exit code 1. Remove unused NotImplementedOutput
and NotImplementedData structs.
Version:
- Include GIT_HASH env var in handle_version output (human and robot).
- Add git_hash field to VersionData with skip_serializing_if for None.
Robot-docs:
- Update exit code table with codes 14-18 (Ollama, NotFound, Ambiguous)
and code 20 (ConfigNotFound). Clarify code 1 and 2 descriptions.
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
Ingest:
- Introduce IngestDisplay struct with show_progress/show_text booleans
to decouple progress bars from text output. Replaces the robot_mode
bool parameter with explicit display control, enabling sync to show
progress without duplicating summary text (progress_only mode).
- Use resolve_project() for --project filtering instead of LIKE queries,
providing proper error messages for ambiguous or missing projects.
List:
- Add colored_cell() helper that checks console::colors_enabled() before
applying comfy-table foreground colors, bridging the gap between the
console and comfy-table crates for --color flag support.
- Use resolve_project() for project filtering (exact ID match).
- Improve since filter to return explicit errors instead of silently
ignoring invalid values.
- Improve format_relative_time for proper singular/plural forms.
Search:
- Validate --after/--updated-after with explicit error messages.
- Handle optional title field (Option<String>) in HydratedRow.
Show:
- Use resolve_project() for project disambiguation.
Sync:
- Thread robot_mode via SyncOptions for IngestDisplay selection.
- Use IngestDisplay::progress_only() in interactive sync mode.
GenerateDocs:
- Use resolve_project() for --project filtering.
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
Global flags:
- --color (auto|always|never) for explicit color control
- --quiet/-q to suppress non-essential output
- Hidden Completions subcommand for bash/zsh/fish/powershell
Flag negation (--no-X) with overrides_with for: has-due, asc, open
(issues/mrs), force/full (ingest/sync), check (stats), explain (search),
retry-failed (embed). Enables scripted flag composition where later flags
override earlier ones.
Validation:
- value_parser on search --mode, --type, --fts-mode for early rejection
- Remove requires="check" from --repair (auto-enabled in handler)
Polish:
- help_heading groups (Filters, Sorting, Output, Actions) on issues,
mrs, and search args for cleaner --help output
- Hide Backup, Reset, and Completions from --help
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
ConfigNotFound previously used exit code 2 which collides with clap's
usage error code. Remap it to exit 20 to avoid ambiguity. Also add
dedicated NotFound (exit 17) and Ambiguous (exit 18) error codes with
proper ErrorCode variants and Display implementations, replacing the
previous incorrect mapping of these errors to GitLabNotFound.
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
Add clap_complete for shell completion generation and libc (unix-only)
for SIGPIPE handling. Create build.rs to embed the git commit hash at
compile time via cargo:rustc-env=GIT_HASH, enabling `lore version` to
display the short hash alongside the version number.
Co-Authored-By: Claude (us.anthropic.claude-opus-4-5-20251101-v1:0) <noreply@anthropic.com>
Add new beads for the checkpoint-3 search pipeline work including
document generation, FTS5 indexing, embedding pipeline, hybrid search,
and CLI command implementations. Update status on completed beads.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- README.md: Add hybrid search and robot mode to feature list. Update
quick start to use new noun-first CLI syntax (lore issues, lore mrs,
lore search). Add embedding configuration section. Update command
examples throughout.
- AGENTS.md: Update robot mode examples to new CLI syntax. Add search,
sync, stats, and generate-docs commands to the robot mode reference.
Update flag conventions (-n for limit, -s for state, -J for JSON).
- docs/prd/checkpoint-3.md: Major expansion with gated milestone
structure (Gate A: lexical, Gate B: hybrid, Gate C: sync). Add
prerequisite rename note, code sample conventions, chunking strategy
details, and sqlite-vec rowid encoding scheme. Clarify that Gate A
requires only SQLite + FTS5 with no sqlite-vec dependency.
- docs/phase-a-spec.md: New detailed specification for Gate A (lexical
search MVP) covering document schema, FTS5 configuration, dirty
queue mechanics, CLI interface, and acceptance criteria.
