fix(engine): don't charge per-file-skipped files against repo-map's aggregate budget

[dhgoal]

Closes #7247.

Problem

buildRepoMap enforces two byte budgets: a per-file cap (maxSourceBytes) and an aggregate cap across all parsed files (maxTotalSourceBytes). It added each file's byte count to totalSourceBytes before the check that skips a file for exceeding the per-file cap:

const sourceBytes = Buffer.byteLength(file.sourceText, "utf8");
totalSourceBytes += sourceBytes;               // charged unconditionally...
if (sourceBytes > maxSourceBytes || totalSourceBytes > maxTotalSourceBytes) {
  entries.push({ ..., skipped: "resource_limit" });   // ...even for a file skipped here, never parsed
  continue;
}

So a single oversized file (a vendored/minified asset or generated bundle that slips into the input list) that is itself skipped for exceeding maxSourceBytes still burned its full byte count against maxTotalSourceBytes — the aggregate budget meant to bound total parsed work. Every subsequent small, legitimate file then also tripped totalSourceBytes > maxTotalSourceBytes and was marked skipped: "resource_limit", producing a near-empty, order-dependent repo map with no error surfaced to the driver that consumes it.

Fix

Accrue to the aggregate after the per-file cap check, so only files that pass the per-file cap (and are actually parsed) count against the aggregate. Files under the per-file cap accrue and gate against the aggregate exactly as before — the only behavior change is that a per-file-oversized, unparsed file no longer poisons the aggregate for the files after it.

Tests

Added a regression test in repo-map.test.ts (imports the engine via its source barrel, so the changed lines are instrumented): an oversized first file (28 bytes, over a maxSourceBytes: 20 / maxTotalSourceBytes: 20 config) is skipped, and the small legitimate file after it is still parsed (symbols extracted) rather than starved of aggregate budget. The existing per-file-cap and aggregate-cap tests continue to cover both skip branches.

Verification

• tsc --noEmit (root typecheck) — clean.
• npm run build --workspace @loopover/engine — clean.
• repo-map.test.ts — 35 tests green.
• Changed file repo-map.ts: 113/113 lines, 12/12 functions, 84/84 branches covered (v8/lcov).

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.81%. Comparing base (3c72fce) to head (676854c).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #7270   +/-   ##
=======================================
  Coverage   93.81%   93.81%           
=======================================
  Files         704      704           
  Lines       69457    69459    +2     
  Branches    18899    18899           
=======================================
+ Hits        65161    65163    +2     
  Misses       3302     3302           
  Partials      994      994           

Flag	Coverage Δ
shard-1	43.60% <0.00%> (-0.01%)	⬇️
shard-2	37.33% <100.00%> (+0.11%)	⬆️
shard-3	33.18% <0.00%> (+0.08%)	⬆️
shard-4	33.74% <0.00%> (-0.59%)	⬇️
shard-5	32.89% <0.00%> (+0.70%)	⬆️
shard-6	45.48% <0.00%> (-0.30%)	⬇️

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

Files with missing lines	Coverage Δ
packages/loopover-engine/src/miner/repo-map.ts	100.00% <100.00%> (ø)

[loopover-orb]

Tip

✅ LoopOver review result - approve/merge recommended

_{Review updated: 2026-07-19 11:46:38 UTC}

2 files · 1 AI reviewer · no blockers · readiness 86/100 · CI green · clean

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
This correctly fixes the ordering bug in buildRepoMap: the per-file cap check now runs before totalSourceBytes is incremented, so a single oversized/unparsed file no longer poisons the aggregate budget for subsequent legitimate files. The fix is minimal, in the right layer (source of the accrual, not a downstream symptom), and the regression test exercises the exact real path described in the issue (28-byte file over a 20-byte cap, followed by a small file that must still parse). The existing per-file-cap and aggregate-cap tests still pass and correctly continue to cover both skip branches independently.

Nits — 3 non-blocking

• The inline comment at repo-map.ts:271-276 is fairly long for a single conditional; could be trimmed to a sentence since the PR description already explains the rationale in detail.
• No test explicitly confirms totalSourceBytes stays at 0 (not partially incremented) when the oversized file is skipped — the existing test infers this indirectly via the second file parsing successfully.
• Consider a brief unit assertion on the byte-accounting invariant directly (e.g., exposing or asserting via a follow-up small file whose size is exactly at the remaining budget) to make the aggregate-budget behavior even more explicit, though the current test already demonstrates the fix adequately.

Decision drivers

• ✅ Code review — No blockers (1 reviewer)
• ✅ Gate result — Passing (No configured blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	✅ Linked	#7247
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	❌ 8/20	High review scope from cached public metadata (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 103 registered-repo PR(s), 63 merged, 35 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	dhgoal; Gittensor profile; 103 PR(s), 35 issue(s).
Improvement	✅ Minor	risk: clean · value: minor · LLM: moderate

Linked issue satisfaction

Addressed
The diff moves totalSourceBytes accrual after the per-file maxSourceBytes check and skips oversized files before charging the aggregate, exactly matching the requested reordering, and adds the specified regression test proving a small file after an oversized one is no longer starved of aggregate budget.

Review context

• Author: dhgoal
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: JavaScript
• Official Gittensor activity: 103 PR(s), 35 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Start here: Add a concise scope and risk note.
• Then work through the remaining 1 step in the Signals table above.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

🧪 Chat with LoopOver

Ask LoopOver a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @loopover ask <question> answers contribution-quality Q&A with source citations and freshness.
• @loopover chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @loopover mention with a real question is routed to the closest matching read-only command automatically — no exact syntax required.

Full command reference: https://loopover.ai/docs/loopover-commands

🧪 Experimental — new and may change.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by LoopOver, a quiet PR intelligence layer for OSS maintainers.

• Re-run LoopOver review

[loopover-orb]

LoopOver approves — the gate is satisfied and CI is green.