docs(skill): align the omnigraph skill with 0.8.1

[aaltshuler]

Post-release alignment sweep for v0.8.1, skill portion.

• Undirected traversal ($a <edge> $b) added to the quick rules, the embedded grammar excerpt, and references/queries.md (worked example; T22 rule; composition with bounds/not{}) — marked >= 0.8.1.
• Enum widening added to references/schema.md as a supported metadata-only apply; narrowing/rename/enum↔String noted as still refused.
• Two pre-existing inaccuracies found by verification during the sweep: the skill claimed {1,} = unbounded traversal (the typechecker rejects unbounded, T15), and the required-property recipe ended with "tighten to required in a follow-up apply" (refused by the planner as a property-type change, OG-MF-106). Both corrected to match engine behavior.
• Skill version → 0.8.1; compatibility line marks version-gated features.

Rest of the sweep (verified aligned, no changes needed): README (no version-bearing claims), docs/ (no stale 0.8.0 refs outside history), Homebrew tap (auto-updated to v0.8.1 by release.yml), omnigraph-ts (0.8.0 sync merged; 0.8.1 sync to follow via its pipeline), omnigraph-web (doc re-sync remains deferred pending the flat-only import script fix).

---

Greptile Summary

This PR aligns the omnigraph skill with the v0.8.1 release by adding new features and correcting two pre-existing inaccuracies discovered during the sweep.

• New: Undirected traversal ($a <edge> $b, >= 0.8.1) added to quick rules, the embedded PEG grammar excerpt, and references/queries.md with a worked example, T22 rule (asymmetric-edge rejection), and bounds/not{} composition notes.
• New: Enum widening added to references/schema.md as a supported metadata-only schema apply (narrowing/rename/enum↔String remain refused per OG-MF-106).
• Corrected: Bounded-traversal — the previous claim that {1,} equals unbounded traversal has been fixed to state the typechecker rejects unbounded bounds outright ("unbounded traversal is disabled"); Corrected: required-property recipe — step 4 formerly advised tightening T? → T in a follow-up apply, which the planner actively refuses (OG-MF-106); both the references/schema.md recipe and the Common Gotchas table in SKILL.md now reflect the correct behavior.

Confidence Score: 5/5

Documentation-only change correcting two previously inaccurate gotchas and adding two new 0.8.1 features; all edits are internally consistent across the three skill files and cross-check cleanly against AGENTS.md.

Every factual claim in the diff was cross-verified: undirected traversal semantics (T22, deduplication, bounds composition) align with AGENTS.md's GraphIndex/CSR description; the bounded-only traversal correction matches the stated typecheck behaviour; enum widening as metadata-only apply is consistent with the schema.md supported-types section; the D₂ explanation update matches the staged-delete path described in AGENTS.md. The prior P1 finding (stale gotcha advising a refused schema apply) is addressed in both SKILL.md and references/schema.md. No logic, runtime, or contract changes are introduced.

No files require special attention; all three changed files are consistent with each other and with AGENTS.md.

Important Files Changed

Filename	Overview
skills/omnigraph/SKILL.md	Version bumped to 0.8.1; compatibility line extended; two pre-existing inaccuracies corrected (bounded-only traversal, required-property tighten refused); undirected traversal added to quick rules, grammar excerpt, and Common Gotchas; all changes are internally consistent
skills/omnigraph/references/queries.md	Undirected traversal section added with a worked example (T22 rule, bounds composition, negation); D₂ rule explanation updated to reflect that deletes now use the same staged publish path as inserts/updates (since 0.8.0), making it a deliberate semantic boundary rather than an implementation gap; both changes align with AGENTS.md
skills/omnigraph/references/schema.md	Required-property backfill recipe step 4 corrected (tightening T?→T is refused, OG-MF-106); enum widening section added describing it as a metadata-only apply with dedup of value order; both additions are consistent with AGENTS.md and schema.md supported-types table

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["Traversal in .gq match block"] --> B{Edge brackets?}
    B -- "directional\n$p edge $q" --> C["Directional traversal\n(all engine versions)"]
    B -- "undirected\n$p <edge> $q\n>= 0.8.1" --> D{Same endpoint type?}
    D -- "yes\ne.g. Issue -> Issue" --> E["Undirected match\n(either direction, deduplicated)"]
    D -- "no\ne.g. Comment -> Issue" --> F["Rejected: T22\nasymmetric edge"]
    C --> G{Hop bounds?}
    E --> G
    G -- "{n,m} finite" --> H["Variable-hop traversal"]
    G -- "{n,} unbounded" --> I["Rejected: T15\nunbounded disabled"]
    G -- "none" --> J["Single-hop traversal"]

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    A["Traversal in .gq match block"] --> B{Edge brackets?}
    B -- "directional\n$p edge $q" --> C["Directional traversal\n(all engine versions)"]
    B -- "undirected\n$p <edge> $q\n>= 0.8.1" --> D{Same endpoint type?}
    D -- "yes\ne.g. Issue -> Issue" --> E["Undirected match\n(either direction, deduplicated)"]
    D -- "no\ne.g. Comment -> Issue" --> F["Rejected: T22\nasymmetric edge"]
    C --> G{Hop bounds?}
    E --> G
    G -- "{n,m} finite" --> H["Variable-hop traversal"]
    G -- "{n,} unbounded" --> I["Rejected: T15\nunbounded disabled"]
    G -- "none" --> J["Single-hop traversal"]

Loading

Comments Outside Diff (1)

1. skills/omnigraph/SKILL.md, line 380 (link)

   Stale gotcha: still advises "tighten in follow-up apply"

   The PR corrects the inaccurate step-4 text in references/schema.md (tightening T? → T is refused, OG-MF-106), but the matching "Common Gotchas" row in SKILL.md still ends with "Make optional → backfill → tighten in follow-up apply". An agent following only the main SKILL.md (the default load path) will be advised to attempt a schema apply that the planner actively refuses, which is the exact footgun this PR was targeting.

   Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

   

_{Reviews (2): Last reviewed commit: "docs(skill): review follow-ups — gotcha-..." | Re-trigger Greptile}

[devin-ai-integration]

Devin Review found 3 potential issues.