claims_to_graph emitted every qualifier on asb:qualifier, but that slot is a
closed Bucur enum in indicium — domain-adapter qualifiers (e.g. aligned_with)
failed SHACL. Route by membership: core Bucur terms stay on asb:qualifier;
domain terms go on the open asb:domainQualifier slot added in indicium 1.7.0.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Tags each claim with verify_quote status, emits a JSONL audit sidecar, and
applies strict (drop unverified) vs fail-open (keep tagged) policy.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…nt path

Addresses B1 review: guard the candidates comprehension against non-dict
passages (the one unguarded line that could raise), document in-place mutation
and the strict-ignored-when-verifier-unavailable behavior, and add a test that
simulates the indicia extra being absent (degrades all to unverified, kept).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Replaces the positional chunk_idx bug (builder.py) with verify_quote's
matched_index; writes quote_exact + oa:TextQuoteSelector on verified/repaired
evidence and only an anchorStatus marker on unverified (no false selector).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

The test bound a graph for "kbAnchor" without redirecting the on-disk
manifest, so after the first run the persisted manifest marked the paper
unchanged and subsequent runs skipped extraction (empty graph -> assertion
failed). Monkeypatch manifest._DATA_DIR to tmp_path, matching every other
build_claim_graph test. Also sort the new imports and drop the now-unused
enumerate index left by the positional->content-match rewrite.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…ig to builder

annotate_anchor_status tags each extracted record (verified/repaired/
unverified/unchecked) before the license rewrite; MCP build_claim_graph now
passes config.anchor.* (strict/near_threshold/audit_dir) into the builder.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

The build_claim_graph MCP tool now reads config.anchor.* (R3), so its unit
test must give the fake MCPState a real Config (was None). Also moves the
type-only pathlib.Path import under TYPE_CHECKING and drops a stale noqa in
anchor.py.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Drives build_claim_graph with a fabricated quote (verbatim in no passage) and
asserts the claim is kept (fail-open) but tagged unverified with zero
quoteExact / oa:TextQuoteSelector triples — pinning the no-laundering
invariant at the builder layer where the selector logic lives.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Wire the MCP claim-extraction tool through the R3 anchor invariant: after
extract_claims, run anchor_claims so every returned claim carries an _anchor
record {status, matched_index, quote_exact, score}. quote_exact is populated
only for verified/repaired claims — never fabricated for an unverified quote
(no-laundering holds at the tool layer, not just the builder).

Also fix a latent blank-passage bug: get_relevant_passages emits flat
{text, source_doi}, but extract_claims and anchor_claims both read
{chunk_text, source:{doi}}. Normalize once up front so the model sees real
passage text and the verifier gets real candidates — otherwise anchoring would
silently see empty text and tag everything unverified.

Pin indicium>=1.8.0 (the floor that ships the verify_quote kernel the anchor
path imports) and add two unit tests: a verbatim quote anchors verified with
quote_exact recovered; a fabricated quote is kept but tagged unverified with
quote_exact=None.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

The retry predicate now requires isinstance(e, Exception) so that
asyncio.CancelledError (a BaseException, which tenacity otherwise catches)
propagates instead of being retried. Without this, every caller-side
asyncio.wait_for/timeout was ineffective — all 3 retry attempts ran their
full HTTP timeout before the stale deadline was noticed, so the structured
extraction tools blew past their 80s cap (observed 115-151s). The
extract_failure_modes / extract_parameters calls now return within the cap.

route_kbs gains a kb_names alias for candidate_kbs: every other multi-KB
tool names that argument kb_names, so agents reach for it first and the
mismatch previously errored the first call.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Flip the CI ruff gate from report-only (--exit-zero) to enforcing, pin
ruff to 0.15.13, and align CI + mypy to Python 3.12 (matches
requires-python). Add .pre-commit-config.yaml running the same pinned
ruff (lint + format-on-touched-files only, per the pyproject note).

Resolve all outstanding ruff findings via the established fix-vs-ignore
policy: fix real issues (B904 exception chaining, F401/F841 dead code,
RUF012 ClassVar, E731, UP035, SIM105, B905 zip strict=) and config-ignore
genuine false positives with justifications (E402, UP042, SIM102, SIM117,
TC001-3, RUF001-3, N806/N814, B008, plus test-idiom per-file ignores).

Also fix two regressions from over-eager autofixes surfaced by the suite:
- jobs/registry.py: SIM118 had rewritten `for k in row.keys()` on an
  aiosqlite.Row (iterating a Row yields values, not keys) -> IndexError.
  Restored .keys() with a justified noqa.
- cite_graph.py: restored openalex_id_for_doi (a test monkeypatch target)
  with a justified noqa: F401.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Adopt a pragmatic-but-enforced mypy config: semantic bug-catchers stay on
(attr-defined, arg-type, return-value, index, assignment, ...) while
annotation-completeness rules stay off to avoid churn in a research codebase.
Drove all findings to zero across src/, fixing several real latent bugs
surfaced along the way: StreamEvent payloads passed a dict where a str was
required (now json.dumps), a wrong base64.binascii.Error reference, non-indexable
Iterable[Row] from execute_fetchall, a referenced-but-deleted except variable,
and a missing DOI None-guard. Added `mypy src/` to the CI lint job so the gate
holds going forward.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Audit of bare `except Exception` swallows: narrow defensive parse/nav
handlers to precise exception types (KeyError/TypeError/AttributeError,
ValueError, json.JSONDecodeError, ImportError) so unexpected errors
propagate instead of being silently eaten; keep best-effort side-effect
handlers broad but add logger.debug observability. Behavior-preserving:
identical control flow, no new suppression.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Move the seven implementation helpers (_start_mcp_and_web, _run_query,
the bibtex KB builders, _build_app_state_for_cli, and the ingest summary
printers) out of the 2000-line cli.py into a dedicated cli_helpers module,
re-importing them so perspicacite.cli.<name> stays a valid monkeypatch
target for the existing unit tests. First slice of the god-module split;
the larger server.py/orchestrator splits remain a dedicated follow-up.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Reconciles the hardening work (ruff/mypy CI gate + cli_helpers
decomposition) with what landed on main after the branch was cut:
the R3 anchor invariant (PR #10) and the claims markdown-fence fix
(PR #9). The anchor logic was already present on this branch via its
feat/domain-qualifier-routing base, so it reconciles cleanly; PR #9's
_strip_markdown_fence is the net new content.