- docs/api-efficiency-findings.md: Analysis of GitLab API pagination
behavior and efficiency observations from production sync runs.
Documents the missing x-next-page header issue and heuristic fix.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Four new test modules covering the search infrastructure:
- tests/embedding.rs: Unit tests for the embedding pipeline including
chunk ID encoding/decoding, change detection, and document chunking
with overlap verification.
- tests/fts_search.rs: Integration tests for FTS5 search including
safe query sanitization, multi-term queries, prefix matching, and
the raw FTS mode for power users.
- tests/hybrid_search.rs: End-to-end tests for hybrid search mode
including RRF fusion correctness, graceful degradation when
embeddings are unavailable, and filter application.
- tests/golden_query_tests.rs: Golden query tests using fixtures
from tests/fixtures/golden_queries.json to verify search quality
against known-good query/result pairs. Ensures ranking stability
across implementation changes.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Extends the CLI with six new commands that complete the search pipeline:
- lore search <QUERY>: Hybrid search with mode selection (lexical,
hybrid, semantic), rich filtering (--type, --author, --project,
--label, --path, --after, --updated-after), result limits, and
optional explain mode showing RRF score breakdowns. Safe FTS mode
sanitizes user input; raw mode passes through for power users.
- lore stats: Document and index statistics with optional --check
for integrity verification and --repair to fix inconsistencies
(orphaned documents, missing FTS entries, stale dirty queue items).
- lore embed: Generate vector embeddings via Ollama. Supports
--retry-failed to re-attempt previously failed embeddings.
- lore generate-docs: Drain the dirty queue to regenerate documents.
--full seeds all entities for complete rebuild. --project scopes
to a single project.
- lore sync: Full pipeline orchestration (ingest issues + MRs,
generate-docs, embed) with --no-embed and --no-docs flags for
partial runs. Reports per-stage results and total elapsed time.
- lore health: Quick pre-flight check (config exists, DB exists,
schema current). Returns exit code 1 if unhealthy. Designed for
agent pre-flight scripts.
- lore robot-docs: Machine-readable command manifest for agent
self-discovery. Returns all commands, flags, examples, exit codes,
and recommended workflows as structured JSON.
Also enhances lore init with --gitlab-url, --token-env-var, and
--projects flags for fully non-interactive robot-mode initialization.
Fixes init's force/non-interactive precedence logic and adds JSON
output for robot mode.
Updates all command files for the GiError -> LoreError rename.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Integrates the dirty tracking system into all four ingestion paths
(issues, MRs, issue discussions, MR discussions). After each entity
is upserted within its transaction, a corresponding dirty_queue entry
is inserted so the document regenerator knows which documents need
rebuilding.
This ensures that document generation stays transactionally consistent
with data changes: if the ingest transaction rolls back, the dirty
marker rolls back too, preventing stale document regeneration attempts.
Also updates GiError references to LoreError in these files as part
of the codebase-wide rename, and adjusts issue discussion logging
from info to debug level to reduce noise during normal sync runs.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements the search module providing three search modes:
- Lexical (FTS5): Full-text search using SQLite FTS5 with safe query
sanitization. User queries are automatically tokenized and wrapped
in proper FTS5 syntax. Supports a "raw" mode for power users who
want direct FTS5 query syntax (NEAR, column filters, etc.).
- Semantic (vector): Embeds the search query via Ollama, then performs
cosine similarity search against stored document embeddings. Results
are deduplicated by doc_id since documents may have multiple chunks.
- Hybrid (default): Executes both lexical and semantic searches in
parallel, then fuses results using Reciprocal Rank Fusion (RRF) with
k=60. This avoids the complexity of score normalization while
producing high-quality merged rankings. Gracefully degrades to
lexical-only when embeddings are unavailable.
Additional components:
- search::filters: Post-retrieval filtering by source_type, author,
project, labels (AND logic), file path prefix, created_after, and
updated_after. Date filters accept relative formats (7d, 2w) and
ISO dates.
