[lipikaramaswamy]

Related Issue

Fixes #334

Plan Document

No plan required: this is a small clarification to existing date-label guidance, examples, and tests.

Summary

The validator currently keeps dates but drops time references that are not "specific dates", while all built-in date examples are complete dates. This leaves models to decide whether a month or year is specific enough.

Clarify that date includes complete and partial calendar references when context establishes their meaning:

• March in "The appointment is in March"
• 2024 in "She joined in 2024"
• March 2024 in "We met in March 2024"
• the 15th in "The appointment is on the 15th"

Keep the requirement for calendar context: "March forward" and "2024 items" are not dates. Preserve the existing exclusion of vague time references such as "today", "now", and "soon".

Align the validator and augmenter instructions, extend the built-in examples, and document the policy. The existing date label remains the umbrella label.

Type of Change

• Bug fix
• Documentation update

Contributor Checklist

• PR title follows Conventional Commits
• Related issue is linked
• No-plan reason is documented above
• Public API impact checked; existing API and agent-skill template remain valid
• Examples and tests use synthetic data
• No credentials or private endpoint details added

Validation

• make check
• make test (with TMPDIR=/private/tmp and a short pytest temporary directory for socket-path compatibility)
• make docs-build

Tests verify the generated validator and augmenter guidance and the built-in examples. Live-model behavior remains probabilistic; these are prompt-content tests rather than a model-quality benchmark.

Documentation and Artifacts

• Detection documentation updated
• make docs-build passes locally

PR stack

Merge/review order: #305 → #267 → #335.

• refactor(detection): establish deterministic entity candidate lifecycle #305: candidate-lifecycle foundation, targeting main.
• fix(detection): augment plain source on candidate lifecycle #267: plain-source augmentation, stacked on refactor(detection): establish deterministic entity candidate lifecycle #305.
• fix(detection): clarify partial calendar date validation #335: partial-calendar-date guidance, stacked on fix(detection): augment plain source on candidate lifecycle #267.
• feat(detection): add regex entity detection #265: separate regex feature built on refactor(detection): establish deterministic entity candidate lifecycle #305; it is not a prerequisite or merge gate for refactor(detection): establish deterministic entity candidate lifecycle #305, fix(detection): augment plain source on candidate lifecycle #267, or fix(detection): clarify partial calendar date validation #335.

After a parent is squash-merged, restack the remaining child commits onto the updated parent/main and retarget their PRs as appropriate. These links describe dependency order, not approval or CI status.

Restack verification (2026-11-04)

Restacked the date-guidance commit onto #267 at 16e9d0d3; the PR adds exactly one commit (b2043cd5, four files, 43 additions and two deletions). The resulting tracked tree is identical to the combined #305 + #267 + #335 candidate 02c58f7a.

• Repository commit hooks, including make check and DCO: passed.
• Unit tests: 1,839 passed initially; five sandbox-blocked socket/SDK-log tests passed on an unsandboxed rerun (1,844 total).
• Strict documentation build and git diff --check: passed.
• No new live-model calls or model-quality claim; feat(detection): add regex entity detection #265's code was not changed.

[greptile-apps]

Confidence Score: 5/5

[Medium impact] The PR appears safe to merge; no actionable issues were found.

Summary

Clarifies that date covers partial calendar references when the surrounding text establishes their meaning.

• Detection prompts now keep partial calendar dates when context makes them dates.

No actionable issues found. Tests were not run during this review.

_{Reviews (3) · Last reviewed commit: "fix(detection): clarify partial calendar..." · Reviewed by Greptile}

[asteier2026]

One, probably minor, thing we should remember is this does slightly increase the chance of replace propagating changes to instances of the value with different meanings. "March" the month and "march" the verb; "2024" the year and "2024" a quantity. But that is an existing issue we may have to tackle at some point. I wouldn't worry about it now.

Also, I'm presuming you ran the slurm benchmarking and verified prec/recall on datasets like OpenPIIand NemotronPII remain good.