Conflicts (both resolved keeping BOTH anchor logic and lint cleanups):
- src/perspicacite/indicium_layer/builder.py: dropped unused
  INDICIUM_NS import (kept the W1 F401 cleanup; anchor code unaffected).
- tests/unit/test_claims_validation.py: kept the isort blank line
  separating third-party from first-party imports.

Verified on the merged tree: ruff clean, mypy clean (241 files),
2487 passed / 1 skipped.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

CI's lint and test jobs install base deps only (no `indicia` extra), so
rdflib/pyoxigraph/indicium are absent there while present locally. Two
pre-existing failures surfaced after the main merge:

- mypy lint: the ignore_missing_imports override omitted rdflib/pyoxigraph,
  and store.py aliased `ConjunctiveGraph = Dataset` — a local assignment that
  is invalid as a type annotation. Add both modules to the override list,
  drop the alias (use Dataset directly), and disable warn_unused_ignores for
  indicium_layer.store where the rdflib union-attr ignore is env-dependent.
- smoke test: _TOOL_ARGS lacked entries for 8 web/skill/claim-graph tools, so
  test_tool_inventory called them with no args and hit TypeErrors. Supply
  minimal valid args; claim-graph tools short-circuit cleanly at `import
  indicium` in CI.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…etry test

CI installs only ".[dev]" via plain pip, which cannot resolve the local-path
`indicium` source, so indicium/rdflib/pyoxigraph/indicium_adapters are never
present in CI. Guard every claim-graph / verify-kernel test with
pytest.importorskip so they skip cleanly there while still running locally
under `uv sync --extra indicia`. Mixed modules use per-test guards to preserve
their non-indicia coverage rather than skipping the whole module.

Also deselect test_domain_aggregator::test_retry_counts_as_one_circuit_failure
from the safe subset: it exercises real exponential-backoff retry sleeps
(~27s) and exceeds the 15s safe-subset timeout. It passes under a longer
timeout and is unrelated to the indicia extra.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…on links

Move the 18 root-level config files into a tidy config/ tree to declutter
the repo root:
  - config/example.yml                (was config.example.yml)
  - config/providers/*.yml            (6 provider presets, was config.*.example.yml)
  - config/embeddings/*.yml           (11 model presets, was config_*.yml)

Names are cleaned (drop config./config_ prefix and .example infix,
underscores -> hyphens). All references updated across docs, README,
help/error strings (agent_cli, claude_cli, check_install), and tests.
The config-audit test glob now points at config/providers + config/example.

Also add proper citation links to the README (CEUR-WS PDF + HAL archive)
in both the human-readable block and the BibTeX entry, and extend
.gitignore with tooling caches (.mypy_cache, .pytest_cache, .coverage)
and local claim-graph stores.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

- CI: enable actions/setup-python pip caching (keyed on pyproject.toml)
  for both the lint and test jobs to speed up dependency installs.
- Add .github/dependabot.yml: weekly grouped updates for pip and
  github-actions ecosystems.
- pyproject.toml: add readme, authors, keywords, trove classifiers, and
  [project.urls] (Homepage/Repository/Issues/Changelog/Paper) for a
  release-quality package description.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Update inbound references to the relocated root files (INSTALL_AGENT.md,
MANUAL_QA.md) in setup.md, docs/development/testing.md, and docs/index.md,
and drop the now-stale root allowlist entries from .gitignore (the files
live under the already-allowlisted docs/ tree).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Watch frontend/package-lock.json (grouped, weekly) so the frontend's npm
dependencies get the same automated update treatment as the Python deps.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Wire a third CI job for the Next.js frontend running lint, typecheck,
unit tests, and build with npm caching. Add eslint flat config (Next 16
core-web-vitals; React-Compiler-era react-hooks rules set to warn for a
green baseline) and a vitest suite covering the cost estimation helpers.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Remove the blanket warn-downgrade of the React-Compiler react-hooks rules
and address all 21 findings: genuinely unnecessary effects are refactored
(redundant setState removed; reset-on-change moved to render-time guards;
ChatPanel's cancel callback reordered above its consuming effect), and the
remaining legitimate effect patterns (async mount fetches, SSR-safe
localStorage hydration, intentional latches) carry per-line justified
disables. The four rules are back at "error" so future violations fail CI.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Add a log line to broad exception handlers that previously swallowed
errors silently, so failures in optional-dep fallbacks, JSON parsing,
download workers, and background jobs surface in logs. Behavior is
preserved: control flow, return values, and re-raises are unchanged.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…chestrator

Completes the silent-handler audit for the two modules deferred from the
first pass. Each previously-silent broad except now logs one line (debug
for best-effort/fallback paths, warning where user-meaningful work is
dropped) before its existing fallback. Behavior is unchanged: control
flow, return values, and caught types are preserved.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>