- search::rrf: Reciprocal Rank Fusion implementation with configurable
k parameter and optional explain mode that annotates each result
with its component ranks and fusion score breakdown.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements the embedding module that generates vector representations
of documents using a local Ollama instance with the nomic-embed-text
model. These embeddings enable semantic (vector) search and the hybrid
search mode that fuses lexical and semantic results via RRF.
Key components:
- embedding::ollama: HTTP client for the Ollama /api/embeddings
endpoint. Handles connection errors with actionable error messages
(OllamaUnavailable, OllamaModelNotFound) and validates response
dimensions.
- embedding::chunking: Splits long documents into overlapping
paragraph-aware chunks for embedding. Uses a configurable max token
estimate (8192 default for nomic-embed-text) with 10% overlap to
preserve cross-chunk context.
- embedding::chunk_ids: Encodes chunk identity as
doc_id * 1000 + chunk_index for the embeddings table rowid. This
allows vector search to map results back to documents and
deduplicate by doc_id efficiently.
- embedding::change_detector: Compares document content_hash against
stored embedding hashes to skip re-embedding unchanged documents,
making incremental embedding runs fast.
- embedding::pipeline: Orchestrates the full embedding flow: detect
changed documents, chunk them, call Ollama in configurable
concurrency (default 4), store results. Supports --retry-failed
to re-attempt previously failed embeddings.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements the documents module that transforms raw ingested entities
(issues, MRs, discussions) into searchable document blobs stored in
the documents table. This is the foundation for both FTS5 lexical
search and vector embedding.
Key components:
- documents::extractor: Renders entities into structured text documents.
Issues include title, description, labels, milestone, assignees, and
threaded discussion summaries. MRs additionally include source/target
branches, reviewers, and approval status. Discussions are rendered
with full note threading.
- documents::regenerator: Drains the dirty_queue table to regenerate
only documents whose source entities changed since last sync. Supports
full rebuild mode (seeds all entities into dirty queue first) and
project-scoped regeneration.
- documents::truncation: Safety cap at 2MB per document to prevent
pathological outliers from degrading FTS or embedding performance.
- ingestion::dirty_tracker: Marks entities as dirty inside the
ingestion transaction so document regeneration stays consistent
with data changes. Uses INSERT OR IGNORE to deduplicate.
- ingestion::discussion_queue: Queue-based discussion fetching that
isolates individual discussion failures from the broader ingestion
pipeline, preventing a single corrupt discussion from blocking
an entire project sync.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Two targeted fixes to the GitLab API client:
1. Pagination: When the x-next-page header is missing but the current
page returned a full page of results, heuristically advance to the
next page instead of stopping. This fixes silent data truncation
observed with certain GitLab instances that omit pagination headers
on intermediate pages. The existing early-exit on empty or partial
pages remains as the termination condition.
2. Rate limiter: Refactor the async acquire() method into a synchronous
check_delay() that computes the required sleep duration and updates
last_request time while holding the mutex, then releases the lock
before sleeping. This eliminates holding the Mutex<RateLimiter>
across an await point, which previously could block other request
tasks unnecessarily during the sleep interval.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Mechanical rename of GiError -> LoreError across the core module to
match the project's rebranding from gitlab-inbox to gitlore/lore.
Updates the error enum name, all From impls, and the Result type alias.
Additionally introduces:
- New error variants for embedding pipeline: OllamaUnavailable,
OllamaModelNotFound, EmbeddingFailed, EmbeddingsNotBuilt. Each
includes actionable suggestions (e.g., "ollama serve", "ollama pull
nomic-embed-text") to guide users through recovery.
- New error codes 14-16 for programmatic handling of Ollama failures.
- Savepoint-based migration execution in db.rs: each migration now
runs inside a SQLite SAVEPOINT so a failed migration rolls back
cleanly without corrupting the schema_version tracking. Previously
a partial migration could leave the database in an inconsistent
state.
- core::backoff module: exponential backoff with jitter utility for
retry loops in the embedding pipeline and discussion queues.
- core::project module: helper for resolving project IDs and paths
from the local database, used by the document regenerator and
search filters.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
